WEBVTT
00:00:19.600 --> 00:00:21.120
Welcome to the future of you.
00:00:21.359 --> 00:00:25.760
Today I'm speaking with Jack Connor, a tech entrepreneur and language enthusiast.
00:00:26.000 --> 00:00:32.240
His latest project aims to deposit endangered language data into the Arctic code vault.
00:00:32.399 --> 00:00:40.560
The project aims to create AI models of vulnerable languages, ensuring cultural and linguistic heritage is preserved for future generations.
00:00:40.880 --> 00:00:54.640
In this chat, he highlights the rapid disappearance of languages due to the globalizing homogenizing effects of the internet and discusses the importance of and challenges with language preservation and how AI is key.
00:00:54.880 --> 00:01:00.079
We also consider language revitalization, with Jack sharing his own personal experiences.
00:01:00.240 --> 00:01:03.920
Jack speaks at least seven languages and has travelled extensively.
00:01:04.079 --> 00:01:13.760
We also go a bit more sci-fi, envisioning real-time AI-driven auto-translation tools that could alleviate the pressure on major languages in the future.
00:01:14.000 --> 00:01:21.280
Of course, language and identity are totally interrelated, so that is always at the heart of this conversation.
00:01:21.599 --> 00:01:28.640
As you listen to this chat, picture Jack in the Arctic, as that is where he will be by the time we go on air.
00:01:28.719 --> 00:01:43.200
And if you'd like to get in touch, help out with the project by providing audio recordings or even cell phone video for the language deposit, or have any endangered languages or dialects that you care deeply about and want to suggest to him.
00:01:43.359 --> 00:01:46.159
I'll leave all the details in the show notes.
00:01:51.280 --> 00:01:53.519
Jack Connor, welcome to the Future of You.
00:01:54.079 --> 00:01:54.480
Thank you.
00:01:54.640 --> 00:01:55.760
Thank you for having me on.
00:01:55.840 --> 00:01:58.719
It's uh it's an honor and uh yeah, great to be here.
00:01:59.040 --> 00:02:06.159
Now it's brilliant to have you here, Jack, because unusually for the guests I have on here, I'm not 100% familiar with your work.
00:02:06.239 --> 00:02:07.680
I mean, I know about it.
00:02:07.920 --> 00:02:15.919
Theo Priestley put me in your direction when we're talking about something on Twitter, and I'm so glad he did because your work is absolutely fascinating.
00:02:16.000 --> 00:02:23.919
And of course, language and the intersection of language and identity is incredibly interesting, nuanced, and complex.
00:02:24.000 --> 00:02:32.080
So I wonder if we'll start off by you just uh explaining to the audience who you are and and and what you're doing and why you're going to the Arctic.
00:02:32.400 --> 00:02:32.800
Mm-hmm.
00:02:33.759 --> 00:02:34.479
Great questions.
00:02:34.719 --> 00:02:35.280
Great questions.
00:02:35.439 --> 00:02:37.039
Well, yeah, thanks for the intro.
00:02:37.120 --> 00:02:48.319
Um I'm Jack Connor, uh based in uh outside Los Angeles, and I'm going to the Arctic to deposit a reel of AI data and language data.
00:02:48.479 --> 00:03:03.439
It's about 550 languages worth of data into this Arctic code vault that's uh, you know, impervious to EMPs, is 250 feet below the permafrost.
00:03:03.919 --> 00:03:08.240
For the listeners and for you, if most people are familiar with the global seed vault.
00:03:08.639 --> 00:03:09.840
It's that for data.
00:03:10.000 --> 00:03:12.639
And they're literally neighbors, they're right next to each other.
00:03:12.719 --> 00:03:15.599
It was it was built as sort of a response to the global seed vault.
00:03:15.840 --> 00:03:16.960
I'm a language lover.
00:03:17.039 --> 00:03:18.240
I've always loved languages.
00:03:18.400 --> 00:03:21.599
I I lived abroad a lot when I was a teenager in my twenties.
00:03:21.840 --> 00:03:24.400
And as an American, this is a little more unusual, I think.
00:03:24.479 --> 00:03:29.599
Uh and so my background is tech, as a lot of people are probably on, like programmer.
00:03:29.680 --> 00:03:37.039
I was I was uh CTO of a couple uh different Silicon Valley startups that did uh pretty good.
00:03:37.360 --> 00:03:40.479
But I always did language stuff as kind of on the side.
00:03:40.639 --> 00:03:45.919
The combination of these two is kind of what got me into language preservation of endangered languages.
00:03:46.159 --> 00:04:03.840
As I continued to work and sort of explore, one of the one of the big topics that I started to take a lot of interest in and didn't see quite so much action around is the fact that languages are disappearing much, much, much faster than at any time in history.
00:04:04.080 --> 00:04:07.520
This is essentially because of the internet, and it's mostly a good thing.
00:04:07.599 --> 00:04:18.480
You know, I had coworkers in Brazil who spoke perfect English because they played Counter-Strike all day or watched Netflix all day and could get jobs that paid way higher than local rates, and it was awesome.
00:04:18.560 --> 00:04:20.560
You know, this is uh globalization.
00:04:20.800 --> 00:04:24.399
There's, you know, there's pros and cons, but uh I think there's a lot of good stuff.
00:04:24.560 --> 00:04:29.920
But one of the negative things is it is quickly pushing out a lot of local languages.
00:04:30.240 --> 00:04:37.279
We're doing a project in Panama right now with uh Ngave, which is a language down there that's spoken by about 200,000 speakers.
00:04:37.519 --> 00:04:43.120
So we're building some tools uh to put in a hospital there because they have a they have a big uh translator shortage.
00:04:43.279 --> 00:04:48.720
You know, people come into the hospital, have a problem, but there's nobody who can do the interpretation work.
00:04:48.800 --> 00:04:52.959
So we're gonna build like an app that has like some text-to-speech and things like that.
00:04:53.199 --> 00:05:09.759
But it's an example of a language, and it 200,000 is actually a lot of speakers, but you have a lot of languages like that where if you're growing up in that culture, is it more valuable to learn English or Spanish, or is it more valuable to learn your local language?
00:05:10.000 --> 00:05:15.040
So of course, like English and Spanish, because you can get jobs, you know, things like that.
00:05:15.279 --> 00:05:28.480
So you just have a lot of pressure on these tiny little niche local languages, uh, you know, of which there's estimated to be 7,000 in the world, and the pressure is, yeah, I mean, inherently causing a lot of them to disappear.
00:05:28.639 --> 00:05:40.959
Uh here in California, there's, you know, many, many, many indigenous languages, but you know, 100 years ago there used to be uh five-ish times as much, or you know, something along those lines.
00:05:41.199 --> 00:05:52.480
And these days, there's there's also a lot of these languages have, you know, five to twenty speakers, and the average age of the native speakers is over 75.
00:05:53.040 --> 00:05:56.000
This is something that I think most people just don't even think about.
00:05:56.160 --> 00:05:56.720
For good reason.
00:05:56.879 --> 00:06:00.399
I mean, you're not but it's something that's really quickly happening.
00:06:00.639 --> 00:06:04.560
It hurts me as someone who just loves languages.
00:06:04.879 --> 00:06:14.720
I uh have lots of uh thoughts about, you know, language and how language affects how we, you know, pattern our thinking and things like that.
00:06:14.959 --> 00:06:22.000
But the other side is uh in a more tangible way, we're starting to do very interesting research.
00:06:22.240 --> 00:06:30.879
We're throwing a bunch of AI at language data and discovering things like whole civilizations that we didn't know existed.
00:06:31.120 --> 00:06:44.480
Because you triangulate from etymologies or you know, finding out that, you know, for a long time there was uh an argument whether Sanskrit and Indo-European, that it was an Indo-European language.
00:06:44.720 --> 00:06:47.279
Linguists don't get mad at me if I'm getting this slightly wrong.
00:06:47.439 --> 00:06:51.680
But basically, you know, trying to figure out was European languages and Sanskrit related?
00:06:51.920 --> 00:06:56.399
And for a long time people said no, and then we found out recently the answer was definitely yes.
00:06:56.639 --> 00:07:10.319
You know, so there's we're sort of at this point when we're opening up all these cool, amazing, like research-oriented uh tools we can use languages for, our languages are starting to disappear.
00:07:10.720 --> 00:07:13.680
How do you save or preserve these languages?
00:07:13.839 --> 00:07:19.360
And how do you sort of turn them into a data in a format that can be preserved?
00:07:19.600 --> 00:07:20.959
Talk us through a bit of the details.
00:07:21.279 --> 00:07:22.319
Because it's fascinating.
00:07:22.720 --> 00:07:23.360
Yeah, yeah.
00:07:23.519 --> 00:07:46.399
I mean, perfect world if you know, if a genie came out of a lamp and said, Jack, like you can have anything you want with this project, I would have it create an AI of every accent and a Chat GPT clone in every language in the world, and uh maybe go back in time and you know get a get a ton of get a ton more.
00:07:46.560 --> 00:07:47.920
This is never gonna happen.
00:07:48.160 --> 00:07:53.199
So for the the preservation is like for a lot of these little languages, there isn't a lot of things.
00:07:53.439 --> 00:07:56.160
Have you heard of the disappearing internet theory?
00:07:56.399 --> 00:08:03.759
It's kind of like a highfalutin sort of title for the fact that essentially just old websites are disappearing.
00:08:04.079 --> 00:08:07.600
Oh, Brian Rommel's been um tweeting this a lot, hasn't he?
00:08:07.759 --> 00:08:12.959
Saying like there's two million, this is fascinating, there's two million academic papers that have just kind of disappeared.
00:08:13.040 --> 00:08:23.360
So if you're following the footnotes, suddenly you come to like the end because there's uh they've been hosted on websites that have disappeared or the or the uh the actual papers have disappeared.
00:08:23.519 --> 00:08:25.759
So yeah, is that is that what you're talking about?
00:08:26.000 --> 00:08:26.399
Mm-hmm.
00:08:26.720 --> 00:08:35.440
Well, with with a lot of niche languages, this is a huge problem because a lot of to give you an example, here in California, there's a language called COWIA.
00:08:35.600 --> 00:08:44.159
There's a couple variations of it, and uh there's a revitalization effort going, but it's like teensy tiny little language with you know a couple dozen speakers.
00:08:44.480 --> 00:09:03.519
But if you go and try and find, you know, like grammar books, uh dictionaries, things like that, there's a couple online and they're being hosted on what is a basically a WordPress site being run by a two-person nonprofit, you know, out in the you know, kind of Joshua Tree Desert.
00:09:04.320 --> 00:09:16.960
And if that goes like if that site goes down, there goes, you know, 85% of of everything in that language, you know, because there's so, so little data for some of these languages.
00:09:17.120 --> 00:09:22.159
And so if you're looking on a time frame of you know, five years, it might still be there.
00:09:22.399 --> 00:09:27.600
Fifty years, probably not, like 500 years, like really, like who knows.
00:09:27.919 --> 00:09:36.960
So for some of these small languages, and you know, I said on this, and this first deposit we're doing is 554 languages worth of data.
00:09:37.200 --> 00:09:41.919
For some of these, it's just you know, PDFs of dictionaries, grammar books.
00:09:42.080 --> 00:09:50.960
Uh we we have a partner named WikiTongues, who has uh probably the biggest collection in the world of niche language audio and video.
00:09:51.919 --> 00:10:12.000
So if nothing else, you know, if uh Tracy, if you, you know, if you spoke, uh were one of the last speakers of you know a very niche language, if nothing else, we want to get like audio, some grammar, um, a dictionary, and then any, you know, basically any like papers or information I could find.
00:10:12.159 --> 00:10:16.399
You know, huge research project, but for some of these, there's just there's just not much.
00:10:16.559 --> 00:10:23.679
And most languages are oral languages too, meaning they really haven't left a lot of physical data.
00:10:23.919 --> 00:10:30.240
But we get what we can, you know, because again, from uh like a linguist standpoint, it's just it's important.
00:10:30.480 --> 00:10:41.679
Like one of the sad things is if you have an oral language that doesn't really have any written content and the last speaker dies, like that's it.
00:10:41.919 --> 00:10:49.919
Like there's there is no coming back, it is gone from the face of the earth, and you will never be able to recreate it ever again.
00:10:50.159 --> 00:11:13.519
So if nothing else, I want to make sure we got representation for that, you know, for those, hence the the breadth of this project, because we were able to get, you know, there's a lot of languages where we effectively grabbed everything that was on the internet, but everything that was on the internet is, you know, maybe uh a couple uh couple hundred-page PDFs and some some white papers uh from some linguists and stuff like that.
00:11:13.919 --> 00:11:36.639
But then on the other side, like on the, you know, for some of the languages that have a lot more, have more extensive things, the goal is honestly to build AI so that our children's, children's, children, children could be great as they could like you know, knock on some box and some AI genie pops out and talks to them the way they talk to us, or you know, some kind of interface or you know, whatever the case may be.
00:11:36.799 --> 00:11:42.399
But basically so that they could know, oh, this is what Grandma Tracy sounded like.
00:11:42.559 --> 00:11:45.600
Uh this is uh you know how language worked back then.
00:11:45.919 --> 00:12:05.039
And you know, as an example, if we discovered a Chat GPT style LLM in Latin, uh I mean it would be the find of the century for many reasons, but from a linguistic standpoint, it would become basically the most important thing that every grad student on the planet would study.
00:12:05.279 --> 00:12:08.799
All the other stuff that we study would become peripheral to that.
00:12:09.120 --> 00:12:10.879
So that would be my goal.
00:12:11.039 --> 00:12:18.480
But uh, you know, it's very resource-intensive, especially for these non languages that don't have vast libraries of content.
00:12:18.639 --> 00:12:20.240
You need to actually collect a lot of data.
00:12:20.320 --> 00:12:23.120
Aaron Powell But how do you find out where these languages are?
00:12:23.200 --> 00:12:36.000
I mean, obviously you work in this space, but you must are you kind of uncovering new or unknown to you languages all the time, or as the world do we know all the languages, or are there still discoveries in language to make?
00:12:36.480 --> 00:12:37.200
Oh, there's so much.
00:12:37.360 --> 00:12:39.600
And and I would challenge everybody out there.
00:12:39.840 --> 00:12:46.960
I guarantee you there's like some niche languages within 50 miles of you that you have no idea exist.
00:12:47.360 --> 00:12:58.000
I tell this to people around me in LA, like, I mean, there's there's half a dozen indigenous languages that are still spoken within 30 miles of Los Angeles.
00:12:58.240 --> 00:13:07.200
Not spoken by lots of people, but there are native speakers, and they'll have like a, you know, a little museum, and there's a you know, the little bit of government money to kind of keep things rolling.
00:13:07.440 --> 00:13:08.960
It's just mostly off the radar.
00:13:09.120 --> 00:13:11.039
And part of that is it's very local.
00:13:11.360 --> 00:13:29.039
So our organization is focused on preservation, meaning we take basically like we're trying to make, you know, we're trying to save this stuff for the future because the worry is that it's gonna get lost, um, which is, you know, a very strong worry.
00:13:29.600 --> 00:13:37.039
Most people in this space are actually doing what's called language revitalization, which is where they're trying to teach new generations to speak it.
00:13:37.200 --> 00:13:43.440
So you have you know five native speakers now, but trying to increase the number of speakers and bring languages back.
00:13:43.919 --> 00:13:45.759
This is happening all over the world.
00:13:45.919 --> 00:13:58.320
Uh you could think of in in England, uh Catalan and and uh Irish are two excellent examples of where they you know kind of brought them back from the brink and now they're effectively like bilingual cultures.
00:13:58.399 --> 00:14:00.080
Aaron Powell Yeah, or Welsh actually here.
00:14:00.399 --> 00:14:00.799
Oh really?
00:14:00.960 --> 00:14:02.080
Okay, I didn't I didn't know quite as much.
00:14:02.639 --> 00:14:14.000
I mean it's not been at the brink, but people are very, you know, as generations get older, they're very keen to make sure that uh you know the signs are in both languages and the Welsh keeps being spoken as we move forward.
00:14:14.159 --> 00:14:14.320
Yeah.
00:14:14.720 --> 00:14:16.159
God, I love that stuff so much too.
00:14:16.320 --> 00:14:21.120
If you walk into a store and it has signs in two languages, I just automatically am like, it's more fun.
00:14:21.440 --> 00:14:26.799
So this is language revitalization, and this is happening all over the world.
00:14:27.039 --> 00:14:28.399
There's amazing stories.
00:14:28.720 --> 00:14:30.879
This happens best at a local level.
00:14:31.039 --> 00:14:35.360
Where it's literally just, you know, the elders teaching the kids, and we're going up to the Arctic.
00:14:35.519 --> 00:14:41.039
One of the things that we're gonna do is visit the uh the Sami in like northern northern Norway.
00:14:41.279 --> 00:14:58.240
We got invited by a researcher up there, and yeah, similar, like it, you know, in the 80s, apparently in the 80s and 90s, uh, the number of Sami speakers was really low, and you know, earlier it had been kind of like, you know, discouraged and everything.
00:14:58.399 --> 00:15:02.240
And then apparently these days it's you know almost bilingual in a lot of places.
00:15:02.480 --> 00:15:11.039
But that can only happen through kind of like slow growth and these like education programs and and parents that want to like speak it at home and things like that.
00:15:11.279 --> 00:15:13.440
So that's revitalization, and that's amazing.
00:15:13.600 --> 00:15:20.080
So we're not doing that ourselves, but we work with a lot of groups.
00:15:20.240 --> 00:15:24.159
I mean, essentially anyone who's doing that, like we are fully in support of.
00:15:24.480 --> 00:15:24.720
Yeah.
00:15:24.960 --> 00:15:42.000
I think it's so important for identity, because this is where the intersection is, I guess, for like the technology and identity and language and society, in that if you don't have access to the the language, if you believe that language defines reality, then you don't have access to your own history in a sense.
00:15:42.240 --> 00:15:58.320
So, you know, then then there's this strange sort of space that we're in, which is very present all the time when things are digitized, like people aren't quite sure the future is very uncertain, some people are better at you know conceptualizing it than the others.
00:15:58.480 --> 00:16:04.159
But the past, the past is constantly being, particularly in the West, denigrated, um, sort of challenged.
00:16:04.240 --> 00:16:04.879
That's fine.
00:16:04.960 --> 00:16:23.600
But you know, if you're going to lose things from your past forever, then how does one sort of have any continuity and self-continuity with your past, either sort of individually or as a as a tribe, or I mean, you know, in the metaphorical sense that as a society, as a culture, you can't reach back into the past.
00:16:23.679 --> 00:16:30.960
And I think this is one of the things that some people are really finding it quite difficult and a challenge to sort of navigate, you know, who are they?
00:16:31.120 --> 00:16:44.000
If you don't have a strong connection to the past for whatever reason, and language is a really important way of doing it, obviously, um, then you're a little bit potentially a little bit lost um knowing your place in the world.
00:16:44.159 --> 00:16:45.600
I don't know if you agree.
00:16:46.320 --> 00:16:47.919
Oh, totally, a hundred percent.
00:16:48.080 --> 00:17:03.200
I mean, I don't think language, I don't go as far to say that language affects, you know, everything about how you think, because there's too much evidence where there's, you know, Spanish speakers who are this way and Spanish speakers who are also that way, and and stuff like that.
00:17:03.360 --> 00:17:08.319
But speaking of though, I mean when I speak Spanish, I personally feel much more extroverted.
00:17:08.720 --> 00:17:10.319
Maybe that's just a me thing.
00:17:10.480 --> 00:17:10.960
I don't know.
00:17:11.119 --> 00:17:14.079
But um, you know, a lot of people make observations like that.
00:17:14.799 --> 00:17:26.720
But I do think there's something to the fact that how you speak and the language you use will affect how you you know, how you think and kind of uh certain tendencies that you have.
00:17:26.960 --> 00:17:36.960
And I also, with you, like I can only speak for myself personally, but a hundred percent I think our connections to languages in the past is like a big deal.
00:17:37.200 --> 00:17:48.559
Like I, you know, I did 23adme and found out I'm 75% Irish, and I don't speak any Irish, but it's still um it made me much more want to learn Irish.
00:17:48.799 --> 00:18:00.079
Like there's such a strong connection between language and culture that to lose that I mean it's a little sad, but it's also like it feels like it's it's missing a big part.
00:18:00.319 --> 00:18:12.079
And another big thing with this project, too, is you know, these tiny little languages that peep that are in danger of just, you know, kind of poof disappearing off the face of the earth because there's so little presence in the first place.
00:18:12.319 --> 00:18:27.440
But if you were the person and you found out, uh, oh, turns out like my grandpa was uh, you know, spoke Luxembourgish, which is a language I actually didn't know existed until a couple weeks ago when we did a, you know, got a little mini library together.
00:18:27.680 --> 00:18:31.279
And then you went to go be like, oh, that's interesting.
00:18:31.440 --> 00:18:32.480
Let's go look that up.
00:18:32.559 --> 00:18:36.559
And then, you know, found that there was some information about Luxembourgish.
00:18:36.640 --> 00:18:43.680
Maybe you don't learn the whole the whole language, but it it becomes another piece of kind of this whole stew, you know.
00:18:43.839 --> 00:18:46.720
It's like knowing your history, like knowing where you came from.
00:18:47.519 --> 00:18:54.160
It's an ancestral connection, I think, that once you cut it, you sort of left uh a little bit adrift potentially.
00:18:54.640 --> 00:19:01.119
It's so interesting you say that about the internet because the internet does have that capability to sort of flatten everything out.
00:19:01.279 --> 00:19:15.039
We've talked about this a lot in terms of identity, that the it kind of creates this homogeneity and anything that's kind of interesting or very unusual is flung to the side kind of sidelines and ends up on the periphery.
00:19:15.359 --> 00:19:19.279
What do you think the internet specifically is doing to language?
00:19:19.440 --> 00:19:23.440
I mean you sort of summarized it there, but what's what's the detail?
00:19:23.599 --> 00:19:24.720
How does it work?
00:19:24.960 --> 00:19:32.319
What happens with either an indigenous language or even in a language that's not, you know, spoken you know majorly?
00:19:32.480 --> 00:19:36.240
What what actually is the effect of the internet or a use of the internet?
00:19:37.119 --> 00:19:39.599
I think you said it proper, flattening.
00:19:39.759 --> 00:19:45.359
Um accent flattening, accents are disappearing like crazy.
00:19:46.240 --> 00:19:50.400
People in this in the states, people in Texas talk a lot like people in California now.
00:19:50.880 --> 00:19:56.880
So one story is I I grew up in Chicago, and uh Chicago has a really strong accent, and I had a really strong accent.
00:19:57.200 --> 00:19:58.720
If you say Chicago, Chicago.
00:19:59.279 --> 00:20:08.720
I moved to California, and about the first six months people pointed out my accent all the time, and I made zero effort to get rid of it.
00:20:08.960 --> 00:20:10.960
I was actually liked my Chicago accent.
00:20:11.279 --> 00:20:13.759
But then after six months, people stopped pointing it out.
00:20:13.920 --> 00:20:21.279
So it's I just naturally kind of like flattened, you know, sort of uh probably lost a couple aspects.
00:20:21.440 --> 00:20:28.480
I definitely stopped saying the word pop and pop machine because people would make fun of me like crazy for saying that for uh you know for a soda.
00:20:29.039 --> 00:20:32.559
And I think that the internet causes that to happen at large scale.
00:20:32.799 --> 00:20:38.880
So flattening and then on the like niche languages, it really disincentivizes.
00:20:39.119 --> 00:20:40.079
There's a couple places.
00:20:40.160 --> 00:20:48.000
I mean, you have like your Facebook groups and you know, there's like uh indigenous language Twitters and all kinds of stuff like that.
00:20:48.160 --> 00:21:04.240
Like there there are places to go, which is great, but I think the overall effect is kind of pressure towards the major languages, putting a lot of pressure towards English here, you know, if you're in Asia, a lot of pressure towards Mandarin, a lot of stuff being done in Mandarin.
00:21:04.559 --> 00:21:05.839
I was going to ask you about that.
00:21:06.000 --> 00:21:10.400
What are the insights about, you know, Asia and Mandarin or what's happening in China?
00:21:10.640 --> 00:21:20.960
Young people used to be told to, well, learn Mandarin because that's going to be the most useful for you, but I'm not sure that's the case anymore because of the potential for even more sort of decoupling between the West and the East.
00:21:21.200 --> 00:21:27.200
I mean, it's always been America is the superpower, speaking English, so that's the major language that everybody automatically learns.
00:21:27.279 --> 00:21:28.960
But is that the case going forward?
00:21:29.759 --> 00:21:34.799
This is a great time for me to make an argument for for learning a niche language too.
00:21:35.200 --> 00:21:38.720
Because personal experience, I lived in Barcelona for a year.
00:21:38.880 --> 00:21:42.240
I was a young student, skateboarder.
00:21:42.559 --> 00:21:43.039
Loved it.
00:21:43.200 --> 00:21:47.440
Barcelona is like possibly the greatest city in the world to live when you're like twenty one years old.
00:21:48.079 --> 00:21:52.480
I did not know that most of my courses would be taught in Catalan.
00:21:52.640 --> 00:21:54.960
And so I had to learn Catalan like really fast.
00:21:55.359 --> 00:21:55.839
But it was great.
00:21:55.920 --> 00:21:56.319
I loved it.
00:21:56.400 --> 00:21:59.440
You know, I tell people that and they're like, uh, what's the point of learning Catalan?
00:21:59.519 --> 00:22:07.759
You know there's Like a billion Spanish speakers, and there's eight million Catalan speakers, or 12 million million Catalan speakers.
00:22:08.000 --> 00:22:14.079
Catalan has been more useful for me in business than Spanish, Italian, and French combined.
00:22:14.400 --> 00:22:30.240
Because every time I meet someone who speaks Catalan and I can like uh you know have a conversation with them, it's so unusual and it's also such a kind of like I mean there's a bit of a compliment in there and things like that that it really has like opened doors and stuff.
00:22:30.400 --> 00:22:35.279
So I would say for people who are trying to decide what language to learn.
00:22:35.519 --> 00:22:39.200
In response to your question, like uh, you know, what are the kids being told?
00:22:39.359 --> 00:22:43.680
Like, you know, French kids back in the day are being told, I learn English because English is the language.
00:22:43.759 --> 00:22:47.759
And then 200 years ago, everyone's being told learn French because it's the lingua franca.
00:22:47.839 --> 00:22:52.799
This is the language of international, you know, politics and trade in Europe and everything.
00:22:53.119 --> 00:22:54.880
I would say don't follow trends.
00:22:55.039 --> 00:23:12.240
Like, I bet a lot of people learned Mandarin who never went to China but, you know, did go on vacation to Germany and actually could have benefited from learning a bunch of German, even if it didn't seem like, oh, it was the quote unquote strategic move.
00:23:12.559 --> 00:23:15.519
Languages isn't really about numbers, it's about meeting individuals.
00:23:15.839 --> 00:23:23.359
So I honestly, for that question, I tell people like, learn the language you're most interested in and the place that you're most interested in going.
00:23:24.240 --> 00:23:37.599
Because there'll be way more advantages if you learn a language that you actually use with people than one that, you know, is sort of uh this like theoretical uh oh maybe this will be good for business one day.
00:23:38.400 --> 00:23:41.920
About ten years ago, people were saying, don't learn a language, learn code.
00:23:42.079 --> 00:23:43.759
Code is the new language.
00:23:43.920 --> 00:23:49.279
What's your I mean, obviously you're working in both areas, you'll you are at the intersection of that.
00:23:49.359 --> 00:23:50.559
What's your point of view on that?
00:23:51.200 --> 00:23:56.160
Kind of a weird thing to say in that there's so there's so the use cases are so different.
00:23:56.559 --> 00:24:03.680
I will say language if you can learn to speak French, you can definitely learn JavaScript really easily.
00:24:03.920 --> 00:24:08.480
Because it is kind of like a uh human language, but without all of the irregular.
00:24:09.039 --> 00:24:12.640
So language people do take to coding like a fish to water.
00:24:12.799 --> 00:24:13.599
But it's totally different.
00:24:13.680 --> 00:24:15.920
I mean, it's t they're totally different jobs.
00:24:16.240 --> 00:24:37.680
What do you think about these um sort of generative AI models that are doing sort of translation on the fly um when you've got your avatar and it's I mean, I don't speak Japanese, but I can easily go and you know log on to Hey Gen and do a little whatever, little presentation, it'll obviously it it claims obviously that it's you it's me speaking Japanese then, but in my own accent.
00:24:38.160 --> 00:24:40.960
What's your view on these sorts of innovations?
00:24:41.279 --> 00:24:54.079
So one is when we make LLMs from these niche languages, the first thing I'm gonna do is try and force companies to put it into their software so that remember I was talking about the pressure, the internet pressure from major languages?
00:24:55.039 --> 00:25:01.680
The AI um you know translation stuff could be the solution to this.
00:25:01.920 --> 00:25:10.880
For my example, if you are living in Brazil and you speak English so that you can get a better job, you know, that's great.
00:25:10.960 --> 00:25:25.839
But if in 10 years we have some auto-translator where you don't need to speak English to do your JIRA tasks and uh talk to your project manager, like it doesn't matter, you're you're you know, you just have it translating for you, all of a sudden that pressure's gone.
00:25:26.000 --> 00:25:30.480
You don't you don't need to have learned the language, um, it's handling it for you.
00:25:30.720 --> 00:25:39.039
You know, and we can argue that, oh, this wouldn't work for sales jobs, and I'm like, well, yeah, it probably wouldn't work for sales jobs because you want to have a personal touch or something.
00:25:39.119 --> 00:25:58.960
But if you're you know, if you're coding up apps and stuff, I could definitely see a world where this like this layer of machine translation, again, just takes off this pressure, and then all of a sudden, you don't need to learn English, but maybe you're gonna learn your parents' language, you know, or or whatever the case might be.
00:25:59.119 --> 00:26:00.880
So I think that that's a really good thing.
00:26:01.039 --> 00:26:18.880
Um, and I think that as we start to see this integrated better throughout society, I think we got to the point where like machine translation just hit the still quite crappy, but barely acceptable for prime time in in some cases.
00:26:19.119 --> 00:26:24.240
But as we know from technology, that usually means that you're about five years from it being pretty damn good.
00:26:24.480 --> 00:26:26.400
So I really like that aspect of it.
00:26:26.559 --> 00:26:34.079
And and so, for example, like if we finished, you know, this uh Navajo LLM, that's what one long-range project we're doing.
00:26:34.240 --> 00:26:39.759
I would love to like put pressure on Netflix to all, you know, be like, dude, we have a translator.
00:26:39.920 --> 00:26:45.440
You don't have, you know, the money makes sense, like go translate your shows right now, you know.
00:26:45.519 --> 00:26:51.279
That and again, it's taking the pressure off, it's making more languages available.
00:26:51.519 --> 00:26:56.799
And I think that you you apply this throughout society, and I think it's an awesome thing.
00:26:57.119 --> 00:27:06.880
My vision for the world is when you walk up to like a McDonald's kiosk or whatever, it has like every language on earth that you can that you can put it in.
00:27:07.119 --> 00:27:09.279
Not because it's absolutely necessary.
00:27:09.440 --> 00:27:17.759
Like I wouldn't argue that's crucial to the functioning of society, but I think it will do a lot to start solving this this problem.
00:27:18.720 --> 00:27:26.960
Maybe in some megacities, though, I mean if mega cities do grow, it's about like 20, 30, there'd be a third more megacities, and they are really populous.
00:27:27.200 --> 00:27:30.559
Maybe maybe it's about the context again, maybe the location.
00:27:30.640 --> 00:27:38.960
You know, you wouldn't have it elsewhere, but maybe in a megacity where people are in and out all the time from lots of different countries, it would be incredibly efficient.
00:27:39.279 --> 00:27:42.400
Is your project global in scope?
00:27:42.799 --> 00:27:43.519
Oh, yeah, yeah.
00:27:43.599 --> 00:28:00.799
If you took uh this deposit we're doing now, or we're gonna do in a couple weeks, if you put it on a map, it would be pretty evenly spread, with actually a a higher concentration in Africa and Latin and South America, because there's just a higher concentration of languages.
00:28:01.119 --> 00:28:02.880
Oh, I wanted to ask you about Africa.
00:28:03.039 --> 00:28:13.920
Because obviously, if you think about sort of the population trends as well, uh obviously the replacement level's so low with the population in the West, but it's really, really high in Africa, something like seven in Nigeria.
00:28:14.160 --> 00:28:26.480
So is that good that surely is going to have an effect, isn't it, on the kinds of languages that become the major languages that are spoken, or will they stay, do you think, in their local environment?
00:28:26.720 --> 00:28:30.640
How did languages sort of move and shift like that?
00:28:31.039 --> 00:28:32.960
I mean, this is a question people debate.
00:28:33.119 --> 00:28:36.720
Like, why, you know, why is English the dominant language?
00:28:36.880 --> 00:28:44.559
Was it because the British, you know, history of colonialism, or people argue it's because English is inherently better.
00:28:44.799 --> 00:28:51.839
Like one fact about English people bring up is English has never not had the highest vocabulary of any language on earth.
00:28:52.079 --> 00:28:55.519
Uh even before, you know, it was the dominant language.
00:28:55.759 --> 00:29:02.400
And so there's people who argue there's something inherent to English that causes it to spread, which I think is like kind of bullshit personally.
00:29:02.480 --> 00:29:04.720
But I I think it's very special and true.
00:29:05.920 --> 00:29:18.079
I mean it's a great but it is, you know, Borges has some famous quote about how English is the language of poetry, because there's so many more words you can use to describe things that are, you know, very descriptive and everything.
00:29:18.319 --> 00:29:24.880
But in terms of population dynamics, the one of the reasons there's so many languages in Africa is because there's so much isolationism.
00:29:25.200 --> 00:29:39.680
I mean, there's obviously trade throughout Africa, but in in a way, if you found a tribe of people who spoke a language no one had ever heard in England in 2024, it would be fairly shocking.
00:29:39.920 --> 00:29:47.599
If you found it in the Congo, it would be less so, you know, just because of the nature of the geography and the politics and everything.
00:29:47.759 --> 00:29:52.559
So yeah, languages in isolation uh go go hand in hand.
00:29:52.880 --> 00:30:06.960
A big question is into our previous thing of if you have, you know, so much uh as as right now it protests in in England, right, about immigration and anti-immigration and anti-anti-immigration or whatever, you know.
00:30:07.200 --> 00:30:13.519
Let's say snap your fingers and England becomes 30% Arabic speaking.
00:30:13.680 --> 00:30:16.000
What what does that mean linguistically?
00:30:16.160 --> 00:30:18.720
Does does it remain English speaking?
00:30:18.880 --> 00:30:20.480
Does it become bilingual?
00:30:20.720 --> 00:30:23.119
Do we start teaching Arabic in schools?
00:30:23.359 --> 00:30:37.920
Does that become just a its own subculture where there's it's more like, oh, there's Arabic private schools and uh, you know, it's sort of the equivalent of like Arabic like Sunday schools and things like that, but you don't actually like learn.
00:30:38.160 --> 00:30:41.039
There's there's a lot of ways these things can go, you know?
00:30:41.200 --> 00:30:50.640
And and authoritarian attempts to tell people how to do language, people don't really like being told how to talk.
00:30:50.799 --> 00:30:54.480
I hate when people tell try and tell me what words to use and stuff like that.
00:30:54.640 --> 00:31:03.839
And if you've seen like Catalan and Basque and all of these, like the top-down stuff doesn't really even in Ukraine, you know, like banning Russian being spoken, things like that.
00:31:03.920 --> 00:31:09.599
I mean, I think wherever that kind of happens, it tends to make the population want to speak it more.
00:31:10.400 --> 00:31:11.680
Just pisses people off.
00:31:12.079 --> 00:31:12.880
Exactly.
00:31:13.119 --> 00:31:15.680
Yeah, because it's an expression of identity.
00:31:16.160 --> 00:31:17.440
So that's kind of my take.
00:31:17.680 --> 00:31:19.920
And then sometimes these things turn into creoles.
00:31:20.079 --> 00:31:25.519
Maybe, maybe in 20, you know, maybe in 50 years you'd have like an English-Arabic new language.
00:31:25.680 --> 00:31:39.599
Like who So I think I think the ability, I think predicting is there's too much, uh too much black swan potential and you know, too much randomness to to make accurate predictions, but enough to make fun predictions.
00:31:39.680 --> 00:31:42.240
You know, this could happen, that could happen, that could happen.
00:31:42.799 --> 00:31:44.799
Yeah, we always love to do our scenarios.
00:31:44.960 --> 00:31:47.920
Now, hey, tell me, what about non-human language?
00:31:48.079 --> 00:31:55.119
What about will I be able to save and then understand or preserve the language of my dogs here?
00:31:55.680 --> 00:31:56.400
I hope so.
00:31:56.559 --> 00:32:12.720
There's a a couple articles that came out from Harvard Business uh Review where they're trying to analyze whale and I I read the paper, and I would say that they're what they're calling whale grammar is a little bit of a stretch.
00:32:13.119 --> 00:32:15.279
I think it would be good if the aliens came down.
00:32:15.599 --> 00:32:18.559
Because at least it would put a little humility into us as a species.
00:32:18.720 --> 00:32:28.000
You know, I think that might be kind of healthy for us to to not be the only, you know, the only uh intelligent uh people around.
00:32:28.160 --> 00:32:30.400
Um and I would love to talk to my dog.
00:32:30.559 --> 00:32:33.200
I I think one of the things there, if we could crack that communication.
00:32:33.440 --> 00:32:54.960
Okay, if I can answer the question semi-seriously, I think it would be brilliant because I think we would realize it would you know, if if all human language is between uh, you know, five and fifteen, it would be interesting to find something that was like at a hundred and fifty.
00:32:55.200 --> 00:33:05.359
You know, that was that was still language but was so far outside of what you know, kinda how we humans think and stuff like that, that it caused I think it would cause us to rethink a lot.
00:33:05.680 --> 00:33:12.160
I'm a bit with um I know Chom everyone likes to disapprove Chomsky with uh some good reason.
00:33:12.240 --> 00:33:16.960
I I love the guy, but uh, you know, not all of his theories have held up, uh have held up perfectly.
00:33:17.200 --> 00:33:28.319
But one thing he said that I do agree with is it if the aliens came and analyzed human language, they wouldn't say we have seven thousand languages.
00:33:28.400 --> 00:33:32.720
They would probably say we have one language with seven thousand accents.
00:33:33.359 --> 00:33:39.279
And I would add on that that I actually think they would include music as part of the uh language spectrum.
00:33:39.519 --> 00:33:42.960
So I argue music came before came before language.
00:33:43.200 --> 00:33:46.640
Well, I argue I think it's almost definitely true.
00:33:46.799 --> 00:34:02.720
Because of the fact that like around the you could play a sad tone for anyone around the world, they'll recognize it as sad, uh, versus you know, a happy upbeat, you know, Beach Boys, they might not love it, they might not, it might not be in their wheelhouse, but they would recognize it as like a happy song, you know.
00:34:02.960 --> 00:34:15.280
So this stuff is built into us, and if you think about like evolutionary terms, uh, you know, birds, things that communicate, all emotional communication is going to be help us as a species over non.
00:34:16.079 --> 00:34:23.760
So even the ability for me to say, I'm sad, I'm happy, you know, to communicate these basic feelings was beneficial.
00:34:23.840 --> 00:34:26.320
And as far as I can tell, this is the origin of music.
00:34:26.559 --> 00:34:31.360
Actually, I mean, music is still insanely like mysterious.
00:34:31.760 --> 00:34:38.000
As an art form, I mean, no other art form grabs us as emotionally, where you can like you know literally feel it in your chest.
00:34:38.480 --> 00:34:40.639
Yeah, connection to the divine, yeah.
00:34:40.800 --> 00:34:43.039
And nobody can explain it beyond that.
00:34:43.360 --> 00:34:50.480
Are there certain languages that either have existed or still exist that are more inherently more emotional than others?
00:34:50.800 --> 00:34:56.800
I'm thinking about, you know, when you hear about things like whether there's 50 words for snow or the Inuits, yeah.
00:34:57.119 --> 00:34:58.159
The Inuits, yeah.
00:34:58.400 --> 00:35:09.440
Are there languages that can't capture certain emotions in the language or certain languages that do have a plethora of different expressions or phrases or words for to capture a certain emotion?
00:35:09.599 --> 00:35:12.480
I was just wondering about the the kind of continuum is for that.
00:35:12.880 --> 00:35:13.920
Oh, absolutely.
00:35:14.159 --> 00:35:19.280
Lots of languages have that fifty words for snow for something.
00:35:19.519 --> 00:35:21.760
This is probably the most famous example.
00:35:22.159 --> 00:35:35.440
Languages that have more colors, i.e., like instead of saying green and blue, but they're like, you know, green, turquoise, aqua, aquamarine, uh, you know, blue, deep blue, you know.
00:35:35.760 --> 00:35:40.239
But for us, we think green and blue, and then those are kind of subcategories.
00:35:40.559 --> 00:35:47.679
There's a lot of languages where they just divide things more for whatever reason, and I'm sure like the the Alaskan Inuits had that with white.
00:35:47.760 --> 00:35:53.679
I bet the word, their words for white, you know, they had one to go with each form of snow, right?
00:35:54.079 --> 00:35:59.280
What has been proven is that they can pick out those colors much, much, much better.
00:35:59.519 --> 00:36:09.760
You know, if you showed them a color spectrum and said, uh, choose all of the different colors, they could, you know, were us, we would look and just see like, well, I see white and then I see gray.
00:36:09.920 --> 00:36:15.519
They could go and say, like, this is that kind of white, that's that kind of white, that's that kind of white, that's that kind of white.
00:36:15.760 --> 00:36:28.719
So if you have a language that emphasizes something, whether it's color and colors is a really common one, you practice it way more and you get way, way, way, way, way better at it.
00:36:29.039 --> 00:36:30.320
Yeah, exactly.
00:36:30.639 --> 00:36:32.639
You mentioned about aliens earlier.
00:36:33.119 --> 00:36:34.480
In jest probably.
00:36:34.639 --> 00:36:38.800
But um, I mean Arrival is one of my well, top three films.
00:36:38.960 --> 00:36:41.039
Oh my god, I could I mean what?
00:36:41.199 --> 00:36:45.599
I mean, amazing story, amazing storyteller, amazing movie made of the story.
00:36:45.679 --> 00:36:48.639
But would you think we would be able to communicate with aliens?
00:36:48.800 --> 00:36:55.199
Or are you even looking at that as the sort of what might be the languages of outer space or interstellar languages?
00:36:55.679 --> 00:36:56.559
Oh, I hope so.
00:36:56.719 --> 00:37:10.320
I think one of the things that movie did really well was uh point out how different and different in one way where first you're like, oh, they don't, you know, they don't speak the language, they draw the language.
00:37:10.559 --> 00:37:31.679
You know, and and there's uh if you read a lot of sci-fi, which it sounds like you do, um you know, there's in science fiction, there's aliens that draw, that use like sense of smell, they emit little pheromones, there's one, you know, that have like little pressure zones, you know, every everything you could possibly imagine to communicate could potentially form a language.
00:37:32.079 --> 00:37:39.519
So I think that that would be incredible to you know, it'd be the the linguistic research project of the century.
00:37:39.760 --> 00:37:45.360
Um and then also I uh back to the like the way they think.
00:37:45.599 --> 00:37:54.239
And this is something a ri I mean Arrival used the time, you know, the essentially like if you speak their language, you can sort of time jump.
00:37:54.880 --> 00:38:06.159
Um which may be a little extreme, but also I think was a really good, if nothing else, a great metaphor for just like how freaking different like this is gonna get.
00:38:06.400 --> 00:38:16.000
Like the aliens came, like it's not gonna be Star Trek where they come down and we're try essentially we're trying to like decode a version of French.
00:38:16.239 --> 00:38:21.199
It's gonna be like they're gonna come down and do something that we don't understand is even language.
00:38:21.920 --> 00:38:31.119
I mean we talk about language, but communication more broadly could be in any way possible and beyond their imagination, I think.
00:38:31.199 --> 00:38:38.000
Um and that's how it's difficult to reach out and try and try and discover it because you wouldn't know where to start.
00:38:38.480 --> 00:38:42.239
What kind of way would you you go about is they're not gonna speak like us.
00:38:42.559 --> 00:38:49.840
What haven't we discussed about your project, Jack, that we need to um uh so that more people can sort of follow it?
00:38:49.920 --> 00:38:54.239
I know you've got a crowdfunder, we did have a crowdfunder, I know you've got investors in it.
00:38:54.320 --> 00:38:55.599
Um, where's it going next?
00:38:55.679 --> 00:38:56.880
What's the timetable for it?
00:38:57.119 --> 00:38:58.320
Yeah, all successful.
00:38:58.559 --> 00:38:59.679
Congratulations, yeah.
00:38:59.920 --> 00:39:03.440
Yeah, so in a couple weeks we head to the Arctic to do the deposit.
00:39:03.599 --> 00:39:06.320
We're making a mini documentary about this.
00:39:06.639 --> 00:39:08.719
Just about endangered languages.
00:39:08.960 --> 00:39:13.599
I would encourage anyone to just investigate languages in your area.
00:39:13.840 --> 00:39:22.719
I would bet everybody out there that there's at least one or two tiny little niche languages within a couple hundred miles that they have no idea exists.
00:39:22.960 --> 00:39:26.239
And for our project, crowdfunding language data.
00:39:26.480 --> 00:39:45.039
If anybody out there speaks or has access to a small or niche language and would like to uh provide, I mean, like literally like cell phone selfie videos and stuff, you know, the easiest stuff in the world, but so that we have audio content.
00:39:45.199 --> 00:39:47.360
So we're always looking for stuff like that.
00:39:47.440 --> 00:39:52.559
And our partner, WikiTongues, and they're part of like the Wikimedia Foundation and all that.
00:39:52.800 --> 00:39:54.320
They host these things.
00:39:54.800 --> 00:39:55.599
So yeah.
00:39:55.760 --> 00:40:01.280
Uh niche language speakers reach out, language dorks uh hit me up on Twitter, love to talk.
00:40:01.440 --> 00:40:05.280
And I mean, honestly, thank you for letting me come on and talk about this.
00:40:05.440 --> 00:40:07.920
Uh like I said, it's a topic not many people even think about.
00:40:08.000 --> 00:40:10.639
So being able to kind of get the word out.
00:40:11.039 --> 00:40:13.920
No, it's so important because time is the issue with this.
00:40:14.079 --> 00:40:15.440
We will run out of time.
00:40:15.599 --> 00:40:16.559
And it's all very well.
00:40:16.639 --> 00:40:21.199
People are great with these um sort of endeavors going, well, we'll do it then, we'll do it then.
00:40:21.440 --> 00:40:22.880
You've got to do it now, haven't you?
00:40:23.039 --> 00:40:24.239
It's urgent and important.
00:40:24.559 --> 00:40:25.039
Absolutely.
00:40:25.119 --> 00:40:26.320
And that's what we're trying to do.
00:40:26.559 --> 00:40:28.239
Well, thank you so much, Tracy.
00:40:29.039 --> 00:40:30.000
Not at all, Jack.
00:40:30.079 --> 00:40:33.199
It's absolutely brilliant to speak to you and fascinating project.
00:40:33.280 --> 00:40:36.079
And we'll follow it along then on the future of you.
00:40:36.320 --> 00:40:46.079
Everyone will follow the progress, and maybe when you come back from the Arctic, you'll come on again and we can um you can update everybody how you got on, but uh, we will follow the projects with interest.
00:40:46.239 --> 00:40:48.320
Thank you so much for your time today.
00:40:53.840 --> 00:40:57.199
Thank you for listening to The Future of You hosted by me, Tracy Follows.
00:40:57.360 --> 00:41:01.280
Be sure to check out the show notes for more info about the topics we covered today.
00:41:01.440 --> 00:41:06.000
If you enjoyed this episode, please like and subscribe wherever you get your podcasts.
00:41:06.159 --> 00:41:09.840
And if you know someone who would love this episode, please share it with them.
00:41:10.000 --> 00:41:15.840
For more on the future of identity in a digital world, visit futuremade.group slash the future of you.
00:41:16.079 --> 00:41:20.960
And to explore the future of everything else, head over to future made.group.
00:41:21.039 --> 00:41:24.719
The Future of View podcast is produced by Big Tent Media.
00:00:19.600 --> 00:00:21.120
Welcome to the future of you.
00:00:21.359 --> 00:00:25.760
Today I'm speaking with Jack Connor, a tech entrepreneur and language enthusiast.
00:00:26.000 --> 00:00:32.240
His latest project aims to deposit endangered language data into the Arctic code vault.
00:00:32.399 --> 00:00:40.560
The project aims to create AI models of vulnerable languages, ensuring cultural and linguistic heritage is preserved for future generations.
00:00:40.880 --> 00:00:54.640
In this chat, he highlights the rapid disappearance of languages due to the globalizing homogenizing effects of the internet and discusses the importance of and challenges with language preservation and how AI is key.
00:00:54.880 --> 00:01:00.079
We also consider language revitalization, with Jack sharing his own personal experiences.
00:01:00.240 --> 00:01:03.920
Jack speaks at least seven languages and has travelled extensively.
00:01:04.079 --> 00:01:13.760
We also go a bit more sci-fi, envisioning real-time AI-driven auto-translation tools that could alleviate the pressure on major languages in the future.
00:01:14.000 --> 00:01:21.280
Of course, language and identity are totally interrelated, so that is always at the heart of this conversation.
00:01:21.599 --> 00:01:28.640
As you listen to this chat, picture Jack in the Arctic, as that is where he will be by the time we go on air.
00:01:28.719 --> 00:01:43.200
And if you'd like to get in touch, help out with the project by providing audio recordings or even cell phone video for the language deposit, or have any endangered languages or dialects that you care deeply about and want to suggest to him.
00:01:43.359 --> 00:01:46.159
I'll leave all the details in the show notes.
00:01:51.280 --> 00:01:53.519
Jack Connor, welcome to the Future of You.
00:01:54.079 --> 00:01:54.480
Thank you.
00:01:54.640 --> 00:01:55.760
Thank you for having me on.
00:01:55.840 --> 00:01:58.719
It's uh it's an honor and uh yeah, great to be here.
00:01:59.040 --> 00:02:06.159
Now it's brilliant to have you here, Jack, because unusually for the guests I have on here, I'm not 100% familiar with your work.
00:02:06.239 --> 00:02:07.680
I mean, I know about it.
00:02:07.920 --> 00:02:15.919
Theo Priestley put me in your direction when we're talking about something on Twitter, and I'm so glad he did because your work is absolutely fascinating.
00:02:16.000 --> 00:02:23.919
And of course, language and the intersection of language and identity is incredibly interesting, nuanced, and complex.
00:02:24.000 --> 00:02:32.080
So I wonder if we'll start off by you just uh explaining to the audience who you are and and and what you're doing and why you're going to the Arctic.
00:02:32.400 --> 00:02:32.800
Mm-hmm.
00:02:33.759 --> 00:02:34.479
Great questions.
00:02:34.719 --> 00:02:35.280
Great questions.
00:02:35.439 --> 00:02:37.039
Well, yeah, thanks for the intro.
00:02:37.120 --> 00:02:48.319
Um I'm Jack Connor, uh based in uh outside Los Angeles, and I'm going to the Arctic to deposit a reel of AI data and language data.
00:02:48.479 --> 00:03:03.439
It's about 550 languages worth of data into this Arctic code vault that's uh, you know, impervious to EMPs, is 250 feet below the permafrost.
00:03:03.919 --> 00:03:08.240
For the listeners and for you, if most people are familiar with the global seed vault.
00:03:08.639 --> 00:03:09.840
It's that for data.
00:03:10.000 --> 00:03:12.639
And they're literally neighbors, they're right next to each other.
00:03:12.719 --> 00:03:15.599
It was it was built as sort of a response to the global seed vault.
00:03:15.840 --> 00:03:16.960
I'm a language lover.
00:03:17.039 --> 00:03:18.240
I've always loved languages.
00:03:18.400 --> 00:03:21.599
I I lived abroad a lot when I was a teenager in my twenties.
00:03:21.840 --> 00:03:24.400
And as an American, this is a little more unusual, I think.
00:03:24.479 --> 00:03:29.599
Uh and so my background is tech, as a lot of people are probably on, like programmer.
00:03:29.680 --> 00:03:37.039
I was I was uh CTO of a couple uh different Silicon Valley startups that did uh pretty good.
00:03:37.360 --> 00:03:40.479
But I always did language stuff as kind of on the side.
00:03:40.639 --> 00:03:45.919
The combination of these two is kind of what got me into language preservation of endangered languages.
00:03:46.159 --> 00:04:03.840
As I continued to work and sort of explore, one of the one of the big topics that I started to take a lot of interest in and didn't see quite so much action around is the fact that languages are disappearing much, much, much faster than at any time in history.
00:04:04.080 --> 00:04:07.520
This is essentially because of the internet, and it's mostly a good thing.
00:04:07.599 --> 00:04:18.480
You know, I had coworkers in Brazil who spoke perfect English because they played Counter-Strike all day or watched Netflix all day and could get jobs that paid way higher than local rates, and it was awesome.
00:04:18.560 --> 00:04:20.560
You know, this is uh globalization.
00:04:20.800 --> 00:04:24.399
There's, you know, there's pros and cons, but uh I think there's a lot of good stuff.
00:04:24.560 --> 00:04:29.920
But one of the negative things is it is quickly pushing out a lot of local languages.
00:04:30.240 --> 00:04:37.279
We're doing a project in Panama right now with uh Ngave, which is a language down there that's spoken by about 200,000 speakers.
00:04:37.519 --> 00:04:43.120
So we're building some tools uh to put in a hospital there because they have a they have a big uh translator shortage.
00:04:43.279 --> 00:04:48.720
You know, people come into the hospital, have a problem, but there's nobody who can do the interpretation work.
00:04:48.800 --> 00:04:52.959
So we're gonna build like an app that has like some text-to-speech and things like that.
00:04:53.199 --> 00:05:09.759
But it's an example of a language, and it 200,000 is actually a lot of speakers, but you have a lot of languages like that where if you're growing up in that culture, is it more valuable to learn English or Spanish, or is it more valuable to learn your local language?
00:05:10.000 --> 00:05:15.040
So of course, like English and Spanish, because you can get jobs, you know, things like that.
00:05:15.279 --> 00:05:28.480
So you just have a lot of pressure on these tiny little niche local languages, uh, you know, of which there's estimated to be 7,000 in the world, and the pressure is, yeah, I mean, inherently causing a lot of them to disappear.
00:05:28.639 --> 00:05:40.959
Uh here in California, there's, you know, many, many, many indigenous languages, but you know, 100 years ago there used to be uh five-ish times as much, or you know, something along those lines.
00:05:41.199 --> 00:05:52.480
And these days, there's there's also a lot of these languages have, you know, five to twenty speakers, and the average age of the native speakers is over 75.
00:05:53.040 --> 00:05:56.000
This is something that I think most people just don't even think about.
00:05:56.160 --> 00:05:56.720
For good reason.
00:05:56.879 --> 00:06:00.399
I mean, you're not but it's something that's really quickly happening.
00:06:00.639 --> 00:06:04.560
It hurts me as someone who just loves languages.
00:06:04.879 --> 00:06:14.720
I uh have lots of uh thoughts about, you know, language and how language affects how we, you know, pattern our thinking and things like that.
00:06:14.959 --> 00:06:22.000
But the other side is uh in a more tangible way, we're starting to do very interesting research.
00:06:22.240 --> 00:06:30.879
We're throwing a bunch of AI at language data and discovering things like whole civilizations that we didn't know existed.
00:06:31.120 --> 00:06:44.480
Because you triangulate from etymologies or you know, finding out that, you know, for a long time there was uh an argument whether Sanskrit and Indo-European, that it was an Indo-European language.
00:06:44.720 --> 00:06:47.279
Linguists don't get mad at me if I'm getting this slightly wrong.
00:06:47.439 --> 00:06:51.680
But basically, you know, trying to figure out was European languages and Sanskrit related?
00:06:51.920 --> 00:06:56.399
And for a long time people said no, and then we found out recently the answer was definitely yes.
00:06:56.639 --> 00:07:10.319
You know, so there's we're sort of at this point when we're opening up all these cool, amazing, like research-oriented uh tools we can use languages for, our languages are starting to disappear.
00:07:10.720 --> 00:07:13.680
How do you save or preserve these languages?
00:07:13.839 --> 00:07:19.360
And how do you sort of turn them into a data in a format that can be preserved?
00:07:19.600 --> 00:07:20.959
Talk us through a bit of the details.
00:07:21.279 --> 00:07:22.319
Because it's fascinating.
00:07:22.720 --> 00:07:23.360
Yeah, yeah.
00:07:23.519 --> 00:07:46.399
I mean, perfect world if you know, if a genie came out of a lamp and said, Jack, like you can have anything you want with this project, I would have it create an AI of every accent and a Chat GPT clone in every language in the world, and uh maybe go back in time and you know get a get a ton of get a ton more.
00:07:46.560 --> 00:07:47.920
This is never gonna happen.
00:07:48.160 --> 00:07:53.199
So for the the preservation is like for a lot of these little languages, there isn't a lot of things.
00:07:53.439 --> 00:07:56.160
Have you heard of the disappearing internet theory?
00:07:56.399 --> 00:08:03.759
It's kind of like a highfalutin sort of title for the fact that essentially just old websites are disappearing.
00:08:04.079 --> 00:08:07.600
Oh, Brian Rommel's been um tweeting this a lot, hasn't he?
00:08:07.759 --> 00:08:12.959
Saying like there's two million, this is fascinating, there's two million academic papers that have just kind of disappeared.
00:08:13.040 --> 00:08:23.360
So if you're following the footnotes, suddenly you come to like the end because there's uh they've been hosted on websites that have disappeared or the or the uh the actual papers have disappeared.
00:08:23.519 --> 00:08:25.759
So yeah, is that is that what you're talking about?
00:08:26.000 --> 00:08:26.399
Mm-hmm.
00:08:26.720 --> 00:08:35.440
Well, with with a lot of niche languages, this is a huge problem because a lot of to give you an example, here in California, there's a language called COWIA.
00:08:35.600 --> 00:08:44.159
There's a couple variations of it, and uh there's a revitalization effort going, but it's like teensy tiny little language with you know a couple dozen speakers.
00:08:44.480 --> 00:09:03.519
But if you go and try and find, you know, like grammar books, uh dictionaries, things like that, there's a couple online and they're being hosted on what is a basically a WordPress site being run by a two-person nonprofit, you know, out in the you know, kind of Joshua Tree Desert.
00:09:04.320 --> 00:09:16.960
And if that goes like if that site goes down, there goes, you know, 85% of of everything in that language, you know, because there's so, so little data for some of these languages.
00:09:17.120 --> 00:09:22.159
And so if you're looking on a time frame of you know, five years, it might still be there.
00:09:22.399 --> 00:09:27.600
Fifty years, probably not, like 500 years, like really, like who knows.
00:09:27.919 --> 00:09:36.960
So for some of these small languages, and you know, I said on this, and this first deposit we're doing is 554 languages worth of data.
00:09:37.200 --> 00:09:41.919
For some of these, it's just you know, PDFs of dictionaries, grammar books.
00:09:42.080 --> 00:09:50.960
Uh we we have a partner named WikiTongues, who has uh probably the biggest collection in the world of niche language audio and video.
00:09:51.919 --> 00:10:12.000
So if nothing else, you know, if uh Tracy, if you, you know, if you spoke, uh were one of the last speakers of you know a very niche language, if nothing else, we want to get like audio, some grammar, um, a dictionary, and then any, you know, basically any like papers or information I could find.
00:10:12.159 --> 00:10:16.399
You know, huge research project, but for some of these, there's just there's just not much.
00:10:16.559 --> 00:10:23.679
And most languages are oral languages too, meaning they really haven't left a lot of physical data.
00:10:23.919 --> 00:10:30.240
But we get what we can, you know, because again, from uh like a linguist standpoint, it's just it's important.
00:10:30.480 --> 00:10:41.679
Like one of the sad things is if you have an oral language that doesn't really have any written content and the last speaker dies, like that's it.
00:10:41.919 --> 00:10:49.919
Like there's there is no coming back, it is gone from the face of the earth, and you will never be able to recreate it ever again.
00:10:50.159 --> 00:11:13.519
So if nothing else, I want to make sure we got representation for that, you know, for those, hence the the breadth of this project, because we were able to get, you know, there's a lot of languages where we effectively grabbed everything that was on the internet, but everything that was on the internet is, you know, maybe uh a couple uh couple hundred-page PDFs and some some white papers uh from some linguists and stuff like that.
00:11:13.919 --> 00:11:36.639
But then on the other side, like on the, you know, for some of the languages that have a lot more, have more extensive things, the goal is honestly to build AI so that our children's, children's, children, children could be great as they could like you know, knock on some box and some AI genie pops out and talks to them the way they talk to us, or you know, some kind of interface or you know, whatever the case may be.
00:11:36.799 --> 00:11:42.399
But basically so that they could know, oh, this is what Grandma Tracy sounded like.
00:11:42.559 --> 00:11:45.600
Uh this is uh you know how language worked back then.
00:11:45.919 --> 00:12:05.039
And you know, as an example, if we discovered a Chat GPT style LLM in Latin, uh I mean it would be the find of the century for many reasons, but from a linguistic standpoint, it would become basically the most important thing that every grad student on the planet would study.
00:12:05.279 --> 00:12:08.799
All the other stuff that we study would become peripheral to that.
00:12:09.120 --> 00:12:10.879
So that would be my goal.
00:12:11.039 --> 00:12:18.480
But uh, you know, it's very resource-intensive, especially for these non languages that don't have vast libraries of content.
00:12:18.639 --> 00:12:20.240
You need to actually collect a lot of data.
00:12:20.320 --> 00:12:23.120
Aaron Powell But how do you find out where these languages are?
00:12:23.200 --> 00:12:36.000
I mean, obviously you work in this space, but you must are you kind of uncovering new or unknown to you languages all the time, or as the world do we know all the languages, or are there still discoveries in language to make?
00:12:36.480 --> 00:12:37.200
Oh, there's so much.
00:12:37.360 --> 00:12:39.600
And and I would challenge everybody out there.
00:12:39.840 --> 00:12:46.960
I guarantee you there's like some niche languages within 50 miles of you that you have no idea exist.
00:12:47.360 --> 00:12:58.000
I tell this to people around me in LA, like, I mean, there's there's half a dozen indigenous languages that are still spoken within 30 miles of Los Angeles.
00:12:58.240 --> 00:13:07.200
Not spoken by lots of people, but there are native speakers, and they'll have like a, you know, a little museum, and there's a you know, the little bit of government money to kind of keep things rolling.
00:13:07.440 --> 00:13:08.960
It's just mostly off the radar.
00:13:09.120 --> 00:13:11.039
And part of that is it's very local.
00:13:11.360 --> 00:13:29.039
So our organization is focused on preservation, meaning we take basically like we're trying to make, you know, we're trying to save this stuff for the future because the worry is that it's gonna get lost, um, which is, you know, a very strong worry.
00:13:29.600 --> 00:13:37.039
Most people in this space are actually doing what's called language revitalization, which is where they're trying to teach new generations to speak it.
00:13:37.200 --> 00:13:43.440
So you have you know five native speakers now, but trying to increase the number of speakers and bring languages back.
00:13:43.919 --> 00:13:45.759
This is happening all over the world.
00:13:45.919 --> 00:13:58.320
Uh you could think of in in England, uh Catalan and and uh Irish are two excellent examples of where they you know kind of brought them back from the brink and now they're effectively like bilingual cultures.
00:13:58.399 --> 00:14:00.080
Aaron Powell Yeah, or Welsh actually here.
00:14:00.399 --> 00:14:00.799
Oh really?
00:14:00.960 --> 00:14:02.080
Okay, I didn't I didn't know quite as much.
00:14:02.639 --> 00:14:14.000
I mean it's not been at the brink, but people are very, you know, as generations get older, they're very keen to make sure that uh you know the signs are in both languages and the Welsh keeps being spoken as we move forward.
00:14:14.159 --> 00:14:14.320
Yeah.
00:14:14.720 --> 00:14:16.159
God, I love that stuff so much too.
00:14:16.320 --> 00:14:21.120
If you walk into a store and it has signs in two languages, I just automatically am like, it's more fun.
00:14:21.440 --> 00:14:26.799
So this is language revitalization, and this is happening all over the world.
00:14:27.039 --> 00:14:28.399
There's amazing stories.
00:14:28.720 --> 00:14:30.879
This happens best at a local level.
00:14:31.039 --> 00:14:35.360
Where it's literally just, you know, the elders teaching the kids, and we're going up to the Arctic.
00:14:35.519 --> 00:14:41.039
One of the things that we're gonna do is visit the uh the Sami in like northern northern Norway.
00:14:41.279 --> 00:14:58.240
We got invited by a researcher up there, and yeah, similar, like it, you know, in the 80s, apparently in the 80s and 90s, uh, the number of Sami speakers was really low, and you know, earlier it had been kind of like, you know, discouraged and everything.
00:14:58.399 --> 00:15:02.240
And then apparently these days it's you know almost bilingual in a lot of places.
00:15:02.480 --> 00:15:11.039
But that can only happen through kind of like slow growth and these like education programs and and parents that want to like speak it at home and things like that.
00:15:11.279 --> 00:15:13.440
So that's revitalization, and that's amazing.
00:15:13.600 --> 00:15:20.080
So we're not doing that ourselves, but we work with a lot of groups.
00:15:20.240 --> 00:15:24.159
I mean, essentially anyone who's doing that, like we are fully in support of.
00:15:24.480 --> 00:15:24.720
Yeah.
00:15:24.960 --> 00:15:42.000
I think it's so important for identity, because this is where the intersection is, I guess, for like the technology and identity and language and society, in that if you don't have access to the the language, if you believe that language defines reality, then you don't have access to your own history in a sense.
00:15:42.240 --> 00:15:58.320
So, you know, then then there's this strange sort of space that we're in, which is very present all the time when things are digitized, like people aren't quite sure the future is very uncertain, some people are better at you know conceptualizing it than the others.
00:15:58.480 --> 00:16:04.159
But the past, the past is constantly being, particularly in the West, denigrated, um, sort of challenged.
00:16:04.240 --> 00:16:04.879
That's fine.
00:16:04.960 --> 00:16:23.600
But you know, if you're going to lose things from your past forever, then how does one sort of have any continuity and self-continuity with your past, either sort of individually or as a as a tribe, or I mean, you know, in the metaphorical sense that as a society, as a culture, you can't reach back into the past.
00:16:23.679 --> 00:16:30.960
And I think this is one of the things that some people are really finding it quite difficult and a challenge to sort of navigate, you know, who are they?
00:16:31.120 --> 00:16:44.000
If you don't have a strong connection to the past for whatever reason, and language is a really important way of doing it, obviously, um, then you're a little bit potentially a little bit lost um knowing your place in the world.
00:16:44.159 --> 00:16:45.600
I don't know if you agree.
00:16:46.320 --> 00:16:47.919
Oh, totally, a hundred percent.
00:16:48.080 --> 00:17:03.200
I mean, I don't think language, I don't go as far to say that language affects, you know, everything about how you think, because there's too much evidence where there's, you know, Spanish speakers who are this way and Spanish speakers who are also that way, and and stuff like that.
00:17:03.360 --> 00:17:08.319
But speaking of though, I mean when I speak Spanish, I personally feel much more extroverted.
00:17:08.720 --> 00:17:10.319
Maybe that's just a me thing.
00:17:10.480 --> 00:17:10.960
I don't know.
00:17:11.119 --> 00:17:14.079
But um, you know, a lot of people make observations like that.
00:17:14.799 --> 00:17:26.720
But I do think there's something to the fact that how you speak and the language you use will affect how you you know, how you think and kind of uh certain tendencies that you have.
00:17:26.960 --> 00:17:36.960
And I also, with you, like I can only speak for myself personally, but a hundred percent I think our connections to languages in the past is like a big deal.
00:17:37.200 --> 00:17:48.559
Like I, you know, I did 23adme and found out I'm 75% Irish, and I don't speak any Irish, but it's still um it made me much more want to learn Irish.
00:17:48.799 --> 00:18:00.079
Like there's such a strong connection between language and culture that to lose that I mean it's a little sad, but it's also like it feels like it's it's missing a big part.
00:18:00.319 --> 00:18:12.079
And another big thing with this project, too, is you know, these tiny little languages that peep that are in danger of just, you know, kind of poof disappearing off the face of the earth because there's so little presence in the first place.
00:18:12.319 --> 00:18:27.440
But if you were the person and you found out, uh, oh, turns out like my grandpa was uh, you know, spoke Luxembourgish, which is a language I actually didn't know existed until a couple weeks ago when we did a, you know, got a little mini library together.
00:18:27.680 --> 00:18:31.279
And then you went to go be like, oh, that's interesting.
00:18:31.440 --> 00:18:32.480
Let's go look that up.
00:18:32.559 --> 00:18:36.559
And then, you know, found that there was some information about Luxembourgish.
00:18:36.640 --> 00:18:43.680
Maybe you don't learn the whole the whole language, but it it becomes another piece of kind of this whole stew, you know.
00:18:43.839 --> 00:18:46.720
It's like knowing your history, like knowing where you came from.
00:18:47.519 --> 00:18:54.160
It's an ancestral connection, I think, that once you cut it, you sort of left uh a little bit adrift potentially.
00:18:54.640 --> 00:19:01.119
It's so interesting you say that about the internet because the internet does have that capability to sort of flatten everything out.
00:19:01.279 --> 00:19:15.039
We've talked about this a lot in terms of identity, that the it kind of creates this homogeneity and anything that's kind of interesting or very unusual is flung to the side kind of sidelines and ends up on the periphery.
00:19:15.359 --> 00:19:19.279
What do you think the internet specifically is doing to language?
00:19:19.440 --> 00:19:23.440
I mean you sort of summarized it there, but what's what's the detail?
00:19:23.599 --> 00:19:24.720
How does it work?
00:19:24.960 --> 00:19:32.319
What happens with either an indigenous language or even in a language that's not, you know, spoken you know majorly?
00:19:32.480 --> 00:19:36.240
What what actually is the effect of the internet or a use of the internet?
00:19:37.119 --> 00:19:39.599
I think you said it proper, flattening.
00:19:39.759 --> 00:19:45.359
Um accent flattening, accents are disappearing like crazy.
00:19:46.240 --> 00:19:50.400
People in this in the states, people in Texas talk a lot like people in California now.
00:19:50.880 --> 00:19:56.880
So one story is I I grew up in Chicago, and uh Chicago has a really strong accent, and I had a really strong accent.
00:19:57.200 --> 00:19:58.720
If you say Chicago, Chicago.
00:19:59.279 --> 00:20:08.720
I moved to California, and about the first six months people pointed out my accent all the time, and I made zero effort to get rid of it.
00:20:08.960 --> 00:20:10.960
I was actually liked my Chicago accent.
00:20:11.279 --> 00:20:13.759
But then after six months, people stopped pointing it out.
00:20:13.920 --> 00:20:21.279
So it's I just naturally kind of like flattened, you know, sort of uh probably lost a couple aspects.
00:20:21.440 --> 00:20:28.480
I definitely stopped saying the word pop and pop machine because people would make fun of me like crazy for saying that for uh you know for a soda.
00:20:29.039 --> 00:20:32.559
And I think that the internet causes that to happen at large scale.
00:20:32.799 --> 00:20:38.880
So flattening and then on the like niche languages, it really disincentivizes.
00:20:39.119 --> 00:20:40.079
There's a couple places.
00:20:40.160 --> 00:20:48.000
I mean, you have like your Facebook groups and you know, there's like uh indigenous language Twitters and all kinds of stuff like that.
00:20:48.160 --> 00:21:04.240
Like there there are places to go, which is great, but I think the overall effect is kind of pressure towards the major languages, putting a lot of pressure towards English here, you know, if you're in Asia, a lot of pressure towards Mandarin, a lot of stuff being done in Mandarin.
00:21:04.559 --> 00:21:05.839
I was going to ask you about that.
00:21:06.000 --> 00:21:10.400
What are the insights about, you know, Asia and Mandarin or what's happening in China?
00:21:10.640 --> 00:21:20.960
Young people used to be told to, well, learn Mandarin because that's going to be the most useful for you, but I'm not sure that's the case anymore because of the potential for even more sort of decoupling between the West and the East.
00:21:21.200 --> 00:21:27.200
I mean, it's always been America is the superpower, speaking English, so that's the major language that everybody automatically learns.
00:21:27.279 --> 00:21:28.960
But is that the case going forward?
00:21:29.759 --> 00:21:34.799
This is a great time for me to make an argument for for learning a niche language too.
00:21:35.200 --> 00:21:38.720
Because personal experience, I lived in Barcelona for a year.
00:21:38.880 --> 00:21:42.240
I was a young student, skateboarder.
00:21:42.559 --> 00:21:43.039
Loved it.
00:21:43.200 --> 00:21:47.440
Barcelona is like possibly the greatest city in the world to live when you're like twenty one years old.
00:21:48.079 --> 00:21:52.480
I did not know that most of my courses would be taught in Catalan.
00:21:52.640 --> 00:21:54.960
And so I had to learn Catalan like really fast.
00:21:55.359 --> 00:21:55.839
But it was great.
00:21:55.920 --> 00:21:56.319
I loved it.
00:21:56.400 --> 00:21:59.440
You know, I tell people that and they're like, uh, what's the point of learning Catalan?
00:21:59.519 --> 00:22:07.759
You know there's Like a billion Spanish speakers, and there's eight million Catalan speakers, or 12 million million Catalan speakers.
00:22:08.000 --> 00:22:14.079
Catalan has been more useful for me in business than Spanish, Italian, and French combined.
00:22:14.400 --> 00:22:30.240
Because every time I meet someone who speaks Catalan and I can like uh you know have a conversation with them, it's so unusual and it's also such a kind of like I mean there's a bit of a compliment in there and things like that that it really has like opened doors and stuff.
00:22:30.400 --> 00:22:35.279
So I would say for people who are trying to decide what language to learn.
00:22:35.519 --> 00:22:39.200
In response to your question, like uh, you know, what are the kids being told?
00:22:39.359 --> 00:22:43.680
Like, you know, French kids back in the day are being told, I learn English because English is the language.
00:22:43.759 --> 00:22:47.759
And then 200 years ago, everyone's being told learn French because it's the lingua franca.
00:22:47.839 --> 00:22:52.799
This is the language of international, you know, politics and trade in Europe and everything.
00:22:53.119 --> 00:22:54.880
I would say don't follow trends.
00:22:55.039 --> 00:23:12.240
Like, I bet a lot of people learned Mandarin who never went to China but, you know, did go on vacation to Germany and actually could have benefited from learning a bunch of German, even if it didn't seem like, oh, it was the quote unquote strategic move.
00:23:12.559 --> 00:23:15.519
Languages isn't really about numbers, it's about meeting individuals.
00:23:15.839 --> 00:23:23.359
So I honestly, for that question, I tell people like, learn the language you're most interested in and the place that you're most interested in going.
00:23:24.240 --> 00:23:37.599
Because there'll be way more advantages if you learn a language that you actually use with people than one that, you know, is sort of uh this like theoretical uh oh maybe this will be good for business one day.
00:23:38.400 --> 00:23:41.920
About ten years ago, people were saying, don't learn a language, learn code.
00:23:42.079 --> 00:23:43.759
Code is the new language.
00:23:43.920 --> 00:23:49.279
What's your I mean, obviously you're working in both areas, you'll you are at the intersection of that.
00:23:49.359 --> 00:23:50.559
What's your point of view on that?
00:23:51.200 --> 00:23:56.160
Kind of a weird thing to say in that there's so there's so the use cases are so different.
00:23:56.559 --> 00:24:03.680
I will say language if you can learn to speak French, you can definitely learn JavaScript really easily.
00:24:03.920 --> 00:24:08.480
Because it is kind of like a uh human language, but without all of the irregular.
00:24:09.039 --> 00:24:12.640
So language people do take to coding like a fish to water.
00:24:12.799 --> 00:24:13.599
But it's totally different.
00:24:13.680 --> 00:24:15.920
I mean, it's t they're totally different jobs.
00:24:16.240 --> 00:24:37.680
What do you think about these um sort of generative AI models that are doing sort of translation on the fly um when you've got your avatar and it's I mean, I don't speak Japanese, but I can easily go and you know log on to Hey Gen and do a little whatever, little presentation, it'll obviously it it claims obviously that it's you it's me speaking Japanese then, but in my own accent.
00:24:38.160 --> 00:24:40.960
What's your view on these sorts of innovations?
00:24:41.279 --> 00:24:54.079
So one is when we make LLMs from these niche languages, the first thing I'm gonna do is try and force companies to put it into their software so that remember I was talking about the pressure, the internet pressure from major languages?
00:24:55.039 --> 00:25:01.680
The AI um you know translation stuff could be the solution to this.
00:25:01.920 --> 00:25:10.880
For my example, if you are living in Brazil and you speak English so that you can get a better job, you know, that's great.
00:25:10.960 --> 00:25:25.839
But if in 10 years we have some auto-translator where you don't need to speak English to do your JIRA tasks and uh talk to your project manager, like it doesn't matter, you're you're you know, you just have it translating for you, all of a sudden that pressure's gone.
00:25:26.000 --> 00:25:30.480
You don't you don't need to have learned the language, um, it's handling it for you.
00:25:30.720 --> 00:25:39.039
You know, and we can argue that, oh, this wouldn't work for sales jobs, and I'm like, well, yeah, it probably wouldn't work for sales jobs because you want to have a personal touch or something.
00:25:39.119 --> 00:25:58.960
But if you're you know, if you're coding up apps and stuff, I could definitely see a world where this like this layer of machine translation, again, just takes off this pressure, and then all of a sudden, you don't need to learn English, but maybe you're gonna learn your parents' language, you know, or or whatever the case might be.
00:25:59.119 --> 00:26:00.880
So I think that that's a really good thing.
00:26:01.039 --> 00:26:18.880
Um, and I think that as we start to see this integrated better throughout society, I think we got to the point where like machine translation just hit the still quite crappy, but barely acceptable for prime time in in some cases.
00:26:19.119 --> 00:26:24.240
But as we know from technology, that usually means that you're about five years from it being pretty damn good.
00:26:24.480 --> 00:26:26.400
So I really like that aspect of it.
00:26:26.559 --> 00:26:34.079
And and so, for example, like if we finished, you know, this uh Navajo LLM, that's what one long-range project we're doing.
00:26:34.240 --> 00:26:39.759
I would love to like put pressure on Netflix to all, you know, be like, dude, we have a translator.
00:26:39.920 --> 00:26:45.440
You don't have, you know, the money makes sense, like go translate your shows right now, you know.
00:26:45.519 --> 00:26:51.279
That and again, it's taking the pressure off, it's making more languages available.
00:26:51.519 --> 00:26:56.799
And I think that you you apply this throughout society, and I think it's an awesome thing.
00:26:57.119 --> 00:27:06.880
My vision for the world is when you walk up to like a McDonald's kiosk or whatever, it has like every language on earth that you can that you can put it in.
00:27:07.119 --> 00:27:09.279
Not because it's absolutely necessary.
00:27:09.440 --> 00:27:17.759
Like I wouldn't argue that's crucial to the functioning of society, but I think it will do a lot to start solving this this problem.
00:27:18.720 --> 00:27:26.960
Maybe in some megacities, though, I mean if mega cities do grow, it's about like 20, 30, there'd be a third more megacities, and they are really populous.
00:27:27.200 --> 00:27:30.559
Maybe maybe it's about the context again, maybe the location.
00:27:30.640 --> 00:27:38.960
You know, you wouldn't have it elsewhere, but maybe in a megacity where people are in and out all the time from lots of different countries, it would be incredibly efficient.
00:27:39.279 --> 00:27:42.400
Is your project global in scope?
00:27:42.799 --> 00:27:43.519
Oh, yeah, yeah.
00:27:43.599 --> 00:28:00.799
If you took uh this deposit we're doing now, or we're gonna do in a couple weeks, if you put it on a map, it would be pretty evenly spread, with actually a a higher concentration in Africa and Latin and South America, because there's just a higher concentration of languages.
00:28:01.119 --> 00:28:02.880
Oh, I wanted to ask you about Africa.
00:28:03.039 --> 00:28:13.920
Because obviously, if you think about sort of the population trends as well, uh obviously the replacement level's so low with the population in the West, but it's really, really high in Africa, something like seven in Nigeria.
00:28:14.160 --> 00:28:26.480
So is that good that surely is going to have an effect, isn't it, on the kinds of languages that become the major languages that are spoken, or will they stay, do you think, in their local environment?
00:28:26.720 --> 00:28:30.640
How did languages sort of move and shift like that?
00:28:31.039 --> 00:28:32.960
I mean, this is a question people debate.
00:28:33.119 --> 00:28:36.720
Like, why, you know, why is English the dominant language?
00:28:36.880 --> 00:28:44.559
Was it because the British, you know, history of colonialism, or people argue it's because English is inherently better.
00:28:44.799 --> 00:28:51.839
Like one fact about English people bring up is English has never not had the highest vocabulary of any language on earth.
00:28:52.079 --> 00:28:55.519
Uh even before, you know, it was the dominant language.
00:28:55.759 --> 00:29:02.400
And so there's people who argue there's something inherent to English that causes it to spread, which I think is like kind of bullshit personally.
00:29:02.480 --> 00:29:04.720
But I I think it's very special and true.
00:29:05.920 --> 00:29:18.079
I mean it's a great but it is, you know, Borges has some famous quote about how English is the language of poetry, because there's so many more words you can use to describe things that are, you know, very descriptive and everything.
00:29:18.319 --> 00:29:24.880
But in terms of population dynamics, the one of the reasons there's so many languages in Africa is because there's so much isolationism.
00:29:25.200 --> 00:29:39.680
I mean, there's obviously trade throughout Africa, but in in a way, if you found a tribe of people who spoke a language no one had ever heard in England in 2024, it would be fairly shocking.
00:29:39.920 --> 00:29:47.599
If you found it in the Congo, it would be less so, you know, just because of the nature of the geography and the politics and everything.
00:29:47.759 --> 00:29:52.559
So yeah, languages in isolation uh go go hand in hand.
00:29:52.880 --> 00:30:06.960
A big question is into our previous thing of if you have, you know, so much uh as as right now it protests in in England, right, about immigration and anti-immigration and anti-anti-immigration or whatever, you know.
00:30:07.200 --> 00:30:13.519
Let's say snap your fingers and England becomes 30% Arabic speaking.
00:30:13.680 --> 00:30:16.000
What what does that mean linguistically?
00:30:16.160 --> 00:30:18.720
Does does it remain English speaking?
00:30:18.880 --> 00:30:20.480
Does it become bilingual?
00:30:20.720 --> 00:30:23.119
Do we start teaching Arabic in schools?
00:30:23.359 --> 00:30:37.920
Does that become just a its own subculture where there's it's more like, oh, there's Arabic private schools and uh, you know, it's sort of the equivalent of like Arabic like Sunday schools and things like that, but you don't actually like learn.
00:30:38.160 --> 00:30:41.039
There's there's a lot of ways these things can go, you know?
00:30:41.200 --> 00:30:50.640
And and authoritarian attempts to tell people how to do language, people don't really like being told how to talk.
00:30:50.799 --> 00:30:54.480
I hate when people tell try and tell me what words to use and stuff like that.
00:30:54.640 --> 00:31:03.839
And if you've seen like Catalan and Basque and all of these, like the top-down stuff doesn't really even in Ukraine, you know, like banning Russian being spoken, things like that.
00:31:03.920 --> 00:31:09.599
I mean, I think wherever that kind of happens, it tends to make the population want to speak it more.
00:31:10.400 --> 00:31:11.680
Just pisses people off.
00:31:12.079 --> 00:31:12.880
Exactly.
00:31:13.119 --> 00:31:15.680
Yeah, because it's an expression of identity.
00:31:16.160 --> 00:31:17.440
So that's kind of my take.
00:31:17.680 --> 00:31:19.920
And then sometimes these things turn into creoles.
00:31:20.079 --> 00:31:25.519
Maybe, maybe in 20, you know, maybe in 50 years you'd have like an English-Arabic new language.
00:31:25.680 --> 00:31:39.599
Like who So I think I think the ability, I think predicting is there's too much, uh too much black swan potential and you know, too much randomness to to make accurate predictions, but enough to make fun predictions.
00:31:39.680 --> 00:31:42.240
You know, this could happen, that could happen, that could happen.
00:31:42.799 --> 00:31:44.799
Yeah, we always love to do our scenarios.
00:31:44.960 --> 00:31:47.920
Now, hey, tell me, what about non-human language?
00:31:48.079 --> 00:31:55.119
What about will I be able to save and then understand or preserve the language of my dogs here?
00:31:55.680 --> 00:31:56.400
I hope so.
00:31:56.559 --> 00:32:12.720
There's a a couple articles that came out from Harvard Business uh Review where they're trying to analyze whale and I I read the paper, and I would say that they're what they're calling whale grammar is a little bit of a stretch.
00:32:13.119 --> 00:32:15.279
I think it would be good if the aliens came down.
00:32:15.599 --> 00:32:18.559
Because at least it would put a little humility into us as a species.
00:32:18.720 --> 00:32:28.000
You know, I think that might be kind of healthy for us to to not be the only, you know, the only uh intelligent uh people around.
00:32:28.160 --> 00:32:30.400
Um and I would love to talk to my dog.
00:32:30.559 --> 00:32:33.200
I I think one of the things there, if we could crack that communication.
00:32:33.440 --> 00:32:54.960
Okay, if I can answer the question semi-seriously, I think it would be brilliant because I think we would realize it would you know, if if all human language is between uh, you know, five and fifteen, it would be interesting to find something that was like at a hundred and fifty.
00:32:55.200 --> 00:33:05.359
You know, that was that was still language but was so far outside of what you know, kinda how we humans think and stuff like that, that it caused I think it would cause us to rethink a lot.
00:33:05.680 --> 00:33:12.160
I'm a bit with um I know Chom everyone likes to disapprove Chomsky with uh some good reason.
00:33:12.240 --> 00:33:16.960
I I love the guy, but uh, you know, not all of his theories have held up, uh have held up perfectly.
00:33:17.200 --> 00:33:28.319
But one thing he said that I do agree with is it if the aliens came and analyzed human language, they wouldn't say we have seven thousand languages.
00:33:28.400 --> 00:33:32.720
They would probably say we have one language with seven thousand accents.
00:33:33.359 --> 00:33:39.279
And I would add on that that I actually think they would include music as part of the uh language spectrum.
00:33:39.519 --> 00:33:42.960
So I argue music came before came before language.
00:33:43.200 --> 00:33:46.640
Well, I argue I think it's almost definitely true.
00:33:46.799 --> 00:34:02.720
Because of the fact that like around the you could play a sad tone for anyone around the world, they'll recognize it as sad, uh, versus you know, a happy upbeat, you know, Beach Boys, they might not love it, they might not, it might not be in their wheelhouse, but they would recognize it as like a happy song, you know.
00:34:02.960 --> 00:34:15.280
So this stuff is built into us, and if you think about like evolutionary terms, uh, you know, birds, things that communicate, all emotional communication is going to be help us as a species over non.
00:34:16.079 --> 00:34:23.760
So even the ability for me to say, I'm sad, I'm happy, you know, to communicate these basic feelings was beneficial.
00:34:23.840 --> 00:34:26.320
And as far as I can tell, this is the origin of music.
00:34:26.559 --> 00:34:31.360
Actually, I mean, music is still insanely like mysterious.
00:34:31.760 --> 00:34:38.000
As an art form, I mean, no other art form grabs us as emotionally, where you can like you know literally feel it in your chest.
00:34:38.480 --> 00:34:40.639
Yeah, connection to the divine, yeah.
00:34:40.800 --> 00:34:43.039
And nobody can explain it beyond that.
00:34:43.360 --> 00:34:50.480
Are there certain languages that either have existed or still exist that are more inherently more emotional than others?
00:34:50.800 --> 00:34:56.800
I'm thinking about, you know, when you hear about things like whether there's 50 words for snow or the Inuits, yeah.
00:34:57.119 --> 00:34:58.159
The Inuits, yeah.
00:34:58.400 --> 00:35:09.440
Are there languages that can't capture certain emotions in the language or certain languages that do have a plethora of different expressions or phrases or words for to capture a certain emotion?
00:35:09.599 --> 00:35:12.480
I was just wondering about the the kind of continuum is for that.
00:35:12.880 --> 00:35:13.920
Oh, absolutely.
00:35:14.159 --> 00:35:19.280
Lots of languages have that fifty words for snow for something.
00:35:19.519 --> 00:35:21.760
This is probably the most famous example.
00:35:22.159 --> 00:35:35.440
Languages that have more colors, i.e., like instead of saying green and blue, but they're like, you know, green, turquoise, aqua, aquamarine, uh, you know, blue, deep blue, you know.
00:35:35.760 --> 00:35:40.239
But for us, we think green and blue, and then those are kind of subcategories.
00:35:40.559 --> 00:35:47.679
There's a lot of languages where they just divide things more for whatever reason, and I'm sure like the the Alaskan Inuits had that with white.
00:35:47.760 --> 00:35:53.679
I bet the word, their words for white, you know, they had one to go with each form of snow, right?
00:35:54.079 --> 00:35:59.280
What has been proven is that they can pick out those colors much, much, much better.
00:35:59.519 --> 00:36:09.760
You know, if you showed them a color spectrum and said, uh, choose all of the different colors, they could, you know, were us, we would look and just see like, well, I see white and then I see gray.
00:36:09.920 --> 00:36:15.519
They could go and say, like, this is that kind of white, that's that kind of white, that's that kind of white, that's that kind of white.
00:36:15.760 --> 00:36:28.719
So if you have a language that emphasizes something, whether it's color and colors is a really common one, you practice it way more and you get way, way, way, way, way better at it.
00:36:29.039 --> 00:36:30.320
Yeah, exactly.
00:36:30.639 --> 00:36:32.639
You mentioned about aliens earlier.
00:36:33.119 --> 00:36:34.480
In jest probably.
00:36:34.639 --> 00:36:38.800
But um, I mean Arrival is one of my well, top three films.
00:36:38.960 --> 00:36:41.039
Oh my god, I could I mean what?
00:36:41.199 --> 00:36:45.599
I mean, amazing story, amazing storyteller, amazing movie made of the story.
00:36:45.679 --> 00:36:48.639
But would you think we would be able to communicate with aliens?
00:36:48.800 --> 00:36:55.199
Or are you even looking at that as the sort of what might be the languages of outer space or interstellar languages?
00:36:55.679 --> 00:36:56.559
Oh, I hope so.
00:36:56.719 --> 00:37:10.320
I think one of the things that movie did really well was uh point out how different and different in one way where first you're like, oh, they don't, you know, they don't speak the language, they draw the language.
00:37:10.559 --> 00:37:31.679
You know, and and there's uh if you read a lot of sci-fi, which it sounds like you do, um you know, there's in science fiction, there's aliens that draw, that use like sense of smell, they emit little pheromones, there's one, you know, that have like little pressure zones, you know, every everything you could possibly imagine to communicate could potentially form a language.
00:37:32.079 --> 00:37:39.519
So I think that that would be incredible to you know, it'd be the the linguistic research project of the century.
00:37:39.760 --> 00:37:45.360
Um and then also I uh back to the like the way they think.
00:37:45.599 --> 00:37:54.239
And this is something a ri I mean Arrival used the time, you know, the essentially like if you speak their language, you can sort of time jump.
00:37:54.880 --> 00:38:06.159
Um which may be a little extreme, but also I think was a really good, if nothing else, a great metaphor for just like how freaking different like this is gonna get.
00:38:06.400 --> 00:38:16.000
Like the aliens came, like it's not gonna be Star Trek where they come down and we're try essentially we're trying to like decode a version of French.
00:38:16.239 --> 00:38:21.199
It's gonna be like they're gonna come down and do something that we don't understand is even language.
00:38:21.920 --> 00:38:31.119
I mean we talk about language, but communication more broadly could be in any way possible and beyond their imagination, I think.
00:38:31.199 --> 00:38:38.000
Um and that's how it's difficult to reach out and try and try and discover it because you wouldn't know where to start.
00:38:38.480 --> 00:38:42.239
What kind of way would you you go about is they're not gonna speak like us.
00:38:42.559 --> 00:38:49.840
What haven't we discussed about your project, Jack, that we need to um uh so that more people can sort of follow it?
00:38:49.920 --> 00:38:54.239
I know you've got a crowdfunder, we did have a crowdfunder, I know you've got investors in it.
00:38:54.320 --> 00:38:55.599
Um, where's it going next?
00:38:55.679 --> 00:38:56.880
What's the timetable for it?
00:38:57.119 --> 00:38:58.320
Yeah, all successful.
00:38:58.559 --> 00:38:59.679
Congratulations, yeah.
00:38:59.920 --> 00:39:03.440
Yeah, so in a couple weeks we head to the Arctic to do the deposit.
00:39:03.599 --> 00:39:06.320
We're making a mini documentary about this.
00:39:06.639 --> 00:39:08.719
Just about endangered languages.
00:39:08.960 --> 00:39:13.599
I would encourage anyone to just investigate languages in your area.
00:39:13.840 --> 00:39:22.719
I would bet everybody out there that there's at least one or two tiny little niche languages within a couple hundred miles that they have no idea exists.
00:39:22.960 --> 00:39:26.239
And for our project, crowdfunding language data.
00:39:26.480 --> 00:39:45.039
If anybody out there speaks or has access to a small or niche language and would like to uh provide, I mean, like literally like cell phone selfie videos and stuff, you know, the easiest stuff in the world, but so that we have audio content.
00:39:45.199 --> 00:39:47.360
So we're always looking for stuff like that.
00:39:47.440 --> 00:39:52.559
And our partner, WikiTongues, and they're part of like the Wikimedia Foundation and all that.
00:39:52.800 --> 00:39:54.320
They host these things.
00:39:54.800 --> 00:39:55.599
So yeah.
00:39:55.760 --> 00:40:01.280
Uh niche language speakers reach out, language dorks uh hit me up on Twitter, love to talk.
00:40:01.440 --> 00:40:05.280
And I mean, honestly, thank you for letting me come on and talk about this.
00:40:05.440 --> 00:40:07.920
Uh like I said, it's a topic not many people even think about.
00:40:08.000 --> 00:40:10.639
So being able to kind of get the word out.
00:40:11.039 --> 00:40:13.920
No, it's so important because time is the issue with this.
00:40:14.079 --> 00:40:15.440
We will run out of time.
00:40:15.599 --> 00:40:16.559
And it's all very well.
00:40:16.639 --> 00:40:21.199
People are great with these um sort of endeavors going, well, we'll do it then, we'll do it then.
00:40:21.440 --> 00:40:22.880
You've got to do it now, haven't you?
00:40:23.039 --> 00:40:24.239
It's urgent and important.
00:40:24.559 --> 00:40:25.039
Absolutely.
00:40:25.119 --> 00:40:26.320
And that's what we're trying to do.
00:40:26.559 --> 00:40:28.239
Well, thank you so much, Tracy.
00:40:29.039 --> 00:40:30.000
Not at all, Jack.
00:40:30.079 --> 00:40:33.199
It's absolutely brilliant to speak to you and fascinating project.
00:40:33.280 --> 00:40:36.079
And we'll follow it along then on the future of you.
00:40:36.320 --> 00:40:46.079
Everyone will follow the progress, and maybe when you come back from the Arctic, you'll come on again and we can um you can update everybody how you got on, but uh, we will follow the projects with interest.
00:40:46.239 --> 00:40:48.320
Thank you so much for your time today.
00:40:53.840 --> 00:40:57.199
Thank you for listening to The Future of You hosted by me, Tracy Follows.
00:40:57.360 --> 00:41:01.280
Be sure to check out the show notes for more info about the topics we covered today.
00:41:01.440 --> 00:41:06.000
If you enjoyed this episode, please like and subscribe wherever you get your podcasts.
00:41:06.159 --> 00:41:09.840
And if you know someone who would love this episode, please share it with them.
00:41:10.000 --> 00:41:15.840
For more on the future of identity in a digital world, visit futuremade.group slash the future of you.
00:41:16.079 --> 00:41:20.960
And to explore the future of everything else, head over to future made.group.
00:41:21.039 --> 00:41:24.719
The Future of View podcast is produced by Big Tent Media.
