Newsletter No.256 — Mar 19, 2023

Jungle Snooker and AI’s High Weirdness (Webb, Klein, Doctorow, and Evans on AI) ⊗ Apocalyptic Infrastructures ⊗ There’s Nothing Unnatural About a Computer

Want to understand the world & imagine better futures?

Also this week → Over reliance as a service ⊗ Rewatch the GPT-4 developer livestream ⊗ Kottke.org is 25 years old ⊗ Future Today Institute’s 2023 Tech Trends Report


Jungle snooker and AI’s high weirdness

Last week I wrote about AI that “it’s still a fascinating topic but I feel I’m going to have to step away a bit, at least for this newsletter, it’s getting a bit redundant” and then GPT-4 arrived. I didn’t pay too much attention to the release over the week, but it was ‘ambiant,’ shall we say. I did bookmark a few things and save a good number of articles in Reader, which I sat down to read Friday morning. I started with the surprising ease and effectiveness of AI in a loop by Matt Webb (my highlights), then this changes everything by Ezra Klein (my highlights), then the AI hype bubble is the new crypto hype bubble by Cory Doctorow (my highlights) (pre GPT-4, to be fair), and finally language is our latent space by Jon Evans (my highlights). I was basically trying to triangulate a decent position/anticipation or where AI might be going and how quickly. (Sidenote: yes it’s all dudes, I try to balance issues and what I read in general, chips fell this way this week.)

If you’ve been reading me for a while, you might have noticed that some names come back regularly, the four above being among them. I don’t have an ordered list of who’s opinion I trust more, but I do have loose groupings of who’s thinking I respect. When people in the same grouping disagree, it’s both a good sign that I’m not toooo much in any echo chamber (even though those are mostly debunked, it’s still a useful image), and harder to make sense of.

In the articles above, Webb used ‘singularity’ in his url and quotes Vinge to end the post. He’s super impressed with GPT-4, LangChain, and ReAct. The rate of amelioration between versions, combined with the tools around them, makes him enthusiastic and optimistic about what’s coming. Klein seems to be roughly in the same group in terms of the scale and speed of what’s coming, while being very worried about the potential dangers and the slim chances that ‘we’ come up with appropriate policies in time. Doctorow basically doesn’t believe any of it, calls bs and focuses more on who’s creating these AIs and their usual biases and exaggerations, not a bad angle. Part of his argument is based on Timnit Gebru and Emily M Bender’s paper and thinking, including the interview in You are not a parrot (which I’m a bit ashamed to say I haven’t read yet).

As Bender says, we’ve made “machines that can mindlessly generate text, but we haven’t learned how to stop imagining the mind behind it.”

Evans (the more technically adept of the four writers) uses this great image:

For the first time in a very long time, the tech industry has discovered an entirely new and unknown land mass, terra incognito we call “AI.” It may be a curious but ultimately fairly barren island ... or a continent larger than our own. We don't know.

You should have a read for his various metaphors of what LLMs actually do, which go a long way in understanding and setting expectations. The ‘money quote’ or most important takeaway from his piece:

The unreasonable effectiveness of LLMs, says me, stems from the fact that we have already implicitly encoded a great deal of the world’s complexity, and our understanding of it, into our language — and LLMs are piggybacking on the dense Kolmogorov complexity of that implicit knowledge.

I just listed the articles in the order I read them, a mix of serendipity and habits. It’s largely blind luck that I read Evans last because it’s the best conclusion to the quartet.

If you attach the four together; AI shows glimpses (sometimes more than glimpses) of great usefulness and massive disruption for certain tasks and jobs. This potential is maturing extremely fast and policy will likely not keep up. Still, be weary of the bullshit thrown left and right, who it’s coming from and their background. Perhaps LLMs work so well because of all the meaning we have embedded in our language, which might mean that it is one piece of a group of technologies to get (if ever) to AGI, not the technology in its infancy. ◼

Apocalyptic infrastructures

Infrastructures have made a number of appearances in this newsletter, often emphasising the need for maintenance, usually with some awe and the goal of showing their importance. You could say that in this piece at Noema Laleh Khalili wraps all of that up with historical context, especially on the use of infrastructure in colonial power and as post-colonial tools, but also in placing infrastructures and their importance within the contemporary issues and opportunities around the climate crisis, degrowth, and inequality. ‘We’ need a new way of framing these projects, of thinking about who is included in decisions, and in the ultimate goals of these infrastructures. Great read.

Across political divides, all infrastructures share one common feature: their detrimental environmental effects. Dams destroy riverine ecosystems and leach the soil. Cement factories and coal-powered electricity spew out pollution across the globe. Sewer lines pour into sensitive riparian and coastal biospheres. Oil fields and pipelines contaminate vast swathes of land, leaking into fragile water tables. Data centers produce carbon dioxide and heat on a monumental scale. […]

The dilemma is how to provide a livable life and livelihood, health, education, basic utilities, clean air and clean water without hitching them to the zero-sum game of growth. Degrowth would entail slowing down fossil fuel consumption, arresting the constant drive toward the financialization of every aspect of life, and contracting the processes that produce waste. It asks of us all to consume less and more thoughtfully. […]

Infrastructures that would emerge out of an ideology of degrowth would incorporate a more redistributive, participatory and egalitarian ethos. And a strategy of degrowth would include ecological wellbeing as an immutable principle in all planning and use. […]

For infrastructure to work, for it to serve the public and steward the world’s air, water and soil for future generations, it has to be planned through more open, egalitarian and environmentally militant processes.

There’s nothing unnatural about a computer

At Grow, Claire L. Evans interviewing James Bridle about their book Ways of being to “take a fresh look at nature’s intelligence.” As with most Q&A formats, I struggle a bit to summarise so just have a read, especially as an introduction or further thinking on the ideas of more-than-human intelligence and various forms of intelligence. Especially relevant in this AI-heavy issue, the first highlighted quote below is especially noteworthy and is something I’ve mentioned a number of times; how AI can be interpreted, thought of, and perhaps guided through new lessons gleaned from animal and plant intelligences?

[G]ardening has allowed me to develop a more active awareness of the fact that we live in a more-than-human world. It’s a really powerful kind of awareness to have, and it’s precisely the kind of awareness that I would want to have about technology. Is making things the technological equivalent of gardening with code? […]

I have this very strong sense that one of the broader roles of AI in the present is really just to broaden our idea of intelligence. The very existence, even the idea of artificial intelligence, is a doorway to acknowledging multiple forms of intelligence and infinite kinds of intelligence, and therefore a really quite radical decentering of the human, which has always accompanied our ideas about AI — but mostly incredibly fearfully. […]

That’s why I’ve always had the fascination with all the glitches and weird edge cases and strangeness of AI. Because some of that is reflecting back whatever trash these things have been fed on. But it’s also genuinely presenting radically new ways of seeing the world that expands our view in ways that we don’t yet fully understand. […]

The reason this is urgent and a fight is because we are literally losing knowledge. Through habitat destruction, through climate change, through loss of biodiversity — it is knowledge that is being lost. But that process is hardly new. It’s been going on for centuries, if not millennia. It’s the main action of colonialism and imperialism.

Kottke.org is 25 years old today and I’m going to write about it

Happy anniversary and congrats to Jason on an incredible (and ongoing) ride through a good chunk of the history of the internet and much of the history of weblogs. I’ve been following along since Osil8 and it was an honour to guest edit a few times. I was definitely nodding my head reading the quote below, thinking of my own blogging and newslettering experience.

I had a personal realization recently: kottke.org isn’t so much a thing I’m making but a process I’m going through. A journey. A journey towards knowledge, discovery, empathy, connection, and a better way of seeing the world. Along the way, I’ve found myself and all of you. I feel so so so lucky to have had this opportunity.

Asides

  • Apply to join the upcoming cohort of Nervous System Mastery — learn evidence-backed protocols for cultivating calm, increase your capacity for focus and improve sleep. Limited seats available. Sponsored
  • 🌌 🔭 🤩 🇺🇸 NASA’s Webb Telescope Captures Rarely Seen Prelude to Supernova. “The rare sight of a Wolf-Rayet star – among the most luminous, most massive, and most briefly detectable stars known – was one of the first observations made by NASA’s James Webb Space Telescope in June 2022. Webb shows the star, WR 124, in unprecedented detail with its powerful infrared instruments. The star is 15,000 light-years away in the constellation Sagitta.”
  • 📖 🤖 🤔 A monthly magazine to showcase the diverse creativity of the Midjourney community. “Each magazine features a selection of artwork curated from the 10,000 most highly rated images, as well as interviews with Midjourney community members.”
  • 🏢 🤔 🇺🇸 So You Want to Turn an Office Building Into a Home? Long piece at The New York Times with some animations. “Cities are eager to do this amid rising remote work. But it’s harder than you might think.”
  • 🗺 ⚔️ 😍 How to Draw Fantasy World Maps. “JP Coovert shows people how to draw maps for fantasy games, books, and other media.”
  • 💩 😖 Hey Cortana, what’s the worse move we can make? Microsoft lays off AI ethics and society team. “The move leaves Microsoft without a dedicated team to ensure its AI principles are closely tied to product design at a time when the company is leading the charge to make AI tools available to the mainstream, current and former employees said.”
  • 🌱 🌿 🤔 🇬🇧 High-tech, soil-less, and brimming with life: the vertical farms of tomorrow - Positive News. “White walls, spotlessly clean floors, bright lights and rows and rows of hardboard and metal frames, out of which spring tiny pots of plants – spinach, basil, all manner of leafy greens, herbs and spices. Those splashes of greenery aside, it looks more like a tech hub than a farm.”
  • ⛪️ 😃 🇫🇷 Notre-Dame Repair Reveals Another Historic First: 800-Year-Old Iron Reinforcements. “Scientists working on the scorched interior of Notre-Dame de Paris have found iron was used in the cathedral’s construction in the mid-12th century. It’s an unexpected discovery that changes how researchers thought the church was built, and provides surprising insights on the iron trade in 12th-century Paris.”