WEBVTT
00:00:00.162 --> 00:00:02.512
So Mike, what fun thing did you see on the internet this week?
00:00:02.895 --> 00:00:07.618
so this one, usually our fun things are, kind of, you know, random, silly, fun things.
00:00:07.618 --> 00:00:23.217
This one actually might be somewhat relevant to people who listen to this podcast, which is this new AI system called Jev from this company called Typesafe, which has just become all the rage among people playing with AI tools this week.
00:00:23.547 --> 00:00:25.987
It is, in some ways, really simple.
00:00:25.998 --> 00:00:27.457
It's not an LLM.
00:00:27.507 --> 00:00:28.998
It doesn't generate text.
00:00:28.998 --> 00:00:30.358
It doesn't talk to you.
00:00:30.588 --> 00:00:32.718
It doesn't do any of the things that we associate with AI.
00:00:32.737 --> 00:00:57.457
It is just sort of a incredibly fast, incredibly cheap universal classifier, which for trust and safety people, having a good classifier is everything, and this just suddenly, like, completely revolutionized the space of classifiers, and I've been just having fun with it and, and I've been seeing lots of people having fun with it.
00:00:57.468 --> 00:00:59.968
It's just a really unique, simple tool.
00:00:59.968 --> 00:01:13.847
You just give it stuff, and you say,"Classify it,""Put it into these buckets," or,"Give me a yes/no," or in some cases you can,"Give me a percentage," like how likely versus unlikely is this.
00:01:14.111 --> 00:01:22.078
and it just does it, incredibly fast and incredibly cheap, and it's just like a sort of new AI tool that lots of people are having fun with.
00:01:22.281 --> 00:01:24.162
Yeah, I can, I can vouch for this as well.
00:01:24.296 --> 00:01:34.418
I'm building a little system with a friend of mine, and it groups articles very accurately that are similar to one another in a way that, like, would've been very much more expensive and time-consuming prior.
00:01:34.459 --> 00:01:37.882
So I think this is a, a geeky if- but still
00:01:38.006 --> 00:01:47.269
a, a, a geeky thing, but I, I've been saying to people I now look around my life and see Jev-shaped problems where, whereas before I didn't.
00:01:47.310 --> 00:01:49.579
And so it's, yeah, it's, it is unique.
00:01:49.579 --> 00:01:50.000
What about you?
00:01:50.000 --> 00:01:51.170
What, what fun thing have you found?
00:01:51.495 --> 00:01:59.174
Well, I, I've come across an app, I'm sure there are other apps like it, called Persodex, which is a weird name.
00:01:59.527 --> 00:02:03.397
it's a kinda hybrid of person and Rolodex, as you might expect.
00:02:03.608 --> 00:02:18.007
but what it does, it essentially is a layer that sits on your Apple, contacts, I think it's Apple only, iOS only at the moment, and allows you to kind of add information, location data, notes to your contact.
00:02:18.206 --> 00:02:25.969
contact's obviously famously good for holding names and numbers, but doesn't-- they're not very good for anything else, and they get outdated, and you can't necessarily...
00:02:26.000 --> 00:02:29.379
They don't become useful until you know exactly who you want to, to contact.
00:02:29.729 --> 00:02:33.699
Whereas Persodex is, is apparently good for showing where people are.
00:02:33.740 --> 00:02:37.993
If you show up in a country, you can pull up Persodex and see if people are around.
00:02:38.003 --> 00:02:43.354
You can, know a bit about the conversation you had six or 12 months ago because you've kind of added it to the app.
00:02:43.593 --> 00:02:46.563
So, you know, I'm actually going to New York in a few weeks.
00:02:46.723 --> 00:02:54.120
I would love to be able to kinda use this to say, Who's in New York and how can I hang out with them while I'm there?" So I'm gonna give it a go.
00:02:54.283 --> 00:02:58.257
if any Ctrl-Alt-Speech listeners are based in New York and are around in October,
00:02:58.370 --> 00:02:59.651
get into Ben's persedex.
00:03:00.387 --> 00:03:01.276
Yeah, yeah.
00:03:01.727 --> 00:03:05.817
Sounds a bit creepy but, uh, yeah, I'm, I'm up for it.
00:03:05.973 --> 00:03:07.347
get in touch and let's talk.
00:03:14.985 --> 00:03:23.718
Hello and welcome to Ctrl-Alt-Speech, the podcast where we make sense of the major debates shaping online speech, platform power, content moderation, and the future of the internet.
00:03:23.972 --> 00:03:34.201
It's September the 24th, 2026, and this week we're making sense of the New York noise about AI regulation, highlighting three ways that internet users are embarking on governance.
00:03:34.551 --> 00:03:41.111
And for our extended Patreon episode, what recent stories about Discord and Alo say about the state of age assurance right now.
00:03:41.562 --> 00:03:51.032
Ctrl-Alt-Speech is the highly effective, age-appropriate musings of Mike Masnick, founder and editor of Techdirt, and me, Ben Whitelaw, founder and editor of Everything in Moderation.
00:03:51.307 --> 00:03:54.175
I think most of our, what we say is, 18 plus.
00:03:54.294 --> 00:03:55.314
You know, it's, it's, it's...
00:03:55.441 --> 00:03:56.770
checking IDs at the door, Ben?
00:03:58.390 --> 00:03:59.360
Do we need a bouncer?
00:03:59.371 --> 00:04:03.997
If we're gonna have to hire a bouncer to, check IDs before you listen to this podcast.
00:04:04.670 --> 00:04:05.661
we must see your ID.
00:04:05.871 --> 00:04:07.471
Uh, here's a question for the audience.
00:04:07.471 --> 00:04:10.980
Do we have any teenagers listening to this show?
00:04:11.197 --> 00:04:14.348
email us, podcast@ctrlaltspeech.com.
00:04:14.627 --> 00:04:15.858
I'm sort of curious.
00:04:15.867 --> 00:04:21.298
I'm thinking we might not get any responses to that, but I'm, I am sort of curious.
00:04:21.350 --> 00:04:30.958
obviously I think our audience skews more professional, but, uh, you know, th-this is a topic that, is important, and we do talk about kid safety stuff and all sorts of things.
00:04:30.958 --> 00:04:43.761
And one of the points that we do try and make is that kids' views on these things absolutely do matter, and people who are trying to make laws around this stuff should be talking more to kids to actually get their perspective and not what adults assume kids' perspective is, so
00:04:43.836 --> 00:04:44.187
Yeah.
00:04:44.437 --> 00:04:44.817
Agreed.
00:04:44.846 --> 00:04:54.166
And there was actually one listener who I chatted to at a conference last year who said that they made their, son or daughter listen for the first time, and they subsequently went on to keep listening.
00:04:54.166 --> 00:04:57.221
So, um, perhaps that, person will get in touch.
00:04:57.360 --> 00:05:03.560
Obviously, kids don't email, so feel free to Snap me or whatever the the latest thing is.
00:05:03.781 --> 00:05:17.930
Um, listeners, if you, if you get to, the end of today's free episode and you want 20 minutes more analysis from us, or you wish the episode dropped yesterday, which is a kind of funny thing to say when you're talking about the future, you can become a Patreon supporter.
00:05:18.194 --> 00:05:27.711
You can find out more about that Go to patreon.com/ctrlaltspeech to see our tiers, to get the great benefits of becoming a supporter, and to help us keep this podcast going each week.
00:05:28.007 --> 00:05:34.117
big love to the supporters who already do fork out for those extended episodes and early episodes.
00:05:34.326 --> 00:05:39.901
Mike, we've got a, couple of really chunky stories, I think, this week that incorporate a lot of different elements.
00:05:40.204 --> 00:05:42.423
and so we're gonna do our best to kinda tie them all together.
00:05:42.454 --> 00:05:49.990
We're gonna start with, I guess, what has been the big news story across a number of different, beats, I'd say, this week, which is the UN General Assembly.
00:05:50.254 --> 00:05:57.187
All the world's major leaders in one place, all of them in a room talking about various things, including AI regulation.
00:05:57.552 --> 00:06:04.060
And, you know, your fair leader, Donald Trump, has has bestowed us with some great wisdom.
00:06:04.321 --> 00:06:08.690
Um, that's just one of a number of elements, I guess, that, we wanted to kinda touch on in the opening segment today.
00:06:09.040 --> 00:06:09.345
Yeah.
00:06:09.355 --> 00:06:27.661
Well, I will note that not all of the world's leaders are there because, uh, China's Xi did not go, though is now in the US, or I don't know if he's left at this point, but flew into DC to meet with Trump there, but did not go to New York and did not present at the UN General Assembly.
00:06:27.749 --> 00:06:28.079
No.
00:06:28.209 --> 00:06:38.273
There's some very good photos that I've been seeing today of them hanging out together, and I don't know if you remember the Obama-Xi Jinping, Winnie the Pooh episode?
00:06:38.314 --> 00:06:52.564
Oh, I am very familiar with the Winnie the Pooh and the fact that this is, this is an interesting, content moderation thing, which is that you're not really allowed to mention Winnie the Pooh in China on the internet because of the comparisons between Xi and, and Winnie the Pooh.
00:06:52.766 --> 00:06:53.096
Yeah.
00:06:53.117 --> 00:07:03.572
I was actually talking to a, a Ctrl-Alt-Speech listener, about that content moderation issue, and then I saw pictures of Trump and, Xi Jinping together, and I thought, it's kind of happening again." Um,
00:07:03.904 --> 00:07:06.334
see, you see the Winnie the Pooh, uh, comparison?
00:07:06.812 --> 00:07:07.562
Yeah, I do.
00:07:07.612 --> 00:07:10.632
It, it's not quite Tigger as Donald Trump, but, uh,
00:07:10.915 --> 00:07:11.795
I was gonna say,
00:07:11.872 --> 00:07:12.910
yeah, the s- the
00:07:12.942 --> 00:07:13.572
it's not, not...
00:07:13.612 --> 00:07:14.161
Ior.
00:07:15.050 --> 00:07:15.889
Yeah, yeah, yeah.
00:07:16.389 --> 00:07:18.000
So yeah, I'm, I'm expecting more memes.
00:07:18.199 --> 00:07:26.387
but kind of talk us through what you parsed from all the noise, about AI regulation and all, all of the different media reports that have come out of New York this week.
00:07:26.673 --> 00:07:32.363
yeah, obviously, as we've been discussing the last few weeks, there's been a whole bunch of discussion about AI regulation.
00:07:32.403 --> 00:07:44.257
I mean, this has been going on for quite some time, but it sort of felt like it was building to some sort of crescendo with, with all of the concerns about AI safety and everything and, discussions about regulation and all this kind of stuff.
00:07:44.257 --> 00:07:49.367
And you have all these world leaders together in one place, and you know that it's going to be a topic of conversation.
00:07:49.603 --> 00:07:54.913
And you had Donald Trump going up there and renaming AI.
00:07:54.994 --> 00:08:00.733
Uh, suddenly it's super intelligence, or as he said,"SI." We should now refer to it as SI.
00:08:01.053 --> 00:08:02.684
So, first of all, that's not gonna happen.
00:08:03.228 --> 00:08:08.543
as much as people dislike the term AI, it is the term, it is the term that is going to stick.
00:08:08.553 --> 00:08:09.824
It is what everyone is used to.
00:08:09.824 --> 00:08:11.314
There's no way to, change that.
00:08:11.625 --> 00:08:19.024
and it is sort of weird because one of the other announcements that he made is that he's launching an AI force, not an SI force.
00:08:19.024 --> 00:08:22.394
If he's changing it to SI, he maybe should have thought about that.
00:08:22.624 --> 00:08:33.053
But he's launching an AI force, which he is saying is modeled on his super successful launch of the Space Force as a another branch of the military.
00:08:33.328 --> 00:08:34.868
what does that mean?
00:08:35.341 --> 00:08:36.711
I don't know.
00:08:36.760 --> 00:08:57.740
And I think that is the larger summary about almost everything that happened, not just from Trump, but from all the world leaders who made grand pronouncements and there was a, you know, what is it, a declaration, from a bunch of UN members that w- like, this grand declaration about and regulation.
00:08:57.994 --> 00:09:01.471
and the summary is that, what does it mean?
00:09:01.471 --> 00:09:02.610
Uh, probably nothing.
00:09:02.874 --> 00:09:07.793
it feels like the sort of thing that everybody felt that they had to say something about.
00:09:08.073 --> 00:09:11.553
They had to announce these big grand plans, and we're going to deal with it.
00:09:11.563 --> 00:09:16.774
We know that it is a challenge, and we know that it needs to be regulated, and therefore, we are going to do something.
00:09:18.058 --> 00:09:20.416
But what does it actually mean?
00:09:20.426 --> 00:09:21.905
What are they actually going to do?
00:09:22.076 --> 00:09:23.365
Nobody has any idea.
00:09:23.395 --> 00:09:26.405
I don't think anyone has any concrete plans.
00:09:26.405 --> 00:09:28.725
I don't think anyone is putting anything into action.
00:09:28.765 --> 00:09:35.196
Everyone is just talking and saying stuff because they feel that they need to comment on it, but nobody seems to have a good idea.
00:09:35.196 --> 00:09:37.278
Nobody seems to know what is actually going to work.
00:09:37.491 --> 00:09:44.022
And so it was a, a bunch of nothing, and to some extent, that feels like a really wasted opportunity, right?
00:09:44.022 --> 00:09:48.229
You had all these people together, and they're not experts in AI.
00:09:48.249 --> 00:09:51.298
They're not experts in how to regulate this stuff, really.
00:09:51.698 --> 00:09:58.371
But you did have this opportunity and, and, you know, Sam Altman and Dario Amodei spoke and, talked to people.
00:09:58.572 --> 00:10:00.631
You had all these chances for education.
00:10:00.631 --> 00:10:05.058
You had all these chances to have, more serious and more thoughtful discussions.
00:10:05.359 --> 00:10:12.028
But, you know, maybe they were happening in the back rooms, but all the stuff that happened publicly felt like noise.
00:10:12.139 --> 00:10:14.229
It didn't feel like there was any useful signal.
00:10:14.229 --> 00:10:17.028
It didn't feel like there was anything that really moved the ball forward.
00:10:17.269 --> 00:10:29.808
Didn't feel like there was anything that was legitimate in any way that, both the governments of the world will actually, bother to do anything real with or that the people developing the technology are going to take seriously.
00:10:30.089 --> 00:10:33.219
So the whole thing felt like pageantry, right?
00:10:33.219 --> 00:10:35.908
it was all for show, which is a very Donald Trump thing.
00:10:35.918 --> 00:10:37.578
Everything he does is sort of pageantry.
00:10:37.918 --> 00:10:52.225
But there was very little out of all of the different announcements and speeches and talks and roundtables that came out of this that felt substantive, that felt like we're going to see some actual change that we can recognize and tie back to this moment.
00:10:52.535 --> 00:10:56.841
And so to me, it was, it was a lot of, noise and very, very little signal.
00:10:57.169 --> 00:11:05.493
Yeah, and a continuation, I'd say, of the kind of regulatory conversations that are being had about the major platforms and about the social media space, I'd say.
00:11:05.493 --> 00:11:17.200
If you go into the, comments of various presidents and prime ministers that spoke, they essentially say the same kind of thing, but applied to frontier labs, big AI companies.
00:11:17.470 --> 00:11:26.460
You know, you take Macron, he was saying, that should be a kind of open source initiative that doesn't make Europe bound to or tied to the US or China.
00:11:26.580 --> 00:11:33.509
So again, a kind of very similar type of message to his tech sovereignty piece that he's been, banging on about for the last few years.
00:11:33.820 --> 00:11:48.076
Andy Burnham, the UK Prime Minister, again, not known for his kind of tech credentials, but was talking about being an honest broker in the global discussion about AI, which I think is what UK has often been, because of where it's kind of geographically placed and, because of its influence.
00:11:48.273 --> 00:11:50.253
and that kind of happened in a few other cases as well.
00:11:50.283 --> 00:11:53.980
So it's, it's like these leaders didn't necessarily do anything new.
00:11:53.980 --> 00:12:00.559
They just kind of continued the same type of, approach, which I think is, probably what you're seeing, right?
00:12:00.620 --> 00:12:02.842
There was no, There's a stalemate as, as a result.
00:12:03.231 --> 00:12:12.586
it's understandable to some extent because I think the reality is that nobody's quite sure what to do and what will actually be effective, but they feel that they need to do something.
00:12:12.639 --> 00:12:20.308
you know, this is the same political conundrum that we had around social media, where there was this general feeling that something needed to be done,
00:12:20.746 --> 00:12:21.067
Mm.
00:12:21.178 --> 00:12:25.769
people would say things or, like, toss out proposals, and occasionally you get somebody doing something.
00:12:26.068 --> 00:12:28.719
But, like, these are really, really complicated topics.
00:12:28.719 --> 00:12:44.451
This is a thing that we are often discussing here that require a ton of nuance and a ton of understanding, and misdiagnosing the problems and the challenges lead to very bad solutions that probably make the, challenges and the problems much worse in the long run.
00:12:44.871 --> 00:12:54.432
And so to some extent, I'm almost glad that this was a lot of nothing because I don't think we really understand what are the real...
00:12:54.442 --> 00:12:58.381
Like, there's easy to sort of cherry-pick little things and say,"This is the problem.
00:12:58.381 --> 00:13:11.568
This is the thing that we're concerned about." There's the sort of, I, I think very distracting discussion on the sort of like the big sexy humanity is gonna die and, very cinematic kinds of stories that I think are pretty silly, and not very helpful.
00:13:11.902 --> 00:13:17.131
But we don't, we don't have a clear definition of what is the challenge that we're trying to solve for.
00:13:17.142 --> 00:13:19.029
What is it that we're trying to fix?
00:13:19.068 --> 00:13:20.639
what harms are we trying to prevent?
00:13:20.938 --> 00:13:25.701
These are all really important things, and they're really nuanced, and there's not-- we don't have clear answers for that.
00:13:25.711 --> 00:13:40.371
And because of that, it's probably better that we're not rushing into the breach and saying,"This is how we're going to fix the things that we haven't even decided are the actual problems yet." but it did feel like here was an opportunity to have some more thoughtful discussions about what, are the real harms?
00:13:40.371 --> 00:13:41.721
What are the real challenges?
00:13:41.721 --> 00:13:43.091
What are the real opportunities?
00:13:43.451 --> 00:13:54.916
We could have that open discussion, but the world is not great right now and hasn't been for some time, I think, in having serious adult conversations about nuance and trade-offs in all of these things.
00:13:55.186 --> 00:14:05.662
And so it may be for the best in the long run that, people, politicians especially feel that they need to speak up, and the media has to build stories out of it because that's what both sides do.
00:14:05.892 --> 00:14:22.011
and you know, the, tech folks are gonna do the tech folk thing and, act humble before the, the, leaders of the world, while at the same time, notably, both OpenAI and Anthropic released new models this week, even right after they were saying,"Well, we're pacing, we're pacing," blah, blah, blah.
00:14:22.336 --> 00:14:35.011
you know, this feels like this should've been a momentous and important week, and the reality is that I don't think we will ever, point back to this week as anything, you know, anything that comes in the future came out of what happened this week.
00:14:35.466 --> 00:14:35.635
Yeah.
00:14:35.696 --> 00:14:39.860
I mean, and, let's be fair to the UN and the kind of people who gather together.
00:14:39.876 --> 00:14:49.466
it might be the case that the conversations were had and that that will facilitate or catalyze a bunch of, smarter, more thoughtful decisions about how AI is regulated.
00:14:49.696 --> 00:14:52.342
there's only so much that a bunch of people can do in a few days in New York.
00:14:52.363 --> 00:14:57.546
But from what was kind of shared publicly, a lot of it felt very similar.
00:14:57.836 --> 00:15:17.870
I think it will come down-- Again, this is, this is a kind of broader question about where intergovernmental, bodies like the UN actually can play a role, and to what extent is it going to be states who are gonna eventually determine how this plays out through regulation that they impose on companies that act in their, jurisdictions, right?
00:15:17.870 --> 00:15:27.980
And I was reading an interview by, with rather, the UN under-secretary general, who's kind of coordinating a lot of the discussions on the sidelines about some of these regulatory letters.
00:15:27.980 --> 00:15:37.216
And he, he said, The emphasis lays with member states and their appetite for a more robust capacity for international oversight." It's kind of, we can only do so much as the UN, frankly.
00:15:37.363 --> 00:15:47.706
if states don't want to do it because they're pursuing a, growth agenda or they feel like AI is important for, economic development, then there's, you know, there's not much we can do, frankly.
00:15:47.956 --> 00:15:50.225
And, and I think that showed in the letter that you mentioned.
00:15:50.500 --> 00:15:55.409
22 signatories, a number of countries that you would expect to be on there weren't on there.
00:15:55.788 --> 00:16:05.364
So can kind of read between the lines and, and maybe guess who is taking that kind of growth innovation approach and, at least taking a back seat while other people figure it out.
00:16:05.533 --> 00:16:16.878
Yeah, and, the reality is, like, the more important discussion was probably the one that was happening a few hundred miles to the south in the White House between the Chinese and, and American, you know.
00:16:17.111 --> 00:16:27.217
I don't think Trump and Xi necessarily would have a, particularly useful or fruitful discussion on AI regulation, but I imagine some of the people around them probably did.
00:16:27.548 --> 00:16:35.280
But right now, the reality is that those are the only two countries that matter because all of the leading models are coming out of those two countries and there is...
00:16:35.441 --> 00:16:49.988
That fight is continuing, and it, feels very sort of cliché and useless to say, like,"Oh, it's this, you know, the battle between the US and Chinese models." but that is where the fight is happening, and that is where the discussion is happening.
00:16:49.998 --> 00:17:08.644
And if those two sides continue to battle it out, and the fear is that, if these tools continue to become more and more useful, as I think they are, and I know some people disagree with that, but I think it's pretty clear at this point that the tools have gotten more and more useful, there is a question of who's going to be providing the underlying, systems.
00:17:08.817 --> 00:17:12.988
And the Chinese models have gotten really, really, really good.
00:17:13.238 --> 00:17:29.867
they are behind the American frontier models, and there is an argument that the Chinese models are really built off of the backs of the American models, so that if the American model makers slow down, that, naturally the Chinese models will slow down as well because they're really relying on them in the first place.
00:17:30.238 --> 00:17:42.574
But, every other country, to some extent, doesn't matter as much if those two countries just keep their, feet on the gas in terms of, driving those companies to, build the next, more powerful model.
00:17:42.925 --> 00:17:43.165
Yeah.
00:17:43.165 --> 00:17:44.826
Can we talk about the US for a second?
00:17:45.364 --> 00:17:46.013
Uh, must we?
00:17:47.066 --> 00:17:50.655
because we, y- y- talked about, you made that kind of driving metaphor there.
00:17:50.665 --> 00:18:02.326
I was gonna say I'm getting kind of car sick by the l- you know, the very jerky left and rights that the US administration is taking in terms of where it's going with AI regulation.
00:18:02.536 --> 00:18:04.576
And there's a couple of more stories this week that showed that.
00:18:04.935 --> 00:18:08.885
You know, Trump announcing at the weekend that he was gonna appoint a new AI czar
00:18:09.423 --> 00:18:11.223
SIzar, AIzar, SI...
00:18:11.746 --> 00:18:11.826
czar,
00:18:11.963 --> 00:18:12.844
Yeah, yeah
00:18:13.135 --> 00:18:16.333
uh, without any other kind of congressman really knowing about it.
00:18:17.066 --> 00:18:29.385
then there was Treasury Secretary Scott Bessent saying on Monday that the government isn't gonna take liability away from the AI labs, which feels like a fairly important addition to the conversation, but wasn't really picked up.
00:18:29.603 --> 00:18:34.816
and raises a question about what an AI or AI czar might do if they are appointed.
00:18:35.031 --> 00:18:35.342
Yep.
00:18:35.343 --> 00:18:41.019
and now there's, there are reports that Bessent may be the leading, candidate to become the AI czar.
00:18:41.359 --> 00:18:43.336
So his, comments actually do matter.
00:18:43.375 --> 00:18:49.746
But apparently, there's a sort of knife fight going on within the administration over who gets to be the AI czar.
00:18:50.022 --> 00:18:53.230
and so there are a number of names that have, been proposed.
00:18:53.230 --> 00:18:56.306
There was a, an article that listed out a, bunch of these names.
00:18:56.675 --> 00:19:13.867
and, a bunch of them are sort of industry-connected people, a lot of people associated with, the venture capital firm a16z, uh, who are now in the administration who are potential AI czar, but there may also be Scott Bessent, who I don't think of as a particularly, knowledgeable person on AI.
00:19:13.907 --> 00:19:15.657
doesn't seem to have that much expertise.
00:19:15.798 --> 00:19:19.884
Well, doesn't seem to have that much expertise on lots of things that he's now opining on.
00:19:20.192 --> 00:19:25.744
But we've had AI czars before, and we had David Sacks for quite some time.
00:19:25.904 --> 00:19:41.407
you know, he's no longer the AI czar, but he's apparently still in the good graces of Trump and advising him on AI policy, and he's been very vocal about all these discussions and sort of mocking both Anthropic and OpenAI over the last few weeks and, and all the existential risk stuff.
00:19:41.435 --> 00:19:48.464
he's sort of been completely mocking them and saying that they're really just going for regulatory capture, which he might not be wrong about.
00:19:48.484 --> 00:19:51.565
I mean, there's lo- very little that I agree with David Sacks on.
00:19:51.565 --> 00:19:53.531
I, I think he's not great.
00:19:53.589 --> 00:19:56.055
but he might be right on that particular point.
00:19:56.284 --> 00:20:00.701
there've been a couple other people who have been very temporarily considered the AI czar.
00:20:01.011 --> 00:20:04.923
I think one of them lasted four days before before he was let go.
00:20:05.112 --> 00:20:11.892
so it'll be interesting to see who actually becomes the AI czar, but also if they have any real power or what it is that they can actually do.
00:20:12.082 --> 00:20:14.857
and again, like US policy is all over the map.
00:20:14.907 --> 00:20:34.765
You have some people in or around the administration who hate big tech and hate the AI companies and wanna punish them and limit them and all this stuff, and then you have other people who are like,"Let everything go." And you had Trump say that didn't need any guardrails other than your very smart and high IQ president, which, you know, is really concerning.
00:20:34.875 --> 00:20:38.984
I mean, he did talk about that, the AI force would be filled with high IQ people.
00:20:38.984 --> 00:20:44.491
He's a little obsessed with IQ, which is kind of embarrassing in all sorts of other ways.
00:20:44.531 --> 00:20:44.701
But
00:20:44.738 --> 00:20:52.894
he also said that the kind of dangers posed by AI, this was in his speech earlier this week, were similar to the climate change hoax.
00:20:53.074 --> 00:20:55.211
Um, hi- his words, not mine.
00:20:55.364 --> 00:20:59.307
important distinction there, um, for people who are listening.
00:20:59.351 --> 00:21:06.371
and so y- that mindset, that view is going to trickle down, to an AI czar or to any kind of approach to AI regulation, I think so.
00:21:06.763 --> 00:21:09.661
it's a hell of a ride kind of watching what the US is doing, I'd say
00:21:09.689 --> 00:21:21.939
Yeah, I mean, the simple fact is that we don't have adult leadership in the US government right now, and that is a problem when you want to have serious discussions about how do you deal with this technology, whether it is through regulatory means or, or other means.
00:21:22.266 --> 00:21:26.368
it would be nice to have, like, serious people in charge, and we just don't have that.
00:21:26.538 --> 00:21:28.692
So that's the reality that we have to deal with.
00:21:28.976 --> 00:21:29.336
Yeah.
00:21:30.420 --> 00:21:31.849
I don't, I don't have anything to say about that.
00:21:33.000 --> 00:21:37.519
Can we find some adults to preside over the US administration, please?
00:21:38.125 --> 00:21:41.016
perfect segue, I'd say, Mike, to our eye-catching section.
00:21:41.398 --> 00:21:55.823
we're, we're bringing a bit of hope this week because there are a few stories that I think are kind of evidence of internet users embarking on governance and, partaking in acts of internet governance in ways that I think are really, really hopeful and we wanted to flag those to
00:21:56.345 --> 00:21:56.835
Hope?
00:21:57.075 --> 00:21:58.125
On this podcast?
00:21:58.174 --> 00:22:00.065
Ben, what is happening?
00:22:00.454 --> 00:22:01.644
Is this a cry for help?
00:22:03.093 --> 00:22:03.653
Don't ask
00:22:03.654 --> 00:22:05.265
We don't do hope here.
00:22:06.232 --> 00:22:06.442
Don't
00:22:06.464 --> 00:22:11.005
We do trade-offs and sadness and misery all around.
00:22:11.007 --> 00:22:11.396
Yeah.
00:22:11.396 --> 00:22:16.196
I g- I think I got six hours sleep last night for the first time in a few months, and it's had an effect on me.
00:22:16.527 --> 00:22:19.940
so, let's go through those and, and, you know, you can be the judge.
00:22:20.116 --> 00:22:26.166
I mean, the, the first thing to kind of flag is this, great new piece from New Public looking at a service called MySky.
00:22:26.407 --> 00:22:29.519
you know, MySky has, has actually been around, A while, let's say.
00:22:29.720 --> 00:22:35.374
I haven't actually been in partaking in it, and, and so I only found out about it as a result of this New Public piece.
00:22:35.374 --> 00:22:41.920
And it's a great interview with Rene Di Resta and Jonathan Stray, who came up with the idea and have kind of rolled it out.
00:22:42.194 --> 00:22:51.997
we talk about algorithms all the time on the podcast, Mike, platforms offer algorithms, but it's incredibly hard to see how they work, why you're being recommended certain content, and you have no control over it.
00:22:52.126 --> 00:22:57.644
You know, you can like certain posts or hover over certain, videos maybe to help tell the algorithm what you're into.
00:22:58.001 --> 00:22:58.480
that's it.
00:22:58.500 --> 00:22:59.599
You have very little control.
00:22:59.920 --> 00:23:11.230
And what MySky does is that it uses the AT protocol and is, accessible via BlueSky to give users some, customizable dials to shape that content that you see.
00:23:11.390 --> 00:23:14.970
And so, so I, because I found out about it this week, I went on and I, did that.
00:23:15.199 --> 00:23:20.529
and you can do that by looking at how much time you want the content to be served over.
00:23:20.529 --> 00:23:23.619
So if you want just to look at content over the last 24 hours, you can do that.
00:23:23.890 --> 00:23:25.339
You can do it over a week if you want.
00:23:25.696 --> 00:23:31.034
it can be kinda tweaked according to whether you want to see more content from people you follow versus people who aren't.
00:23:31.364 --> 00:23:35.344
You can do it based upon whether it's, you know, content is popular on the platform and trending.
00:23:35.614 --> 00:23:37.700
And so it's a really, it's a great way.
00:23:37.700 --> 00:23:41.180
I didn't expect to enjoy it as much as I, as, as much as I did.
00:23:41.539 --> 00:23:42.829
And now it won't be for everyone.
00:23:43.017 --> 00:23:45.336
it's not gonna be something that everyone will go on and do.
00:23:45.727 --> 00:23:50.568
But I think it's a great example that we should hold up as, as something that's positive and, yeah, hopeful.
00:23:50.798 --> 00:23:51.104
Yeah.
00:23:51.115 --> 00:24:07.663
it's really wonderful and, actually if you go back to my original"Protocols, Not Platforms" paper, which in part inspired the At Protocol in the first place, this was part of the discussion that I had, was that, it would be great if people could, you know, manually tweak their own dials, right?
00:24:07.683 --> 00:24:31.758
I mean, I think so much of the concern, and including much of the, the sort of general hatred of algorithms, where a lot of people are like,"Oh, algorithms bad," which, I've always had an, an issue with, is because of the lack of transparency, that they don't understand who is making the decisions, who is turning the knobs, who is deciding what it is that you wanna see, and therefore there's the opportunity for there to be manipulation.
00:24:31.820 --> 00:24:42.478
a lot of people have the natural concern and they think that, oh, it's like, you know, I'm being manipulated to make me angry because, you know, a lot of people are fed a lot of content that makes them feel angry, and so they, they fear that.
00:24:42.478 --> 00:24:46.077
Or, or it might be, you know, to try and sell me stuff or whatever it is.
00:24:46.077 --> 00:25:03.875
And there's just a lot of concern from that lack of transparency, and this brings in You know, it gives you so much transparency and control over this and, and so you can understand why certain things were recommended, and then you can adjust that if you wanna see, different types of things or different reasons for having things recommended to you.
00:25:04.075 --> 00:25:07.085
Having that control is really, really powerful.
00:25:07.198 --> 00:25:18.557
you know, it's probably too much and overwhelming for some people, but just having it there and having the ability to, play around and turn the knobs yourself, and even to learn like what that means.
00:25:18.798 --> 00:25:28.131
Because again, like, you know, when you have this assumption that all the, algorithmic decisions are done for evil reasons, sometimes you don't even realize like how things turn out.
00:25:28.151 --> 00:25:44.904
Even if you have control of the knobs, you might begin to realize like,"Oh, it's interesting to see what things get recommended when I, tweak the knobs this way or that." And so this, as a demonstration of that and clearly showing it and doing it in a sort of nice UI fashion, I think is really, really powerful and educational at the same time.
00:25:44.924 --> 00:25:47.055
And so I'm, I'm a huge fan of this project.
00:25:47.055 --> 00:25:48.157
I think it's, fantastic.
00:25:48.415 --> 00:25:49.165
Yeah, agreed.
00:25:49.306 --> 00:25:53.296
I'm gonna update listeners when I have a further tweak, in the coming weeks.
00:25:53.296 --> 00:25:57.893
But great start, great example of, people building, in interesting ways.
00:25:58.202 --> 00:26:02.252
and we're seeing, you know, we're seeing the influence of this spread elsewhere.
00:26:02.252 --> 00:26:06.945
And so YouTube just announced giving users more control over their algorithms.
00:26:06.955 --> 00:26:17.326
It's not nearly to this level at all, and it's not to the level of, like, even a Blue Sky where, you can bring in your own algorithm and, and do things that way.
00:26:17.546 --> 00:26:23.425
But, I think we're beginning to see this recognition and, you know, we've seen others do similar things.
00:26:23.425 --> 00:26:31.756
Threads, people would joke, on Threads about dear algo, like, you know, trying to tell the algo what to do, and then Threads actually implemented that.
00:26:31.836 --> 00:26:40.875
So allowing users to have some influence, direct influence over the algorithm is now becoming more of a standard thing, and I think that's a good thing.
00:26:41.096 --> 00:26:45.445
And the My Sky feed is an extreme version of that, but a really cool one.
00:26:45.933 --> 00:26:46.683
Yeah, agreed.
00:26:47.080 --> 00:27:10.911
talking of YouTube, Mike, we shouldn't gloss over the fact that at the event, the, the Made On YouTube event that, a lot of this was announced this week, Neal Mohan, who we've covered on the podcast before, he's the CEO of YouTube, he said that YouTube wouldn't stump up the£6 billion as part of the big Meta settlement, the multi-state settlement that we covered on the podcast a few weeks ago.
00:27:11.010 --> 00:27:14.631
and he claimed that YouTube has been focusing on child safety for a long time.
00:27:14.631 --> 00:27:23.490
It would continue to do so, and basically that means that the Meta will now pay 70% of that overall 18 billion uh,'cause that was the kind of deal, right?
00:27:23.528 --> 00:27:23.977
we can get
00:27:23.980 --> 00:27:39.141
So yeah, a reminder, some of the payment that Meta would make to the states was conditional on getting TikTok and YouTube to agree to the same restrictions, which is, to some extent an anti-competitive, you know, way to, build an industry standard.
00:27:39.346 --> 00:27:52.234
so this would save Meta some money, obviously, save YouTube some money if they're not paying into the system as well, but also means that Meta might not then be able to define the specific features and standards.
00:27:52.654 --> 00:28:07.627
And, you know, one of the points that I had raised regarding that whole discussion was, this was Meta deciding for the rest of the industry what they wanted to do, and my fear was that meant that we wouldn't get more interesting experimentation from others or, different attempts to do different things.
00:28:08.028 --> 00:28:26.214
Whereas, what Neil Mohan is saying here is basically like:"Hey, we have other ideas on what actually helps keep kids safe, and we don't wanna have our hands tied by some other company and a bunch of state AGs that tell us what we have to do when we actually have better ideas." And so that is encouraging.
00:28:26.345 --> 00:28:37.141
I still think what we will likely see is states and state AGs continuing to go after YouTube and pressuring them to accept these same sorts of feature changes.
00:28:37.451 --> 00:28:50.381
But, I think it's a good thing Because, you know, so far, TikTok and YouTube had been pretty quiet in re- in response to, to Meta's, all-out war to try and get them to sign on to the same deal.
00:28:50.580 --> 00:28:54.411
And so it's good for him to just take a stand and say like:"No.
00:28:54.451 --> 00:28:55.417
Look, that's stupid.
00:28:55.417 --> 00:28:56.218
We're not gonna do that."
00:28:56.579 --> 00:28:56.849
Yeah.
00:28:56.880 --> 00:28:59.069
And the question is what, what will Meta do next?
00:28:59.230 --> 00:29:03.813
Because I expect they will have a plan for f- for this.
00:29:04.173 --> 00:29:20.083
You know, I, I said on the podcast that, I feel like they were playing checkers and everyone else was playing chess when they concocted this idea, and I wouldn't be surprised if they have a next move, because it was fairly predictable that YouTube would turn around and say,"We're not paying." so I don't know what it will be, but I think they'll have something up their sleeve.
00:29:20.166 --> 00:29:20.507
Yeah.
00:29:20.527 --> 00:29:24.683
I mean, it's, it's not even the payment I think that matters so much as the feature changes, right?
00:29:24.693 --> 00:29:26.824
That, that's the, the bigger issue to me.
00:29:27.095 --> 00:29:35.334
and again, like the features, and I don't necessarily think that the features that Meta agreed to put in place even make sense on Instagram and, Facebook.
00:29:35.630 --> 00:29:40.124
but, YouTube is a very different platform, and I, I like, yeah, I keep trying to make this point too.
00:29:40.124 --> 00:29:44.933
Like, each of these platforms, uh, is used differently and in different ways and with different audiences.
00:29:45.324 --> 00:29:50.540
And YouTube especially is much more of a, a sit back and, watch kind of platform.
00:29:50.540 --> 00:30:01.547
It's much more akin to television than traditional social media, and yet some of these feature requirements that Meta would be placing on them are really specific to specifically social media.
00:30:01.856 --> 00:30:04.207
And so the whole thing seems very strange.
00:30:04.227 --> 00:30:10.020
It would be nice for YouTube to, take a really strong stand here and say like, No, that's, that's stupid for us.
00:30:10.020 --> 00:30:10.790
We're not gonna do that."
00:30:11.048 --> 00:30:11.368
Yeah.
00:30:11.749 --> 00:30:15.249
we're, we're doing what we never do, Mike, which is kind of rapidly going through our eye-catching stories.
00:30:15.269 --> 00:30:16.439
So we can, fit...
00:30:16.525 --> 00:30:19.894
one more eye-catching story in before we go to our extended Patreon episode.
00:30:20.214 --> 00:30:26.858
And, again, fun example of, users on Reddit this time enacting a bit of power to define what, their online space looks like
00:30:27.282 --> 00:30:28.818
Yeah, I really love this story.
00:30:28.929 --> 00:30:34.659
I think this is, this is a really interesting, example of sort of community governance, for interesting challenges.
00:30:34.659 --> 00:30:44.721
We're sort of used to so many of the trust and safety stories that we talk about, or content moderation stories that we talk about, are sort of big companies making decisions and, going back to what I was saying before with the algorithms.
00:30:44.991 --> 00:30:58.781
Like, there is some feeling of uncomfortableness with the idea that you have big, powerful companies with different interests than you making big decisions about what kind of speech or what kinds of things you can do online, and that, it has sort of a weird feeling.
00:30:59.067 --> 00:31:06.501
But, we know that, communities have, setups that, have rules and, and moderation themselves, and seeing that in action is really interesting.
00:31:06.501 --> 00:31:12.442
So this is a story about a specific subreddit, the Crafts subreddit, which is a very big subreddit,
00:31:12.644 --> 00:31:13.864
I'm not a member, but, but
00:31:13.951 --> 00:31:15.531
I am not a member either.
00:31:15.571 --> 00:31:21.989
Uh, but, you know, it talks about crafting and making stuff and knitting and, and all sorts of things along those lines.
00:31:22.308 --> 00:31:31.134
And there was a debate about whether or not people should be allowed to post Harry crafts and discussions about them.
00:31:31.535 --> 00:31:38.795
And what was happening is that, as I'm sure many listeners on the podcast know, J.K.
00:31:38.795 --> 00:31:43.525
Rowling, the author of the Harry Potter books, has become somewhat controversial and
00:31:43.722 --> 00:31:45.093
Don't start unpacking that
00:31:45.472 --> 00:31:48.817
I will not unpack that other than, you know, yes.
00:31:49.106 --> 00:32:02.363
She's become very controversial because she's said some things that have upset lots of people and, and I would say, the one thing I will say on this is I think has done a lot of damage, that was, really unnecessary and, deeply harmful.
00:32:02.553 --> 00:32:12.272
and so that has led to whenever people are posting"Harry Potter" related crafts, the discussion related to them has gone off the rails within this particular subreddit.
00:32:12.613 --> 00:32:23.242
And that has become a moderation challenge for the moderators within that subreddit, and often those discussions have nothing to do with the actual crafts and very much everything to do with J.K.
00:32:23.242 --> 00:32:26.303
Rowling and the mess that she has made.
00:32:26.740 --> 00:32:35.047
and so the entire subreddit basically had a discussion of whether or not we should just stop allowing any discussions on"Harry Potter" related crafts.
00:32:35.497 --> 00:32:47.817
And they had an incredibly open and pretty thorough process in which they asked everybody to weigh in, and there was a big discussion and then voting, and it wasn't a entirely one-sided thing.
00:32:47.826 --> 00:32:50.009
It w- there were, viewpoints across the board.
00:32:50.240 --> 00:33:01.336
And then on the voting, it sort of narrowly came out with a narrow majority saying that they thought that we should just ban discussions of, Harry Potter" crafts.
00:33:01.356 --> 00:33:05.920
And so the moderator team of the subreddit says that that is the new rule.
00:33:06.136 --> 00:33:10.960
you cannot post anything related to"Harry Potter." You can post things that are...
00:33:11.109 --> 00:33:19.522
If it's not directly, if it just sort of looks like a"Harry Potter" thing but doesn't directly reference"Harry Potter" or that world, you can still do that.
00:33:19.769 --> 00:33:35.136
but what I thought was, was really interesting was the openness of this discussion and the thoughtfulness, and people talked about like slippery slope,"Oh, now we're getting political, now we're gonna ban this." And, what they said very, very clearly is like,"No." this is not even a political statement about J.K.
00:33:35.136 --> 00:33:37.146
Rowling or her views on things.
00:33:37.366 --> 00:33:48.653
It is very much that it becomes impossible to have a sensible community when any time any"Harry Potter" content shows up here, the conversation just goes completely off the rails.
00:33:48.932 --> 00:33:53.997
And so it's almost a, you know, a, form of preservation to say,"We're just not gonna deal with that.
00:33:54.096 --> 00:33:56.436
That kind of content is just not allowed."
00:33:56.615 --> 00:33:56.944
Yeah.
00:33:57.095 --> 00:34:04.980
I mean, like massive credit to these moderators, you know, which we forget and we gloss over because Reddit is, essentially kind of helmed by these subreddit moderators.
00:34:04.980 --> 00:34:21.385
But these are volunteers who are, moderating gigantic communities, at some points with often hundreds of thousands or millions of users, and they're doing this very kind of deliberative, inclusive approach to how standards and, and rules are set, and I think that's fantastic.
00:34:21.675 --> 00:34:28.385
We've seen examples of these kind of community council, or jury approaches working really well in some, in some areas.
00:34:28.385 --> 00:34:39.045
And I think, again, like it's not very sexy'cause it's kind of slow and, deliberate, and it involves polls and processes and being transparent and documenting thinking.
00:34:39.262 --> 00:34:43.541
but again, that's, what I think the grunt work of, looking after online spaces is about.
00:34:43.898 --> 00:34:44.208
Yeah.
00:34:44.208 --> 00:35:21.538
I mean, there's a lot of work and a lot of research and, plenty of academic research on sort of like community governance that often gets ignored in these discussions and-- but I think is really, really important and sort of understanding that and understanding that, community governance really, really matters, and it's not all just like one super rich, super powerful dude deciding how the community works, and recognizing these other kinds of models and ones that are more engaging and involve people and actually, try to come to some sort of agreement recognizing pros and cons to different approaches and just saying like,"This is the reality, this is what we're trying to deal with.
00:35:21.557 --> 00:35:44.150
how should we best deal with it?" And coming to a conclusion saying,"These are the new rules based on this open discussion," is, whether or not you agree with any of the, the position, it's somewhat refreshing and somewhat encouraging that, like, a community can make an adult decision that is like, let's talk about the trade-offs, let's talk about the different issues here, and then let's just make a call on what's gonna be best for our community and go with it." that's, really great.
00:35:44.413 --> 00:35:45.152
Yeah, agreed.
00:35:45.391 --> 00:35:49.932
we're gonna touch on a bit of that in the extended Patreon episode, Mike, which we'll go into now.
00:35:50.208 --> 00:35:51.016
we'll wrap up today.
00:35:51.016 --> 00:35:57.106
I, I hope listeners don't feel too much hope and positivity because that's not really our thing.
00:35:57.295 --> 00:35:59.088
We, we could, we could change the podcast.
00:36:00.246 --> 00:36:01.905
we can, We can be hopeful for the future
00:36:01.952 --> 00:36:02.262
Yeah.
00:36:02.302 --> 00:36:02.862
Yeah, okay.
00:36:02.893 --> 00:36:04.092
Well, don't promise anything too much.
00:36:04.293 --> 00:36:05.972
Um, listeners, thanks for joining us.
00:36:06.030 --> 00:36:09.327
if you enjoy the podcast, rate and review it wherever you get your podcasts.
00:36:09.606 --> 00:36:14.067
become a Patreon supporter to get this next segment, and we'll jump into that now.
00:36:14.117 --> 00:36:26.936
It's all about Discord's age assurance rollout and how Ofcom is going after Aylo, the owner of the largest porn site in the world, Pornhub, and what that means for the future of the internet
00:36:29.077 --> 00:36:32.358
That's the end of this week's free edition of Ctrl Alt Speech.
00:36:32.739 --> 00:36:39.617
To hear the extended episode with additional coverage, become a Patreon supporter at patreon.com/ctrlaltspeech.
00:36:39.878 --> 00:36:42.418
That's C-T-R-L Alt Speech.
00:36:42.998 --> 00:36:49.418
If your organization is interested in sponsoring the podcast, contact us via our website at ctrlaltspeech.com.
00:36:49.858 --> 00:36:50.577
Thanks for listening
00:00:00.162 --> 00:00:02.512
So Mike, what fun thing did you see on the internet this week?
00:00:02.895 --> 00:00:07.618
so this one, usually our fun things are, kind of, you know, random, silly, fun things.
00:00:07.618 --> 00:00:23.217
This one actually might be somewhat relevant to people who listen to this podcast, which is this new AI system called Jev from this company called Typesafe, which has just become all the rage among people playing with AI tools this week.
00:00:23.547 --> 00:00:25.987
It is, in some ways, really simple.
00:00:25.998 --> 00:00:27.457
It's not an LLM.
00:00:27.507 --> 00:00:28.998
It doesn't generate text.
00:00:28.998 --> 00:00:30.358
It doesn't talk to you.
00:00:30.588 --> 00:00:32.718
It doesn't do any of the things that we associate with AI.
00:00:32.737 --> 00:00:57.457
It is just sort of a incredibly fast, incredibly cheap universal classifier, which for trust and safety people, having a good classifier is everything, and this just suddenly, like, completely revolutionized the space of classifiers, and I've been just having fun with it and, and I've been seeing lots of people having fun with it.
00:00:57.468 --> 00:00:59.968
It's just a really unique, simple tool.
00:00:59.968 --> 00:01:13.847
You just give it stuff, and you say,"Classify it,""Put it into these buckets," or,"Give me a yes/no," or in some cases you can,"Give me a percentage," like how likely versus unlikely is this.
00:01:14.111 --> 00:01:22.078
and it just does it, incredibly fast and incredibly cheap, and it's just like a sort of new AI tool that lots of people are having fun with.
00:01:22.281 --> 00:01:24.162
Yeah, I can, I can vouch for this as well.
00:01:24.296 --> 00:01:34.418
I'm building a little system with a friend of mine, and it groups articles very accurately that are similar to one another in a way that, like, would've been very much more expensive and time-consuming prior.
00:01:34.459 --> 00:01:37.882
So I think this is a, a geeky if- but still
00:01:38.006 --> 00:01:47.269
a, a, a geeky thing, but I, I've been saying to people I now look around my life and see Jev-shaped problems where, whereas before I didn't.
00:01:47.310 --> 00:01:49.579
And so it's, yeah, it's, it is unique.
00:01:49.579 --> 00:01:50.000
What about you?
00:01:50.000 --> 00:01:51.170
What, what fun thing have you found?
00:01:51.495 --> 00:01:59.174
Well, I, I've come across an app, I'm sure there are other apps like it, called Persodex, which is a weird name.
00:01:59.527 --> 00:02:03.397
it's a kinda hybrid of person and Rolodex, as you might expect.
00:02:03.608 --> 00:02:18.007
but what it does, it essentially is a layer that sits on your Apple, contacts, I think it's Apple only, iOS only at the moment, and allows you to kind of add information, location data, notes to your contact.
00:02:18.206 --> 00:02:25.969
contact's obviously famously good for holding names and numbers, but doesn't-- they're not very good for anything else, and they get outdated, and you can't necessarily...
00:02:26.000 --> 00:02:29.379
They don't become useful until you know exactly who you want to, to contact.
00:02:29.729 --> 00:02:33.699
Whereas Persodex is, is apparently good for showing where people are.
00:02:33.740 --> 00:02:37.993
If you show up in a country, you can pull up Persodex and see if people are around.
00:02:38.003 --> 00:02:43.354
You can, know a bit about the conversation you had six or 12 months ago because you've kind of added it to the app.
00:02:43.593 --> 00:02:46.563
So, you know, I'm actually going to New York in a few weeks.
00:02:46.723 --> 00:02:54.120
I would love to be able to kinda use this to say, Who's in New York and how can I hang out with them while I'm there?" So I'm gonna give it a go.
00:02:54.283 --> 00:02:58.257
if any Ctrl-Alt-Speech listeners are based in New York and are around in October,
00:02:58.370 --> 00:02:59.651
get into Ben's persedex.
00:03:00.387 --> 00:03:01.276
Yeah, yeah.
00:03:01.727 --> 00:03:05.817
Sounds a bit creepy but, uh, yeah, I'm, I'm up for it.
00:03:05.973 --> 00:03:07.347
get in touch and let's talk.
00:03:14.985 --> 00:03:23.718
Hello and welcome to Ctrl-Alt-Speech, the podcast where we make sense of the major debates shaping online speech, platform power, content moderation, and the future of the internet.
00:03:23.972 --> 00:03:34.201
It's September the 24th, 2026, and this week we're making sense of the New York noise about AI regulation, highlighting three ways that internet users are embarking on governance.
00:03:34.551 --> 00:03:41.111
And for our extended Patreon episode, what recent stories about Discord and Alo say about the state of age assurance right now.
00:03:41.562 --> 00:03:51.032
Ctrl-Alt-Speech is the highly effective, age-appropriate musings of Mike Masnick, founder and editor of Techdirt, and me, Ben Whitelaw, founder and editor of Everything in Moderation.
00:03:51.307 --> 00:03:54.175
I think most of our, what we say is, 18 plus.
00:03:54.294 --> 00:03:55.314
You know, it's, it's, it's...
00:03:55.441 --> 00:03:56.770
checking IDs at the door, Ben?
00:03:58.390 --> 00:03:59.360
Do we need a bouncer?
00:03:59.371 --> 00:04:03.997
If we're gonna have to hire a bouncer to, check IDs before you listen to this podcast.
00:04:04.670 --> 00:04:05.661
we must see your ID.
00:04:05.871 --> 00:04:07.471
Uh, here's a question for the audience.
00:04:07.471 --> 00:04:10.980
Do we have any teenagers listening to this show?
00:04:11.197 --> 00:04:14.348
email us, podcast@ctrlaltspeech.com.
00:04:14.627 --> 00:04:15.858
I'm sort of curious.
00:04:15.867 --> 00:04:21.298
I'm thinking we might not get any responses to that, but I'm, I am sort of curious.
00:04:21.350 --> 00:04:30.958
obviously I think our audience skews more professional, but, uh, you know, th-this is a topic that, is important, and we do talk about kid safety stuff and all sorts of things.
00:04:30.958 --> 00:04:43.761
And one of the points that we do try and make is that kids' views on these things absolutely do matter, and people who are trying to make laws around this stuff should be talking more to kids to actually get their perspective and not what adults assume kids' perspective is, so
00:04:43.836 --> 00:04:44.187
Yeah.
00:04:44.437 --> 00:04:44.817
Agreed.
00:04:44.846 --> 00:04:54.166
And there was actually one listener who I chatted to at a conference last year who said that they made their, son or daughter listen for the first time, and they subsequently went on to keep listening.
00:04:54.166 --> 00:04:57.221
So, um, perhaps that, person will get in touch.
00:04:57.360 --> 00:05:03.560
Obviously, kids don't email, so feel free to Snap me or whatever the the latest thing is.
00:05:03.781 --> 00:05:17.930
Um, listeners, if you, if you get to, the end of today's free episode and you want 20 minutes more analysis from us, or you wish the episode dropped yesterday, which is a kind of funny thing to say when you're talking about the future, you can become a Patreon supporter.
00:05:18.194 --> 00:05:27.711
You can find out more about that Go to patreon.com/ctrlaltspeech to see our tiers, to get the great benefits of becoming a supporter, and to help us keep this podcast going each week.
00:05:28.007 --> 00:05:34.117
big love to the supporters who already do fork out for those extended episodes and early episodes.
00:05:34.326 --> 00:05:39.901
Mike, we've got a, couple of really chunky stories, I think, this week that incorporate a lot of different elements.
00:05:40.204 --> 00:05:42.423
and so we're gonna do our best to kinda tie them all together.
00:05:42.454 --> 00:05:49.990
We're gonna start with, I guess, what has been the big news story across a number of different, beats, I'd say, this week, which is the UN General Assembly.
00:05:50.254 --> 00:05:57.187
All the world's major leaders in one place, all of them in a room talking about various things, including AI regulation.
00:05:57.552 --> 00:06:04.060
And, you know, your fair leader, Donald Trump, has has bestowed us with some great wisdom.
00:06:04.321 --> 00:06:08.690
Um, that's just one of a number of elements, I guess, that, we wanted to kinda touch on in the opening segment today.
00:06:09.040 --> 00:06:09.345
Yeah.
00:06:09.355 --> 00:06:27.661
Well, I will note that not all of the world's leaders are there because, uh, China's Xi did not go, though is now in the US, or I don't know if he's left at this point, but flew into DC to meet with Trump there, but did not go to New York and did not present at the UN General Assembly.
00:06:27.749 --> 00:06:28.079
No.
00:06:28.209 --> 00:06:38.273
There's some very good photos that I've been seeing today of them hanging out together, and I don't know if you remember the Obama-Xi Jinping, Winnie the Pooh episode?
00:06:38.314 --> 00:06:52.564
Oh, I am very familiar with the Winnie the Pooh and the fact that this is, this is an interesting, content moderation thing, which is that you're not really allowed to mention Winnie the Pooh in China on the internet because of the comparisons between Xi and, and Winnie the Pooh.
00:06:52.766 --> 00:06:53.096
Yeah.
00:06:53.117 --> 00:07:03.572
I was actually talking to a, a Ctrl-Alt-Speech listener, about that content moderation issue, and then I saw pictures of Trump and, Xi Jinping together, and I thought, it's kind of happening again." Um,
00:07:03.904 --> 00:07:06.334
see, you see the Winnie the Pooh, uh, comparison?
00:07:06.812 --> 00:07:07.562
Yeah, I do.
00:07:07.612 --> 00:07:10.632
It, it's not quite Tigger as Donald Trump, but, uh,
00:07:10.915 --> 00:07:11.795
I was gonna say,
00:07:11.872 --> 00:07:12.910
yeah, the s- the
00:07:12.942 --> 00:07:13.572
it's not, not...
00:07:13.612 --> 00:07:14.161
Ior.
00:07:15.050 --> 00:07:15.889
Yeah, yeah, yeah.
00:07:16.389 --> 00:07:18.000
So yeah, I'm, I'm expecting more memes.
00:07:18.199 --> 00:07:26.387
but kind of talk us through what you parsed from all the noise, about AI regulation and all, all of the different media reports that have come out of New York this week.
00:07:26.673 --> 00:07:32.363
yeah, obviously, as we've been discussing the last few weeks, there's been a whole bunch of discussion about AI regulation.
00:07:32.403 --> 00:07:44.257
I mean, this has been going on for quite some time, but it sort of felt like it was building to some sort of crescendo with, with all of the concerns about AI safety and everything and, discussions about regulation and all this kind of stuff.
00:07:44.257 --> 00:07:49.367
And you have all these world leaders together in one place, and you know that it's going to be a topic of conversation.
00:07:49.603 --> 00:07:54.913
And you had Donald Trump going up there and renaming AI.
00:07:54.994 --> 00:08:00.733
Uh, suddenly it's super intelligence, or as he said,"SI." We should now refer to it as SI.
00:08:01.053 --> 00:08:02.684
So, first of all, that's not gonna happen.
00:08:03.228 --> 00:08:08.543
as much as people dislike the term AI, it is the term, it is the term that is going to stick.
00:08:08.553 --> 00:08:09.824
It is what everyone is used to.
00:08:09.824 --> 00:08:11.314
There's no way to, change that.
00:08:11.625 --> 00:08:19.024
and it is sort of weird because one of the other announcements that he made is that he's launching an AI force, not an SI force.
00:08:19.024 --> 00:08:22.394
If he's changing it to SI, he maybe should have thought about that.
00:08:22.624 --> 00:08:33.053
But he's launching an AI force, which he is saying is modeled on his super successful launch of the Space Force as a another branch of the military.
00:08:33.328 --> 00:08:34.868
what does that mean?
00:08:35.341 --> 00:08:36.711
I don't know.
00:08:36.760 --> 00:08:57.740
And I think that is the larger summary about almost everything that happened, not just from Trump, but from all the world leaders who made grand pronouncements and there was a, you know, what is it, a declaration, from a bunch of UN members that w- like, this grand declaration about and regulation.
00:08:57.994 --> 00:09:01.471
and the summary is that, what does it mean?
00:09:01.471 --> 00:09:02.610
Uh, probably nothing.
00:09:02.874 --> 00:09:07.793
it feels like the sort of thing that everybody felt that they had to say something about.
00:09:08.073 --> 00:09:11.553
They had to announce these big grand plans, and we're going to deal with it.
00:09:11.563 --> 00:09:16.774
We know that it is a challenge, and we know that it needs to be regulated, and therefore, we are going to do something.
00:09:18.058 --> 00:09:20.416
But what does it actually mean?
00:09:20.426 --> 00:09:21.905
What are they actually going to do?
00:09:22.076 --> 00:09:23.365
Nobody has any idea.
00:09:23.395 --> 00:09:26.405
I don't think anyone has any concrete plans.
00:09:26.405 --> 00:09:28.725
I don't think anyone is putting anything into action.
00:09:28.765 --> 00:09:35.196
Everyone is just talking and saying stuff because they feel that they need to comment on it, but nobody seems to have a good idea.
00:09:35.196 --> 00:09:37.278
Nobody seems to know what is actually going to work.
00:09:37.491 --> 00:09:44.022
And so it was a, a bunch of nothing, and to some extent, that feels like a really wasted opportunity, right?
00:09:44.022 --> 00:09:48.229
You had all these people together, and they're not experts in AI.
00:09:48.249 --> 00:09:51.298
They're not experts in how to regulate this stuff, really.
00:09:51.698 --> 00:09:58.371
But you did have this opportunity and, and, you know, Sam Altman and Dario Amodei spoke and, talked to people.
00:09:58.572 --> 00:10:00.631
You had all these chances for education.
00:10:00.631 --> 00:10:05.058
You had all these chances to have, more serious and more thoughtful discussions.
00:10:05.359 --> 00:10:12.028
But, you know, maybe they were happening in the back rooms, but all the stuff that happened publicly felt like noise.
00:10:12.139 --> 00:10:14.229
It didn't feel like there was any useful signal.
00:10:14.229 --> 00:10:17.028
It didn't feel like there was anything that really moved the ball forward.
00:10:17.269 --> 00:10:29.808
Didn't feel like there was anything that was legitimate in any way that, both the governments of the world will actually, bother to do anything real with or that the people developing the technology are going to take seriously.
00:10:30.089 --> 00:10:33.219
So the whole thing felt like pageantry, right?
00:10:33.219 --> 00:10:35.908
it was all for show, which is a very Donald Trump thing.
00:10:35.918 --> 00:10:37.578
Everything he does is sort of pageantry.
00:10:37.918 --> 00:10:52.225
But there was very little out of all of the different announcements and speeches and talks and roundtables that came out of this that felt substantive, that felt like we're going to see some actual change that we can recognize and tie back to this moment.
00:10:52.535 --> 00:10:56.841
And so to me, it was, it was a lot of, noise and very, very little signal.
00:10:57.169 --> 00:11:05.493
Yeah, and a continuation, I'd say, of the kind of regulatory conversations that are being had about the major platforms and about the social media space, I'd say.
00:11:05.493 --> 00:11:17.200
If you go into the, comments of various presidents and prime ministers that spoke, they essentially say the same kind of thing, but applied to frontier labs, big AI companies.
00:11:17.470 --> 00:11:26.460
You know, you take Macron, he was saying, that should be a kind of open source initiative that doesn't make Europe bound to or tied to the US or China.
00:11:26.580 --> 00:11:33.509
So again, a kind of very similar type of message to his tech sovereignty piece that he's been, banging on about for the last few years.
00:11:33.820 --> 00:11:48.076
Andy Burnham, the UK Prime Minister, again, not known for his kind of tech credentials, but was talking about being an honest broker in the global discussion about AI, which I think is what UK has often been, because of where it's kind of geographically placed and, because of its influence.
00:11:48.273 --> 00:11:50.253
and that kind of happened in a few other cases as well.
00:11:50.283 --> 00:11:53.980
So it's, it's like these leaders didn't necessarily do anything new.
00:11:53.980 --> 00:12:00.559
They just kind of continued the same type of, approach, which I think is, probably what you're seeing, right?
00:12:00.620 --> 00:12:02.842
There was no, There's a stalemate as, as a result.
00:12:03.231 --> 00:12:12.586
it's understandable to some extent because I think the reality is that nobody's quite sure what to do and what will actually be effective, but they feel that they need to do something.
00:12:12.639 --> 00:12:20.308
you know, this is the same political conundrum that we had around social media, where there was this general feeling that something needed to be done,
00:12:20.746 --> 00:12:21.067
Mm.
00:12:21.178 --> 00:12:25.769
people would say things or, like, toss out proposals, and occasionally you get somebody doing something.
00:12:26.068 --> 00:12:28.719
But, like, these are really, really complicated topics.
00:12:28.719 --> 00:12:44.451
This is a thing that we are often discussing here that require a ton of nuance and a ton of understanding, and misdiagnosing the problems and the challenges lead to very bad solutions that probably make the, challenges and the problems much worse in the long run.
00:12:44.871 --> 00:12:54.432
And so to some extent, I'm almost glad that this was a lot of nothing because I don't think we really understand what are the real...
00:12:54.442 --> 00:12:58.381
Like, there's easy to sort of cherry-pick little things and say,"This is the problem.
00:12:58.381 --> 00:13:11.568
This is the thing that we're concerned about." There's the sort of, I, I think very distracting discussion on the sort of like the big sexy humanity is gonna die and, very cinematic kinds of stories that I think are pretty silly, and not very helpful.
00:13:11.902 --> 00:13:17.131
But we don't, we don't have a clear definition of what is the challenge that we're trying to solve for.
00:13:17.142 --> 00:13:19.029
What is it that we're trying to fix?
00:13:19.068 --> 00:13:20.639
what harms are we trying to prevent?
00:13:20.938 --> 00:13:25.701
These are all really important things, and they're really nuanced, and there's not-- we don't have clear answers for that.
00:13:25.711 --> 00:13:40.371
And because of that, it's probably better that we're not rushing into the breach and saying,"This is how we're going to fix the things that we haven't even decided are the actual problems yet." but it did feel like here was an opportunity to have some more thoughtful discussions about what, are the real harms?
00:13:40.371 --> 00:13:41.721
What are the real challenges?
00:13:41.721 --> 00:13:43.091
What are the real opportunities?
00:13:43.451 --> 00:13:54.916
We could have that open discussion, but the world is not great right now and hasn't been for some time, I think, in having serious adult conversations about nuance and trade-offs in all of these things.
00:13:55.186 --> 00:14:05.662
And so it may be for the best in the long run that, people, politicians especially feel that they need to speak up, and the media has to build stories out of it because that's what both sides do.
00:14:05.892 --> 00:14:22.011
and you know, the, tech folks are gonna do the tech folk thing and, act humble before the, the, leaders of the world, while at the same time, notably, both OpenAI and Anthropic released new models this week, even right after they were saying,"Well, we're pacing, we're pacing," blah, blah, blah.
00:14:22.336 --> 00:14:35.011
you know, this feels like this should've been a momentous and important week, and the reality is that I don't think we will ever, point back to this week as anything, you know, anything that comes in the future came out of what happened this week.
00:14:35.466 --> 00:14:35.635
Yeah.
00:14:35.696 --> 00:14:39.860
I mean, and, let's be fair to the UN and the kind of people who gather together.
00:14:39.876 --> 00:14:49.466
it might be the case that the conversations were had and that that will facilitate or catalyze a bunch of, smarter, more thoughtful decisions about how AI is regulated.
00:14:49.696 --> 00:14:52.342
there's only so much that a bunch of people can do in a few days in New York.
00:14:52.363 --> 00:14:57.546
But from what was kind of shared publicly, a lot of it felt very similar.
00:14:57.836 --> 00:15:17.870
I think it will come down-- Again, this is, this is a kind of broader question about where intergovernmental, bodies like the UN actually can play a role, and to what extent is it going to be states who are gonna eventually determine how this plays out through regulation that they impose on companies that act in their, jurisdictions, right?
00:15:17.870 --> 00:15:27.980
And I was reading an interview by, with rather, the UN under-secretary general, who's kind of coordinating a lot of the discussions on the sidelines about some of these regulatory letters.
00:15:27.980 --> 00:15:37.216
And he, he said, The emphasis lays with member states and their appetite for a more robust capacity for international oversight." It's kind of, we can only do so much as the UN, frankly.
00:15:37.363 --> 00:15:47.706
if states don't want to do it because they're pursuing a, growth agenda or they feel like AI is important for, economic development, then there's, you know, there's not much we can do, frankly.
00:15:47.956 --> 00:15:50.225
And, and I think that showed in the letter that you mentioned.
00:15:50.500 --> 00:15:55.409
22 signatories, a number of countries that you would expect to be on there weren't on there.
00:15:55.788 --> 00:16:05.364
So can kind of read between the lines and, and maybe guess who is taking that kind of growth innovation approach and, at least taking a back seat while other people figure it out.
00:16:05.533 --> 00:16:16.878
Yeah, and, the reality is, like, the more important discussion was probably the one that was happening a few hundred miles to the south in the White House between the Chinese and, and American, you know.
00:16:17.111 --> 00:16:27.217
I don't think Trump and Xi necessarily would have a, particularly useful or fruitful discussion on AI regulation, but I imagine some of the people around them probably did.
00:16:27.548 --> 00:16:35.280
But right now, the reality is that those are the only two countries that matter because all of the leading models are coming out of those two countries and there is...
00:16:35.441 --> 00:16:49.988
That fight is continuing, and it, feels very sort of cliché and useless to say, like,"Oh, it's this, you know, the battle between the US and Chinese models." but that is where the fight is happening, and that is where the discussion is happening.
00:16:49.998 --> 00:17:08.644
And if those two sides continue to battle it out, and the fear is that, if these tools continue to become more and more useful, as I think they are, and I know some people disagree with that, but I think it's pretty clear at this point that the tools have gotten more and more useful, there is a question of who's going to be providing the underlying, systems.
00:17:08.817 --> 00:17:12.988
And the Chinese models have gotten really, really, really good.
00:17:13.238 --> 00:17:29.867
they are behind the American frontier models, and there is an argument that the Chinese models are really built off of the backs of the American models, so that if the American model makers slow down, that, naturally the Chinese models will slow down as well because they're really relying on them in the first place.
00:17:30.238 --> 00:17:42.574
But, every other country, to some extent, doesn't matter as much if those two countries just keep their, feet on the gas in terms of, driving those companies to, build the next, more powerful model.
00:17:42.925 --> 00:17:43.165
Yeah.
00:17:43.165 --> 00:17:44.826
Can we talk about the US for a second?
00:17:45.364 --> 00:17:46.013
Uh, must we?
00:17:47.066 --> 00:17:50.655
because we, y- y- talked about, you made that kind of driving metaphor there.
00:17:50.665 --> 00:18:02.326
I was gonna say I'm getting kind of car sick by the l- you know, the very jerky left and rights that the US administration is taking in terms of where it's going with AI regulation.
00:18:02.536 --> 00:18:04.576
And there's a couple of more stories this week that showed that.
00:18:04.935 --> 00:18:08.885
You know, Trump announcing at the weekend that he was gonna appoint a new AI czar
00:18:09.423 --> 00:18:11.223
SIzar, AIzar, SI...
00:18:11.746 --> 00:18:11.826
czar,
00:18:11.963 --> 00:18:12.844
Yeah, yeah
00:18:13.135 --> 00:18:16.333
uh, without any other kind of congressman really knowing about it.
00:18:17.066 --> 00:18:29.385
then there was Treasury Secretary Scott Bessent saying on Monday that the government isn't gonna take liability away from the AI labs, which feels like a fairly important addition to the conversation, but wasn't really picked up.
00:18:29.603 --> 00:18:34.816
and raises a question about what an AI or AI czar might do if they are appointed.
00:18:35.031 --> 00:18:35.342
Yep.
00:18:35.343 --> 00:18:41.019
and now there's, there are reports that Bessent may be the leading, candidate to become the AI czar.
00:18:41.359 --> 00:18:43.336
So his, comments actually do matter.
00:18:43.375 --> 00:18:49.746
But apparently, there's a sort of knife fight going on within the administration over who gets to be the AI czar.
00:18:50.022 --> 00:18:53.230
and so there are a number of names that have, been proposed.
00:18:53.230 --> 00:18:56.306
There was a, an article that listed out a, bunch of these names.
00:18:56.675 --> 00:19:13.867
and, a bunch of them are sort of industry-connected people, a lot of people associated with, the venture capital firm a16z, uh, who are now in the administration who are potential AI czar, but there may also be Scott Bessent, who I don't think of as a particularly, knowledgeable person on AI.
00:19:13.907 --> 00:19:15.657
doesn't seem to have that much expertise.
00:19:15.798 --> 00:19:19.884
Well, doesn't seem to have that much expertise on lots of things that he's now opining on.
00:19:20.192 --> 00:19:25.744
But we've had AI czars before, and we had David Sacks for quite some time.
00:19:25.904 --> 00:19:41.407
you know, he's no longer the AI czar, but he's apparently still in the good graces of Trump and advising him on AI policy, and he's been very vocal about all these discussions and sort of mocking both Anthropic and OpenAI over the last few weeks and, and all the existential risk stuff.
00:19:41.435 --> 00:19:48.464
he's sort of been completely mocking them and saying that they're really just going for regulatory capture, which he might not be wrong about.
00:19:48.484 --> 00:19:51.565
I mean, there's lo- very little that I agree with David Sacks on.
00:19:51.565 --> 00:19:53.531
I, I think he's not great.
00:19:53.589 --> 00:19:56.055
but he might be right on that particular point.
00:19:56.284 --> 00:20:00.701
there've been a couple other people who have been very temporarily considered the AI czar.
00:20:01.011 --> 00:20:04.923
I think one of them lasted four days before before he was let go.
00:20:05.112 --> 00:20:11.892
so it'll be interesting to see who actually becomes the AI czar, but also if they have any real power or what it is that they can actually do.
00:20:12.082 --> 00:20:14.857
and again, like US policy is all over the map.
00:20:14.907 --> 00:20:34.765
You have some people in or around the administration who hate big tech and hate the AI companies and wanna punish them and limit them and all this stuff, and then you have other people who are like,"Let everything go." And you had Trump say that didn't need any guardrails other than your very smart and high IQ president, which, you know, is really concerning.
00:20:34.875 --> 00:20:38.984
I mean, he did talk about that, the AI force would be filled with high IQ people.
00:20:38.984 --> 00:20:44.491
He's a little obsessed with IQ, which is kind of embarrassing in all sorts of other ways.
00:20:44.531 --> 00:20:44.701
But
00:20:44.738 --> 00:20:52.894
he also said that the kind of dangers posed by AI, this was in his speech earlier this week, were similar to the climate change hoax.
00:20:53.074 --> 00:20:55.211
Um, hi- his words, not mine.
00:20:55.364 --> 00:20:59.307
important distinction there, um, for people who are listening.
00:20:59.351 --> 00:21:06.371
and so y- that mindset, that view is going to trickle down, to an AI czar or to any kind of approach to AI regulation, I think so.
00:21:06.763 --> 00:21:09.661
it's a hell of a ride kind of watching what the US is doing, I'd say
00:21:09.689 --> 00:21:21.939
Yeah, I mean, the simple fact is that we don't have adult leadership in the US government right now, and that is a problem when you want to have serious discussions about how do you deal with this technology, whether it is through regulatory means or, or other means.
00:21:22.266 --> 00:21:26.368
it would be nice to have, like, serious people in charge, and we just don't have that.
00:21:26.538 --> 00:21:28.692
So that's the reality that we have to deal with.
00:21:28.976 --> 00:21:29.336
Yeah.
00:21:30.420 --> 00:21:31.849
I don't, I don't have anything to say about that.
00:21:33.000 --> 00:21:37.519
Can we find some adults to preside over the US administration, please?
00:21:38.125 --> 00:21:41.016
perfect segue, I'd say, Mike, to our eye-catching section.
00:21:41.398 --> 00:21:55.823
we're, we're bringing a bit of hope this week because there are a few stories that I think are kind of evidence of internet users embarking on governance and, partaking in acts of internet governance in ways that I think are really, really hopeful and we wanted to flag those to
00:21:56.345 --> 00:21:56.835
Hope?
00:21:57.075 --> 00:21:58.125
On this podcast?
00:21:58.174 --> 00:22:00.065
Ben, what is happening?
00:22:00.454 --> 00:22:01.644
Is this a cry for help?
00:22:03.093 --> 00:22:03.653
Don't ask
00:22:03.654 --> 00:22:05.265
We don't do hope here.
00:22:06.232 --> 00:22:06.442
Don't
00:22:06.464 --> 00:22:11.005
We do trade-offs and sadness and misery all around.
00:22:11.007 --> 00:22:11.396
Yeah.
00:22:11.396 --> 00:22:16.196
I g- I think I got six hours sleep last night for the first time in a few months, and it's had an effect on me.
00:22:16.527 --> 00:22:19.940
so, let's go through those and, and, you know, you can be the judge.
00:22:20.116 --> 00:22:26.166
I mean, the, the first thing to kind of flag is this, great new piece from New Public looking at a service called MySky.
00:22:26.407 --> 00:22:29.519
you know, MySky has, has actually been around, A while, let's say.
00:22:29.720 --> 00:22:35.374
I haven't actually been in partaking in it, and, and so I only found out about it as a result of this New Public piece.
00:22:35.374 --> 00:22:41.920
And it's a great interview with Rene Di Resta and Jonathan Stray, who came up with the idea and have kind of rolled it out.
00:22:42.194 --> 00:22:51.997
we talk about algorithms all the time on the podcast, Mike, platforms offer algorithms, but it's incredibly hard to see how they work, why you're being recommended certain content, and you have no control over it.
00:22:52.126 --> 00:22:57.644
You know, you can like certain posts or hover over certain, videos maybe to help tell the algorithm what you're into.
00:22:58.001 --> 00:22:58.480
that's it.
00:22:58.500 --> 00:22:59.599
You have very little control.
00:22:59.920 --> 00:23:11.230
And what MySky does is that it uses the AT protocol and is, accessible via BlueSky to give users some, customizable dials to shape that content that you see.
00:23:11.390 --> 00:23:14.970
And so, so I, because I found out about it this week, I went on and I, did that.
00:23:15.199 --> 00:23:20.529
and you can do that by looking at how much time you want the content to be served over.
00:23:20.529 --> 00:23:23.619
So if you want just to look at content over the last 24 hours, you can do that.
00:23:23.890 --> 00:23:25.339
You can do it over a week if you want.
00:23:25.696 --> 00:23:31.034
it can be kinda tweaked according to whether you want to see more content from people you follow versus people who aren't.
00:23:31.364 --> 00:23:35.344
You can do it based upon whether it's, you know, content is popular on the platform and trending.
00:23:35.614 --> 00:23:37.700
And so it's a really, it's a great way.
00:23:37.700 --> 00:23:41.180
I didn't expect to enjoy it as much as I, as, as much as I did.
00:23:41.539 --> 00:23:42.829
And now it won't be for everyone.
00:23:43.017 --> 00:23:45.336
it's not gonna be something that everyone will go on and do.
00:23:45.727 --> 00:23:50.568
But I think it's a great example that we should hold up as, as something that's positive and, yeah, hopeful.
00:23:50.798 --> 00:23:51.104
Yeah.
00:23:51.115 --> 00:24:07.663
it's really wonderful and, actually if you go back to my original"Protocols, Not Platforms" paper, which in part inspired the At Protocol in the first place, this was part of the discussion that I had, was that, it would be great if people could, you know, manually tweak their own dials, right?
00:24:07.683 --> 00:24:31.758
I mean, I think so much of the concern, and including much of the, the sort of general hatred of algorithms, where a lot of people are like,"Oh, algorithms bad," which, I've always had an, an issue with, is because of the lack of transparency, that they don't understand who is making the decisions, who is turning the knobs, who is deciding what it is that you wanna see, and therefore there's the opportunity for there to be manipulation.
00:24:31.820 --> 00:24:42.478
a lot of people have the natural concern and they think that, oh, it's like, you know, I'm being manipulated to make me angry because, you know, a lot of people are fed a lot of content that makes them feel angry, and so they, they fear that.
00:24:42.478 --> 00:24:46.077
Or, or it might be, you know, to try and sell me stuff or whatever it is.
00:24:46.077 --> 00:25:03.875
And there's just a lot of concern from that lack of transparency, and this brings in You know, it gives you so much transparency and control over this and, and so you can understand why certain things were recommended, and then you can adjust that if you wanna see, different types of things or different reasons for having things recommended to you.
00:25:04.075 --> 00:25:07.085
Having that control is really, really powerful.
00:25:07.198 --> 00:25:18.557
you know, it's probably too much and overwhelming for some people, but just having it there and having the ability to, play around and turn the knobs yourself, and even to learn like what that means.
00:25:18.798 --> 00:25:28.131
Because again, like, you know, when you have this assumption that all the, algorithmic decisions are done for evil reasons, sometimes you don't even realize like how things turn out.
00:25:28.151 --> 00:25:44.904
Even if you have control of the knobs, you might begin to realize like,"Oh, it's interesting to see what things get recommended when I, tweak the knobs this way or that." And so this, as a demonstration of that and clearly showing it and doing it in a sort of nice UI fashion, I think is really, really powerful and educational at the same time.
00:25:44.924 --> 00:25:47.055
And so I'm, I'm a huge fan of this project.
00:25:47.055 --> 00:25:48.157
I think it's, fantastic.
00:25:48.415 --> 00:25:49.165
Yeah, agreed.
00:25:49.306 --> 00:25:53.296
I'm gonna update listeners when I have a further tweak, in the coming weeks.
00:25:53.296 --> 00:25:57.893
But great start, great example of, people building, in interesting ways.
00:25:58.202 --> 00:26:02.252
and we're seeing, you know, we're seeing the influence of this spread elsewhere.
00:26:02.252 --> 00:26:06.945
And so YouTube just announced giving users more control over their algorithms.
00:26:06.955 --> 00:26:17.326
It's not nearly to this level at all, and it's not to the level of, like, even a Blue Sky where, you can bring in your own algorithm and, and do things that way.
00:26:17.546 --> 00:26:23.425
But, I think we're beginning to see this recognition and, you know, we've seen others do similar things.
00:26:23.425 --> 00:26:31.756
Threads, people would joke, on Threads about dear algo, like, you know, trying to tell the algo what to do, and then Threads actually implemented that.
00:26:31.836 --> 00:26:40.875
So allowing users to have some influence, direct influence over the algorithm is now becoming more of a standard thing, and I think that's a good thing.
00:26:41.096 --> 00:26:45.445
And the My Sky feed is an extreme version of that, but a really cool one.
00:26:45.933 --> 00:26:46.683
Yeah, agreed.
00:26:47.080 --> 00:27:10.911
talking of YouTube, Mike, we shouldn't gloss over the fact that at the event, the, the Made On YouTube event that, a lot of this was announced this week, Neal Mohan, who we've covered on the podcast before, he's the CEO of YouTube, he said that YouTube wouldn't stump up the£6 billion as part of the big Meta settlement, the multi-state settlement that we covered on the podcast a few weeks ago.
00:27:11.010 --> 00:27:14.631
and he claimed that YouTube has been focusing on child safety for a long time.
00:27:14.631 --> 00:27:23.490
It would continue to do so, and basically that means that the Meta will now pay 70% of that overall 18 billion uh,'cause that was the kind of deal, right?
00:27:23.528 --> 00:27:23.977
we can get
00:27:23.980 --> 00:27:39.141
So yeah, a reminder, some of the payment that Meta would make to the states was conditional on getting TikTok and YouTube to agree to the same restrictions, which is, to some extent an anti-competitive, you know, way to, build an industry standard.
00:27:39.346 --> 00:27:52.234
so this would save Meta some money, obviously, save YouTube some money if they're not paying into the system as well, but also means that Meta might not then be able to define the specific features and standards.
00:27:52.654 --> 00:28:07.627
And, you know, one of the points that I had raised regarding that whole discussion was, this was Meta deciding for the rest of the industry what they wanted to do, and my fear was that meant that we wouldn't get more interesting experimentation from others or, different attempts to do different things.
00:28:08.028 --> 00:28:26.214
Whereas, what Neil Mohan is saying here is basically like:"Hey, we have other ideas on what actually helps keep kids safe, and we don't wanna have our hands tied by some other company and a bunch of state AGs that tell us what we have to do when we actually have better ideas." And so that is encouraging.
00:28:26.345 --> 00:28:37.141
I still think what we will likely see is states and state AGs continuing to go after YouTube and pressuring them to accept these same sorts of feature changes.
00:28:37.451 --> 00:28:50.381
But, I think it's a good thing Because, you know, so far, TikTok and YouTube had been pretty quiet in re- in response to, to Meta's, all-out war to try and get them to sign on to the same deal.
00:28:50.580 --> 00:28:54.411
And so it's good for him to just take a stand and say like:"No.
00:28:54.451 --> 00:28:55.417
Look, that's stupid.
00:28:55.417 --> 00:28:56.218
We're not gonna do that."
00:28:56.579 --> 00:28:56.849
Yeah.
00:28:56.880 --> 00:28:59.069
And the question is what, what will Meta do next?
00:28:59.230 --> 00:29:03.813
Because I expect they will have a plan for f- for this.
00:29:04.173 --> 00:29:20.083
You know, I, I said on the podcast that, I feel like they were playing checkers and everyone else was playing chess when they concocted this idea, and I wouldn't be surprised if they have a next move, because it was fairly predictable that YouTube would turn around and say,"We're not paying." so I don't know what it will be, but I think they'll have something up their sleeve.
00:29:20.166 --> 00:29:20.507
Yeah.
00:29:20.527 --> 00:29:24.683
I mean, it's, it's not even the payment I think that matters so much as the feature changes, right?
00:29:24.693 --> 00:29:26.824
That, that's the, the bigger issue to me.
00:29:27.095 --> 00:29:35.334
and again, like the features, and I don't necessarily think that the features that Meta agreed to put in place even make sense on Instagram and, Facebook.
00:29:35.630 --> 00:29:40.124
but, YouTube is a very different platform, and I, I like, yeah, I keep trying to make this point too.
00:29:40.124 --> 00:29:44.933
Like, each of these platforms, uh, is used differently and in different ways and with different audiences.
00:29:45.324 --> 00:29:50.540
And YouTube especially is much more of a, a sit back and, watch kind of platform.
00:29:50.540 --> 00:30:01.547
It's much more akin to television than traditional social media, and yet some of these feature requirements that Meta would be placing on them are really specific to specifically social media.
00:30:01.856 --> 00:30:04.207
And so the whole thing seems very strange.
00:30:04.227 --> 00:30:10.020
It would be nice for YouTube to, take a really strong stand here and say like, No, that's, that's stupid for us.
00:30:10.020 --> 00:30:10.790
We're not gonna do that."
00:30:11.048 --> 00:30:11.368
Yeah.
00:30:11.749 --> 00:30:15.249
we're, we're doing what we never do, Mike, which is kind of rapidly going through our eye-catching stories.
00:30:15.269 --> 00:30:16.439
So we can, fit...
00:30:16.525 --> 00:30:19.894
one more eye-catching story in before we go to our extended Patreon episode.
00:30:20.214 --> 00:30:26.858
And, again, fun example of, users on Reddit this time enacting a bit of power to define what, their online space looks like
00:30:27.282 --> 00:30:28.818
Yeah, I really love this story.
00:30:28.929 --> 00:30:34.659
I think this is, this is a really interesting, example of sort of community governance, for interesting challenges.
00:30:34.659 --> 00:30:44.721
We're sort of used to so many of the trust and safety stories that we talk about, or content moderation stories that we talk about, are sort of big companies making decisions and, going back to what I was saying before with the algorithms.
00:30:44.991 --> 00:30:58.781
Like, there is some feeling of uncomfortableness with the idea that you have big, powerful companies with different interests than you making big decisions about what kind of speech or what kinds of things you can do online, and that, it has sort of a weird feeling.
00:30:59.067 --> 00:31:06.501
But, we know that, communities have, setups that, have rules and, and moderation themselves, and seeing that in action is really interesting.
00:31:06.501 --> 00:31:12.442
So this is a story about a specific subreddit, the Crafts subreddit, which is a very big subreddit,
00:31:12.644 --> 00:31:13.864
I'm not a member, but, but
00:31:13.951 --> 00:31:15.531
I am not a member either.
00:31:15.571 --> 00:31:21.989
Uh, but, you know, it talks about crafting and making stuff and knitting and, and all sorts of things along those lines.
00:31:22.308 --> 00:31:31.134
And there was a debate about whether or not people should be allowed to post Harry crafts and discussions about them.
00:31:31.535 --> 00:31:38.795
And what was happening is that, as I'm sure many listeners on the podcast know, J.K.
00:31:38.795 --> 00:31:43.525
Rowling, the author of the Harry Potter books, has become somewhat controversial and
00:31:43.722 --> 00:31:45.093
Don't start unpacking that
00:31:45.472 --> 00:31:48.817
I will not unpack that other than, you know, yes.
00:31:49.106 --> 00:32:02.363
She's become very controversial because she's said some things that have upset lots of people and, and I would say, the one thing I will say on this is I think has done a lot of damage, that was, really unnecessary and, deeply harmful.
00:32:02.553 --> 00:32:12.272
and so that has led to whenever people are posting"Harry Potter" related crafts, the discussion related to them has gone off the rails within this particular subreddit.
00:32:12.613 --> 00:32:23.242
And that has become a moderation challenge for the moderators within that subreddit, and often those discussions have nothing to do with the actual crafts and very much everything to do with J.K.
00:32:23.242 --> 00:32:26.303
Rowling and the mess that she has made.
00:32:26.740 --> 00:32:35.047
and so the entire subreddit basically had a discussion of whether or not we should just stop allowing any discussions on"Harry Potter" related crafts.
00:32:35.497 --> 00:32:47.817
And they had an incredibly open and pretty thorough process in which they asked everybody to weigh in, and there was a big discussion and then voting, and it wasn't a entirely one-sided thing.
00:32:47.826 --> 00:32:50.009
It w- there were, viewpoints across the board.
00:32:50.240 --> 00:33:01.336
And then on the voting, it sort of narrowly came out with a narrow majority saying that they thought that we should just ban discussions of, Harry Potter" crafts.
00:33:01.356 --> 00:33:05.920
And so the moderator team of the subreddit says that that is the new rule.
00:33:06.136 --> 00:33:10.960
you cannot post anything related to"Harry Potter." You can post things that are...
00:33:11.109 --> 00:33:19.522
If it's not directly, if it just sort of looks like a"Harry Potter" thing but doesn't directly reference"Harry Potter" or that world, you can still do that.
00:33:19.769 --> 00:33:35.136
but what I thought was, was really interesting was the openness of this discussion and the thoughtfulness, and people talked about like slippery slope,"Oh, now we're getting political, now we're gonna ban this." And, what they said very, very clearly is like,"No." this is not even a political statement about J.K.
00:33:35.136 --> 00:33:37.146
Rowling or her views on things.
00:33:37.366 --> 00:33:48.653
It is very much that it becomes impossible to have a sensible community when any time any"Harry Potter" content shows up here, the conversation just goes completely off the rails.
00:33:48.932 --> 00:33:53.997
And so it's almost a, you know, a, form of preservation to say,"We're just not gonna deal with that.
00:33:54.096 --> 00:33:56.436
That kind of content is just not allowed."
00:33:56.615 --> 00:33:56.944
Yeah.
00:33:57.095 --> 00:34:04.980
I mean, like massive credit to these moderators, you know, which we forget and we gloss over because Reddit is, essentially kind of helmed by these subreddit moderators.
00:34:04.980 --> 00:34:21.385
But these are volunteers who are, moderating gigantic communities, at some points with often hundreds of thousands or millions of users, and they're doing this very kind of deliberative, inclusive approach to how standards and, and rules are set, and I think that's fantastic.
00:34:21.675 --> 00:34:28.385
We've seen examples of these kind of community council, or jury approaches working really well in some, in some areas.
00:34:28.385 --> 00:34:39.045
And I think, again, like it's not very sexy'cause it's kind of slow and, deliberate, and it involves polls and processes and being transparent and documenting thinking.
00:34:39.262 --> 00:34:43.541
but again, that's, what I think the grunt work of, looking after online spaces is about.
00:34:43.898 --> 00:34:44.208
Yeah.
00:34:44.208 --> 00:35:21.538
I mean, there's a lot of work and a lot of research and, plenty of academic research on sort of like community governance that often gets ignored in these discussions and-- but I think is really, really important and sort of understanding that and understanding that, community governance really, really matters, and it's not all just like one super rich, super powerful dude deciding how the community works, and recognizing these other kinds of models and ones that are more engaging and involve people and actually, try to come to some sort of agreement recognizing pros and cons to different approaches and just saying like,"This is the reality, this is what we're trying to deal with.
00:35:21.557 --> 00:35:44.150
how should we best deal with it?" And coming to a conclusion saying,"These are the new rules based on this open discussion," is, whether or not you agree with any of the, the position, it's somewhat refreshing and somewhat encouraging that, like, a community can make an adult decision that is like, let's talk about the trade-offs, let's talk about the different issues here, and then let's just make a call on what's gonna be best for our community and go with it." that's, really great.
00:35:44.413 --> 00:35:45.152
Yeah, agreed.
00:35:45.391 --> 00:35:49.932
we're gonna touch on a bit of that in the extended Patreon episode, Mike, which we'll go into now.
00:35:50.208 --> 00:35:51.016
we'll wrap up today.
00:35:51.016 --> 00:35:57.106
I, I hope listeners don't feel too much hope and positivity because that's not really our thing.
00:35:57.295 --> 00:35:59.088
We, we could, we could change the podcast.
00:36:00.246 --> 00:36:01.905
we can, We can be hopeful for the future
00:36:01.952 --> 00:36:02.262
Yeah.
00:36:02.302 --> 00:36:02.862
Yeah, okay.
00:36:02.893 --> 00:36:04.092
Well, don't promise anything too much.
00:36:04.293 --> 00:36:05.972
Um, listeners, thanks for joining us.
00:36:06.030 --> 00:36:09.327
if you enjoy the podcast, rate and review it wherever you get your podcasts.
00:36:09.606 --> 00:36:14.067
become a Patreon supporter to get this next segment, and we'll jump into that now.
00:36:14.117 --> 00:36:26.936
It's all about Discord's age assurance rollout and how Ofcom is going after Aylo, the owner of the largest porn site in the world, Pornhub, and what that means for the future of the internet
00:36:29.077 --> 00:36:32.358
That's the end of this week's free edition of Ctrl Alt Speech.
00:36:32.739 --> 00:36:39.617
To hear the extended episode with additional coverage, become a Patreon supporter at patreon.com/ctrlaltspeech.
00:36:39.878 --> 00:36:42.418
That's C-T-R-L Alt Speech.
00:36:42.998 --> 00:36:49.418
If your organization is interested in sponsoring the podcast, contact us via our website at ctrlaltspeech.com.
00:36:49.858 --> 00:36:50.577
Thanks for listening
