สำเนาบทสนทนา 🔗
WEBVTT
00:00:02.720 --> 00:00:11.759
Hello and welcome to a podcast where we use the parallel worlding framework to explore what could have been and what might still be possible.
00:00:11.759 --> 00:00:14.800
It's great to have your company, my name's Daz Chandler.
00:00:14.800 --> 00:00:25.199
In this episode, we're stepping into a world where societies refused to treat technology as a neutral force, or an inevitable march toward profit and power.
00:00:25.199 --> 00:00:32.880
Instead, digital systems were shaped deliberately in service of justice, care, and collective accountability.
00:00:32.880 --> 00:00:39.039
Big tech, as we know it here, either never took hold or was transformed from within.
00:00:39.039 --> 00:00:42.399
Communities governed their own digital infrastructures.
00:00:42.399 --> 00:00:46.719
Data was treated as a shared responsibility, not a commodity.
00:00:46.719 --> 00:00:55.200
And innovation was measured not by scale or speed, but by its capacity to support human dignity and planetary well-being.
00:00:55.200 --> 00:01:01.920
So how does that world work and what can it teach us about where we currently find ourselves in this present reality?
00:01:01.920 --> 00:01:06.400
To help us explore that question, we'll soon be joined by Many World's Traveller, Dr.
00:01:06.400 --> 00:01:14.560
Alex Hanna, who's spent years interrogating the power, politics and possibilities of our digital past, present, and future.
00:01:14.560 --> 00:01:17.120
This is The Parallel Effect. Alex Hanna, thanks for joining us.
00:01:41.840 --> 00:01:43.040
Thank you for having me.
00:01:43.359 --> 00:02:00.560
So in our current reality, AI systems are largely developed within white, affluent countries, embedding biases and assumptions that fail to reflect the cultural, social, and economic contexts of the global majority...
00:02:00.560 --> 00:02:15.840
so in the parallel world you visited, how have design principles been transformed to ensure that AI systems genuinely serve diverse populations rather than the interests of powerful corporations and governments?
00:02:16.319 --> 00:02:17.360
Yes, absolutely.
00:02:17.360 --> 00:02:20.319
So I think there's a lot in here...
00:02:20.319 --> 00:02:35.199
The focus on bias often makes us go down a path in which we think about what data is being used to say train different machine learning or AI models, including people and growing that scope.
00:02:35.199 --> 00:02:45.439
But I think what we need to do is take a step back and really think about how that as a mechanism in a frame gets easily co-opted.
00:02:45.439 --> 00:03:00.639
So talking about bias often means, well, all we need to do is integrate this other type of data and create a system which not only addresses those in white, affluent countries, but also addresses those in the majority world.
00:03:00.639 --> 00:03:12.639
But the problem is that quickly goes into modes of inclusion which are predatory: if you are going to participate with us, we're going to have to take your data.
00:03:12.639 --> 00:03:16.479
And this is the condition of being included at all.
00:03:16.479 --> 00:03:22.319
And that itself leads to these very pernicious types of incentive structures.
00:03:22.319 --> 00:03:36.240
Many companies actually in China are having arrangements with people in governments in, say, Zimbabwe to give, for instance, photos of their citizens to train those models.
00:03:36.240 --> 00:03:40.080
But the citizens of Zimbabwe are not getting the benefit of it.
00:03:40.080 --> 00:04:11.759
And so if we're going to step back and think of a parallel world, we have to think about how technology would actually serve people, what kind of technologies that we need to have, but also their governance structures, also where the computing is happening, how the computing is happening in such a way that is not environmentally ruinous, how we're respecting the labour that goes into the construction of those systems...
00:04:11.759 --> 00:04:25.120
So a lot of that means we have to incorporate people into design processes, but also think about what are alternative governance and economic structures of designing technology?
00:04:25.120 --> 00:04:35.759
And, you know, a lot of that goes back to really critical questions about government and ownership of not only the means of production, but the means of computation.
00:04:35.759 --> 00:04:41.920
And there's a lot of different kinds of design principles that folks have made to think about that.
00:04:41.920 --> 00:04:48.000
I mean, one of the frameworks is the Design Justice Network's principles.
00:04:48.000 --> 00:04:54.879
So I think this has been well articulated by folks like uh Sasha Costanza-Chock in her book, 'Design Justice'.
00:04:54.879 --> 00:05:03.439
But also, thinking about other kinds of frameworks that folks have really proffered for technology development.
00:05:03.439 --> 00:05:29.279
And so here, I'm thinking with folks who do things on Indigenous data sovereignty, for instance, and really thinking about what it means to consider and honour data itself and to know that differential access is very important if you're thinking about things like oral histories, or recordings of indigenous languages, and treating those as sacred artefacts.
00:05:29.279 --> 00:05:46.560
And then also folks coming from the critical disability justice movement as well, and thinking about the kind of principles of making sure that people's accessibility needs are met and that these things that get attached to technology are not just what are called "disability dongles".
00:05:46.560 --> 00:05:54.560
I don't know who coined that, but it's kind of a fix for things rather than really building with disabled people in front of mind.
00:05:54.560 --> 00:06:00.000
It's not just about the tech, it's not just about the data, it's about the full stack and the governance structure.
00:06:00.480 --> 00:06:01.040
Okay, sure.
00:06:01.040 --> 00:06:06.879
So let's
Load full transcript
take a closer look then at the governance structures at play in the parallel world.
00:06:06.879 --> 00:06:12.000
Are there any global agreements or governance structures...
00:06:12.000 --> 00:06:26.720
uh perhaps even something like an AI human rights treaty, that ensure AI technologies are developed and used in ways that uphold universal human rights, that protect the environment and generally serve the public good?
00:06:26.720 --> 00:06:30.000
And if so, how do these mechanisms work?
00:06:30.959 --> 00:06:32.959
Yeah, I mean, what a good question.
00:06:32.959 --> 00:06:42.399
I mean, I think maybe not starting so top-down and saying that we need some kind of a treaty, but really thinking about what are kinds of ways that we can think bottom-up?
00:06:42.399 --> 00:06:54.000
What are ways that we ensure that we have allocations in which people are making decisions about technology that are very locally structured?
00:06:54.000 --> 00:07:01.839
We have a situation right now where compute is happening all around the world and making climate change worse.
00:07:01.839 --> 00:07:09.839
And that's so people in the West can, you know, use ChatGPT, or generate an image, or generate a short movie.
00:07:09.839 --> 00:07:20.480
But that's making areas in the world which are already affected by climate change more desolate, more um, more exposed to climate catastrophe.
00:07:20.480 --> 00:07:32.160
What would it mean to have local structures or local governance structures in which individuals can decide on what their technological needs are and then make some kind of decision or allocation towards that?
00:07:32.160 --> 00:07:41.680
And especially, in this parallel realm view, I'm thinking a lot with folks like Erik Olin Wright in his work on 'Envisioning Real Utopias'.
00:07:41.680 --> 00:07:50.319
And a lot of what's in that book is a framework for thinking about who's claiming economic, social, and political control over the economy.
00:07:50.319 --> 00:07:58.079
And then also there's some really interesting examples there about practices in which these are approaching that case.
00:07:58.079 --> 00:08:16.079
And so one of the practices is this process of participatory budgeting that's happened in cities like Porto Alegre in Brazil, and the ways in which local neighbourhood councils are making assessments of their own economic needs and their city, city budgetary needs.
00:08:16.079 --> 00:08:23.279
So I'd imagine there'd be some kind of way in which there's local councils that are deciding on their technological needs.
00:08:23.279 --> 00:08:32.159
And I don't think we need to call them "AI needs", or you know, we need something like Sam Altman has said: "universal basic compute", because I don't know what that means.
00:08:32.159 --> 00:08:34.879
It's something that needs to be closer to...
00:08:34.879 --> 00:08:43.360
Honestly, there needs to be people who can fix the computers, [laughs] and ensure that our elders are able to use them.
00:08:43.360 --> 00:08:50.480
And maybe, there's technology that gets developed on a certain level where it makes it accessible to our elders.
00:08:50.480 --> 00:08:58.159
Myself and Timnit Gebru and Asmelash Teka, we wrote an article which is called 'An Internet for Our Elders'.
00:08:58.159 --> 00:09:04.159
And what that talks about is how, you know, our grandmothers had no idea how to use...
00:09:04.159 --> 00:09:09.759
My last grandmother passed away when I was young, and I don't think really interacted with the internet.
00:09:09.759 --> 00:09:14.799
But if she had been around, or still was around, what would that look like?
00:09:14.799 --> 00:09:34.879
I think she would have to consider thinking about how to type in say Egyptian Arabic or in Coptic, you know, because many people from the early 20th century still do speak in Coptic, which is a language that's primarily uh used in the Coptic church right now, but is something that is...
00:09:34.879 --> 00:09:42.320
is, or was, spoken in homes and descends from Middle Egyptian and it is an indigenous language.
00:09:42.320 --> 00:09:49.679
And so if I wanted my grandmother to interact on the internet, she might want to interact in Egyptian Arabic, Coptic...
00:09:49.679 --> 00:09:57.759
To interact with the internet now, you have to type, or you have to try to hope some of these automatic speech recognition technologies work.
00:09:57.759 --> 00:10:06.320
So I think it would look a lot different, and technological structures would be very much tied to those local governance structures.
00:10:06.879 --> 00:10:18.159
Here, as we know, AI development comes with significant environmental costs from mining to high energy consumption, to growing e-waste...
00:10:18.159 --> 00:10:24.960
So in the parallel world you visited, how are these environmental impacts understood and mitigated?
00:10:24.960 --> 00:10:29.840
What allows AI, if anything, to develop sustainably there?
00:10:30.480 --> 00:10:34.080
Well, you have to think about what, I mean, how it's developed right now.
00:10:34.080 --> 00:10:35.200
I mean, it's...
00:10:35.200 --> 00:10:45.360
Right now we've got a handful of companies that are producing hardware and doing chip fabrication and mining rare earth minerals.
00:10:45.360 --> 00:11:06.960
I mean, the whole stack is exploitative all the way down from the coltan and the lithium that gets mined in the DRC under conflict conditions with child labour, to the chip wafer fabrication that's done all over the world, under the auspices of Intel and TSMC...
00:11:06.960 --> 00:11:23.679
and these different chip manufacturing companies produce incredible amounts of waste in terms of the kinds of environmental runoff, the forever chemicals, to the implementation in data centers and the scaling up of hyperscalers.
00:11:23.679 --> 00:11:26.799
And so, I mean, there's so many different elements of that, right?
00:11:26.799 --> 00:11:29.679
I mean, from top to bottom.
00:11:29.679 -->
00:11:47.519
It's hard to imagine, I mean, just given that we have these tools that are so ubiquitous and we're being asked to buy new phones every two years and buy new computers, and there's all these rare earth elements that go into that.
00:11:47.519 --> 00:11:52.240
And so, thinking about sustainability would have to be completely different.
00:11:52.240 --> 00:12:04.559
It would have to be: these elements are mined, you know, in a way that is not destroying that earth in the Congo, in Chile, at the lithium, salt baths...
00:12:04.559 --> 00:12:14.399
That would entail kind of a different relationship to the land, to how we extract that, to the kind of labour that it takes to extract that.
00:12:14.399 --> 00:12:35.200
And that would mean that there's a need to address the core periphery dynamics of mineral extraction and the exploitation of big, multinational tech companies, and semiconductor manufacturers, and these intermediaries that sell the rare earth elements.
00:12:35.200 --> 00:12:36.320
I think there's a...
00:12:36.320 --> 00:12:47.360
such a focus on data centres now because of the water consumption and the energy consumption and the air pollution, and rightly so.
00:12:47.360 --> 00:12:49.360
Those are major issues.
00:12:49.360 --> 00:13:02.559
But also know that those would not be possible if not for the massive mineral trade that goes on in the semiconductor manufacturing and how these are all elements in that supply chain.
00:13:02.559 --> 00:13:06.080
So, what would it mean to imagine that?
00:13:06.080 --> 00:13:31.279
It would mean, you know, these kinds of deals that I think would would not rely on that conflict labour, it would rely on trade that was fair, it would rely on producing those in a ways in which the chemical elements could be mitigated through materials engineering, or other types of waste management that is actually thoughtful.
00:13:31.279 --> 00:13:35.600
And there's effort put into the research to mitigating that.
00:13:35.600 --> 00:13:46.879
I'm not an expert in those areas, but also have seen some indication that they can do this in such a way that is protective of workers and of the environment.
00:13:46.879 --> 00:13:55.039
And then I think if you're developing local models for local usage, I mean the models don't have to be super huge.
00:13:55.039 --> 00:14:04.639
They don't have to be trained on hundreds of thousands of GPUs or TPUs to make them these 'everything machines'.
00:14:04.639 --> 00:14:10.080
The models can be smaller and they don't have to require this intensive training element.
00:14:10.080 --> 00:14:19.039
And then I think the other part of that is if the models are being used locally, then you can think about what would it mean to have local energy suppliers provide that?
00:14:19.039 --> 00:14:23.679
And I guess like the other element of it is like the models don't have to be working all the time.
00:14:23.679 --> 00:14:25.919
Like, if it's not working, it's not working.
00:14:25.919 --> 00:14:28.320
Like the sun's not out, you don't get your thing today.
00:14:28.320 --> 00:14:29.600
Like, read a book, I don't know?
00:14:29.600 --> 00:14:30.320
Do something else.
00:14:30.320 --> 00:14:34.240
Like it doesn't have to be always ready, always up.
00:14:34.240 --> 00:14:40.720
Also, this culture of like, you know, 99.99% SLAs that need to be fulfilled...
00:14:40.720 --> 00:14:42.480
I mean, we don't have to live like that!
00:14:42.960 --> 00:14:55.679
Coming from our present reality, it's quite challenging to perceive a realm where people can contribute data without risking surveillance or exploitation.
00:14:55.679 --> 00:15:10.720
So when it comes to sharing data in the parallel world, what sort of consent protocols are in place that enable and empower individuals and communities to share their data without fear of harm?
00:15:11.440 --> 00:15:15.440
Yeah, I mean, I think that's a really important thing to highlight.
00:15:15.440 --> 00:15:35.039
And so one of the things that I think about a bit is the Traditional Knowledge labels, or TK labels, which are discerning in a community who has access to particular types of data or knowledge and limiting based on that.
00:15:35.039 --> 00:15:45.919
I think another kind of thing that we could consider is our schemas in which consent has to be continually offered, and can be revoked at any time.
00:15:45.919 --> 00:15:57.679
Say you publish something, that was public at a time, and it was attributed to your own data subjectivity, what would it mean to then say, well, I would like to revoke that?
00:15:57.679 --> 00:16:08.480
And that any model that had been trained on your data subjectivity or your, let's say, your knowledge or something that you produced, could be then revoked?
00:16:08.480 --> 00:16:29.519
I mean, it would take a lot of tracing! I mean, of our own images, of our own data traces, we are producing bites and bites of it a day, and then having and ensuring that you have some kind of way in which you have control over that, in which you are revoking your consent, that's adding quite a lot of technical ability about that.
00:16:29.519 --> 00:16:39.759
But imagine if you could say, "I don't want this here." You know, "this has become part of, you know, data trace part of myself that I don't want." And we can think about that in different guises, too.
00:16:39.759 --> 00:17:01.679
I mean, from a trans perspective, imagine if I could say, "I don't want any mentions of my dead name anywhere, and I just want it immediately removed." Or, "it's really difficult to see these very old pictures of myself, I don't want any of these things here." Or, that visage attributed with a gender that I no longer identify with, maybe never identified with.
00:17:01.679 --> 00:17:04.400
And so, I think there is a lot there.
00:17:04.400 --> 00:17:13.359
What would it mean to then think about those kinds of elements of technology in a future where you can trace where some kind of
aspect of you goes?
00:17:13.359 --> 00:17:21.119
And I think the kind of thing that's very difficult about it is that there's an incentive to get this data in the current regime.
00:17:21.119 --> 00:17:33.279
Companies that are building these huge AI models have a pretty huge incentive to keep on scaling them out, both making deals with people in corporations that have a lot of data.
00:17:33.279 --> 00:17:42.480
So this is why you've seen these long lists of deals that companies have had with newspapers that are sitting on these troves of data.
00:17:42.480 --> 00:18:08.079
And then also why Disney has granted OpenAI this kind of licensing, because Disney is sort of saying, "well, we've got all this intellectual property, rather than sue you and perhaps get a settlement deal 10 years from now, let's sign a billion-dollar deal right now, and you can use our IP for this sort of type period." So these companies are really data hungry.
00:18:08.079 --> 00:18:24.960
So, what would it mean to actually decouple the value of data from that kind of way in which there's a hunger around it, to there's a way in which we can determine what we're allowing our data for and whether we want to be in it at all?
00:18:24.960 --> 00:18:32.000
Data still has value, but it shouldn't have the value that it currently does in these huge AI models.
00:18:32.000 --> 00:18:39.839
And then I think we have to really be imaginative to think about what data governance schema we might need to make that a reality.
00:18:40.640 --> 00:18:51.599
So, in this other place, a realm where transparency, accountability, local governance, and ethical data stewardship are foundational.
00:18:51.599 --> 00:18:58.880
What types of AI projects have become possible as a result of these conditions?
00:18:58.880 --> 00:19:06.960
Can you take us through some examples that have emerged that genuinely serve the well-being of people and the planet?
00:19:07.519 --> 00:19:17.440
So I'm thinking about a knowledge infrastructure that would be available online with differential access to different communities.
00:19:17.440 --> 00:19:26.880
And, in this world, I can imagine different kinds of epistemologies are accepted on this informational infrastructure.
00:19:26.880 --> 00:19:35.759
And so, when I say 'epistemologies', I'm thinking about what kinds of knowledges are considered valuable, or what even counts as 'knowledge'.
00:19:35.759 --> 00:19:44.960
So I would love to know more about my own, say, lineage, or know more about who my family has been.
00:19:44.960 --> 00:19:51.119
And I wouldn't want that in our present world because that might be used against me.
00:19:51.119 --> 00:19:55.200
I mean, that might be used to limit my opportunities.
00:19:55.200 --> 00:20:10.000
Thinking about how we kind of shepherd and maintain that information and that kind of knowledge of which is often sacred to a specific group of people, I mean, that would be one element of it.
00:20:10.000 --> 00:20:13.039
Then,
how do you access that knowledge infrastructure?
00:20:13.039 --> 00:20:27.680
So, you know, our elders could access these tools without kind of, you know, having to use the interface of the keyboard or the mouse, or who could they engage with, you know, in terms of building that kind of ancestral knowledge?
00:20:27.680 --> 00:20:30.319
And here I'm riffing and thinking a little bit with Dr.
00:20:30.319 --> 00:20:37.839
Ruha Benjamin, who talks about AI as not artificial intelligence, but we have ancestral intelligence.
00:20:37.839 --> 00:20:48.640
We have these kinds of things based on what we know, or what we've had in our bones and what we've had passed down through different oral and family traditions.
00:20:48.640 --> 00:20:53.839
What would it mean to maintain that and to have different ways of interacting with that?
00:20:53.839 --> 00:20:55.680
I don't know if that looks like AI.
00:20:55.680 --> 00:20:58.240
It could look like AI, it depends on what AI means.
00:20:58.240 --> 00:21:16.799
If it means, like, machine learning models that support some kind of filtering, or pattern recognition, or speeding up kind of access, or optimising our mechanisms of data storage, I mean that could be great.
00:21:16.799 --> 00:21:30.079
If it could be just kind of sequence-to-sequence translations of voice to text, or voice to even some other modality that allows us to navigate these new informational infrastructures, then that'd be incredible too.
00:21:30.480 --> 00:21:43.440
Thinking back now to when you first arrived in this other place, what did you notice that felt different from here or maybe hinted that you weren't in our world anymore?
00:21:44.240 --> 00:21:48.480
Have you read Um 'The Dispossessed' by Ursula K Le Guinn?
00:21:48.880 --> 00:21:49.839
I sure have.
00:21:50.559 --> 00:21:52.319
I'm thinking a little bit like that.
00:21:52.319 --> 00:21:58.240
You know, the person who visits from the hyper capitalist planet to the more anarchist planet, right?
00:21:58.240 --> 00:21:59.680
I mean, I'd feel like a traveller.
00:21:59.680 --> 00:22:00.960
Like a stranger.
00:22:00.960 --> 00:22:19.119
I mean, could you imagine going from a world like this where one has to be defensive about one's data, one has to be aggressively, in some cases, protectionist: I don't want to sign up for an app, I don't want to use my credit card, I don't want to use an ATM, I don't want to set the cookies, I don't want to...
00:22:19.119 --> 00:22:21.759
And we're bombarded by all these different things.
00:22:21.759 --> 00:22:28.799
And I even think in this new world, it would look much more like that shock that the character feels in that novel.
00:22:28.799 --> 00:22:36.400
And then at some point, that kind of culture of local control pervades the technology itself.
00:22:36.400 --> 00:22:44.079
I think that kind of enmeshment would look so much different, not only in technological arrangement, but in social arrangement.
00:22:44.400 --> 00:22:49.519
I'd love to hear about what AI literacy looks like in this other place.
00:22:49.519 --> 00:22:58.240
How have people who aren't in the tech space been empowered to understand how these systems work?
00:22:59.759 --> 00:23:02.640
Let's think about what AI literacy looks like right now.
00:23:02.640 --> 00:23:23.440
Because I think what AI literacy looks like right now is that if you're using, let's say, a Large Language Model and it's producing some output, the kind of boilerplate warnings are well, you know, you have to double-check everything here because Large Language Models produce falsehoods.
00:23:23.440 --> 00:23:29.440
You have to amend this to ensure it's say, in your own voice.
00:23:29.440 --> 00:23:34.799
You have to have kind of a general understanding of what these things are doing.
00:23:34.799 --> 00:23:40.079
But I often think most people, even who are considered AI literate, do not.
00:23:40.079 --> 00:23:54.240
So I think the kind of notion of AI literacy is about confirming information or outputs, and ensuring that they comport with what you would maybe produce on your own if you had more time.
00:23:54.240 --> 00:23:58.799
Although we see this huge mismatch in how many people use LLMs.
00:23:58.799 --> 00:24:07.279
It's basically producing things that one cannot confirm, anyways, especially when students are using it, or code writers are using it, right?
00:24:07.279 --> 00:24:15.839
So I think AI literacy would look very different because I don't think it would be about the confirmation of information.
00:24:15.839 --> 00:24:21.359
AI, you know, as we write about in the book and in other places, it gets obfuscated.
00:24:21.359 --> 00:24:28.000
There's an idea that the technology is more than it is, and then it lends this kind of veneer of magic to it.
00:24:28.000 --> 00:24:31.119
And I mean, education maybe can counter some of that.
00:24:31.119 --> 00:24:35.279
I mean, I think that can be a helpful corrective in some guises.
00:24:35.279 --> 00:24:46.319
And I think it would entail maybe a more organic kind of notion of how we think about technology, you know, or how we would think about a mental model of a process.
00:24:46.319 --> 00:25:10.640
If there is um technology that maybe gets called 'AI', I mean maybe it gets called something different, and that is matching how one conceptualises the world, then that is starting much more from a place in which the understanding, I think, is that this thing is under your control.
00:25:10.640 --> 00:25:16.240
It is building mental models that you can process.
00:25:16.240 --> 00:25:30.319
It is something that I think supports this collective knowledge ownership, and it is not this, like, black box, that you have to do post hoc confirmation of the outputs.
00:25:30.319 --> 00:25:37.440
Maybe it gives you a little bit more insight in the interface, allowing a little more understanding of what's under the hood.
00:25:37.440 --> 00:25:43.599
I'm building this boat while I'm I'm
talking, so...[laughs] But I I certainly know what it would not be.
00:25:43.599 --> 00:25:50.720
Which, I think, is post hoc corrections and this really black box kind of notion of a system.
00:25:51.440 --> 00:26:01.839
As we've discussed in our present reality, companies accumulate enormous power through exclusive access to, and ownership of, data.
00:26:01.839 --> 00:26:25.759
So, in the realm you described, where this power has been redistributed and the very understanding of data has been transformed from something that can be captured and owned, into more of a commons to be collectively stewarded, how are collaborative data sets being used to address global challenges?
00:26:26.640 --> 00:26:28.000
It's a very different world.
00:26:28.000 --> 00:26:31.680
I'm sort of presupposing a huge societal difference.
00:26:31.680 --> 00:26:38.480
Issues around global poverty, around people who are unhoused, climate change...
00:26:38.480 --> 00:26:45.519
I mean, these are the kinds of things which in our current society are the most intractable problems.
00:26:45.519 --> 00:26:57.440
In a society that's more governed from the bottom up, I mean, I think those become less of a tech problem and more of a governance problem that have been addressed by that kind of structure.
00:26:57.440 --> 00:27:02.079
The thing is, like, so many of the big problems we have right now are because capitalism [laughs].
00:27:02.079 --> 00:27:12.559
And so, you know, and so thinking about things that are that are not human-made or societal or structurally determined are the kinds of things that I'm thinking about.
00:27:12.559 --> 00:27:14.480
So health, I think, is one.
00:27:14.480 --> 00:27:30.240
Joint efforts in creating therapies for cancer, or therapies for viruses; and then how we grow food, making those kinds of elements of food production sustainable in local environments.
00:27:30.240 --> 00:27:40.799
What would it mean to think about local communities that don't have to be displaced because of sort of depletion of resources in a certain place?
00:27:40.799 --> 00:27:51.119
What would it mean to not rely on a huge kind of, you know, industrial scale agriculture that really forces you to not use so many parts of the land?
00:27:51.119 --> 00:28:02.640
What would it mean to think about engineering agriculturally so that a certain community could uh stay on a land longer and not have to be as migratory?
00:28:02.640 --> 00:28:15.119
And, I mean, I think a lot of that has to do both with I think the ancestral knowledge, but also maybe this pattern matching that we can do more at scale with kind of a locally controlled model.
00:28:15.759 --> 00:28:31.119
Looking squarely at our present reality for a moment, what do you believe is the most urgent area of focus, particularly for those listening who are really concerned about the societal impact of AI-driven technologies?
00:28:31.599 --> 00:28:33.680
Yeah, I mean, centralisation, right?
00:28:33.680 --> 00:28:36.319
I mean, centralisation is the big issue.
00:28:36.319 --> 00:28:50.160
And the things that terrify me the most and keep me up at night are the fact that we have four companies that are effectively directing the trajectory of the world economy.
00:28:50.160 --> 00:28:53.839
And how do we make interventions with those around us?
00:28:53.839 --> 00:29:13.519
So thinking again locally, and then thinking about how we can connect it to this large common enemy and imagine what it would mean to think about alternatives that are, you know, locally based and could also seed what happens after.
00:29:13.920 --> 00:29:30.319
I'd love to talk to you now about potential divergence points that may have helped shape this other world and effectively reseed the story of technology and of course, in turn, reshape the story of AI.
00:29:30.319 --> 00:29:46.799
Uh, I I imagine it'd be safe to assume that in this other place, instead of its origin story being in the hands of the post-World War II moment where military funding was dominating early computing, it took a different path.
00:29:46.799 --> 00:29:48.559
[laughs] Thanks for sticking with this.
00:29:48.559 --> 00:29:55.119
I know divergence points are tricky and it can be really challenging to abandon our paradigm...
00:29:55.599 --> 00:30:02.559
It's challenging, but it's a good exercise too! And it's challenging because I don't think we flex that muscle a lot.
00:30:02.559 --> 00:30:10.240
You know, we often don't get a chance, and I think it's really important to get that chance, and I think that's part of what we need to do.
00:30:10.240 --> 00:30:12.880
And and there's many folks, I mentioned Dr.
00:30:12.880 --> 00:30:33.440
Ruha Benjamin, you know, in her book 'Viral Justice and Imagination', and it's sort of thinking about what does it mean to imagine these things, not only as imagining redresses for societal harms, which I think are very critical, but also to think about what would it mean to have a whole new world altogether?
00:30:33.440 --> 00:30:41.519
Maybe in the post-World War II moment, we were all sufficiently horrified and suggested that we needed to do something else, right?
00:30:41.519 --> 00:30:53.599
I mean, I'm reading a lot of Octavia Butler lately, and I've I'm coming to the close of her Patternist Series, and she's got in that book, you know, kind of almost multiple dystopias...
00:30:53.599 --> 00:31:11.920
And, and you know, the kind of ever-present lesson of Octavia Butler is that, you know, no matter what kind of technology, whether it is physiological, or genetic, or pathological, humans still have this very strong disposition for hierarchy.
00:31:11.920 --> 00:31:12.720
And that's...
00:31:12.720 --> 00:31:16.160
that's a main Butlerian lesson [laughs].
00:31:16.160 --> 00:31:23.119
Thinking of that in mind, what would it mean to construct social structures that can help mitigate that?
00:31:23.119 --> 00:31:28.319
And then from that, the technology flows that also reinforces that.
00:31:30.480 --> 00:31:32.640
And that was Many Worlds Traveller, Dr.
00:31:32.640 --> 00:31:33.519
Alex Hanna.
00:31:33.519 --> 00:31:42.079
In this reality, among many other things, Alex is director of research at DAIR, the Distributed AI Research Institute.
00:31:42.079 --> 00:31:52.720
Her work centres on the data used in new, computational technologies and the ways in which these data exacerbate racial, gender, and class inequality.
00:31:52.720 --> 00:31:53.200
Dr.
00:31:53.200 --> 00:32:00.240
Hanna is also the co-author of the book 'The AI Con', available through Harper; and co-creator with Emily M.
00:32:00.240 --> 00:32:04.880
Bender of the podcast, 'Mystery AI Hype Theatre 3000'.
00:32:04.880 --> 00:32:09.279
The artist Warren Armstrong has responded to this parallel world.
00:32:09.279 --> 00:32:15.119
To check out what he has to say and to explore all of our episodes, visit and subscribe to theparalleleffect.substack.com.
00:32:17.519 --> 00:32:28.319
You've been listening to The Parallel Effect Podcast, supported by people like you and assisted by the Australian government through Creative Australia, its principal arts investment and advisory body.
00:32:28.319 --> 00:32:30.559
Thanks so much for joining me, I'm Daz Chandler.
00:32:30.559 --> 00:32:39.200
Until next time, as you tend to the technologies you touch, may you find innovation guided by community in service of all life.