Frameworks ⊗ Think better ⊗ Three reasons we’re in an AI bubble

Frameworks

“Knowledge is often a matter of discovery. But when the nature of an enquiry itself is at question, it is an act of creation.” Céline Henne’s article discusses the importance of understanding different types of enquiries, particularly distinguishing between framed (using a framework, whether you know you are or not) and framing enquiries. She argues that facts often do not change minds because it is the underlying framework that shapes our analysis of those facts and two people in a discussion or a speaker to a crowd aren’t necessarily using the same framework.

Henne also highlights that frameworks are often transparent, limiting our awareness of their influence on our thought processes, thus complicating our ability to think critically about our enquiries. Ultimately, she suggests that knowledge is not just about representation of facts but involves an iterative process of articulation and construction that reflects our human needs and interests.

The first part is somewhat academic but the conclusion makes the whole piece useful, conceptually connecting to political discourse, debates, fading consensus reality, and different languages/angles through which people interact with fields of knowledge.

Once the framework is settled, it delimits the questions we can ask and the range of their possible answers, although the correct answer itself is not up to us. […]

The issue is not just what to think about things, but how to think about them. Such enquiries are concerned with the creation, revision or expansion of our frameworks. […]

Frameworks are crystallisations of our understanding of the world, and they remain transparent most of the time. We see through them instead of looking at them.

Think better

Mike Loukides urges readers to take responsibility for their own thinking rather than relying on AI or computers as oracles. “Don’t listen to oracles, AI or otherwise. Take responsibility for thinking.” While generative AI can aid in generating ideas and summarising information, it should not replace human creativity and independent thought. Loukides argues that our ability to innovate and synthesise new ideas requires a deep understanding of the technologies we use, rather than simply accepting AI-generated outputs.

People who just repeat what generative AI tells them, without understanding the answer, without thinking through the answer and making it their own, aren’t doing anything an AI can’t do. They are replaceable. They will lose their jobs to someone who can bring insights that go beyond what an AI can do. […]

Viewed from this angle, generative AI is just another abstraction layer, another layer that generates distance between the programmer, the machines they program, and the problems they solve.

Three reasons we’re in an AI bubble (and four reasons we’re not)

Casey Newton with a good piece on whether this is an AI bubble or not (and what kind of bubble). I find the microwave oven example intriguing. “Invented in 1947, and by 1971 were only in 1 percent of American homes. They didn’t reach 90 percent of American homes until 1997.“ As Newton mentions, ChatGPT set a record for “fastest growing consumer application ever,” but did it?

How many people would have tested a microwave if it could be teleported in their home to try a meatloaf a couple of times? “Tastes all wrong” would have been the result and they’d teleport it back. ChatGPT was the fastest growing tested platform ever but a huge percentage of people never went back to it. AI is a rare combination of being seen as software but required massive investment, like computer hardware and anything require an assembly line. That new combination of development, ‘testing,’ and adoption curves might mean we need to evaluate it differently in terms.

It’s possible that AI applications are “never going to be cost-efficient, are never going to actually work right, will take up too much energy, or will prove to be untrustworthy,” the hedge fund wrote. […]

Whether GPT-5 and its peers can deliver a step-change in functionality from their predecessors is a real and important question. But focusing on that too much can obscure just how much innovation is left to be wrung out of the models we have today. […]

On the other hand, startups have a much harder road. The natural advantages they normally have over giants, such as being small and nimble, can’t make up for the high cost of training models and offering AI-powered services.

Turning brownfields to blooming meadows, with the help of fungi

I’ve featured similar articles in the past, but I remain fascinated by the idea of bioremediation with fungi. In this case Danielle Stevenson uses a mix of plants (for heavy metals) and fungi (for carbon-based pollutants) to resuscitate brownfields—”industrial and commercial properties polluted with hazardous substances.” She has also worked with a bike shop for their lubricant-soaked rags, and tribal communities to learn how to use these methods to remediate toxic sites on their own.

if there isn’t any cellulose, and instead there’s diesel or something that’s essentially made out of carbon, they will find a way to eat it, applying their enzymes the same way they would if it was cellulose. […]

In three months we saw a more than 50 percent reduction in all [petrochemical] pollutants. And then by the 12-month period, they were pretty much not detectable. […]

What I have been told is that a lot of times [these communities] haven’t felt listened to. And then I come in with the science that basically backs up the knowledge that the people already have — affirming the power of nature to restore itself. I’m saying the same thing they are, except I’m using the scientific language that regulatory agencies and other groups are more likely to listen to.

§ Trajectories for future festivals. “This article delves into futures for festivals, exploring their potential to address societal challenges, foster resilience, and create sustainable futures through culture and creativity. It’s based on the insight that, as spaces where society reinvents itself, festivals are more than just events; they are dynamic laboratories of change, places where new forms of culture and citizenship are explored and rehearsed.”

The Future Festival Summit is part of MUTEK Forum, held this coming week. It’s an always fascinating event here in Montréal. An “international gathering and ideas market on digital culture, art, technology, and society.” It’s very late, but Sentiers readers still get 25% off. Just head to the box office and use promo code SENTIERS25. Be sure to also hit reply and tell me if you’re going to be in town, we should chat!

Futures, Fictions & Fabulations

Fiction+Facts=Futures
“Metamodernism allows me to see Design Fiction as more than just speculative imagination or a critique of the present. It’s a methodology that bridges the gap between what is, what could be, and what should become. Design Fiction thrives in the oscillation between the possible and the actual, between what is not true yet and what might be possible.”

Singing better worlds to life
Stuart Candy’s notes on a collaboration with Brian Eno, using music to grow public imagination. “The focus is on harnessing the promise and pleasures of worldbuilding on one hand, and musical co-creation on the other, to help people not just think or imagine but play and perform their way into future possibilities.”

Tides of tomorrow
“There’s no mistaking Tides of Tomorrow for anything other than anxious “cli-fi”, but its tone is exuberant, brash and irreverent rather than moody or dread-laden. The setting is the fictional planet of Elynd which, says lead game designer Adrien Poncet, lets him and his colleagues take liberties with the science and technology they are depicting.”

Algorithms, Automations & Augmentations

Transformer explainer
Can’t say I’ve had time to properly dig through and understand everything in there, but great explainer anyway, with superb visuals and a step by step overview of how LLM transformers work.

Under the radar?
“Global policy proposals for ensuring the safety of advanced artificial intelligence (AI) systems have centred on foundation model evaluations as an important method to identify and mitigate the risks these systems pose. The core goals of foundation model evaluations are to understand the foundation model and / or its impacts, including the model’s capabilities, risks, performance, behaviour and social impact.”

Google claims math breakthrough with proof-solving AI models
“DeepMind announced that AI systems called AlphaProof and AlphaGeometry 2 reportedly solved four out of six problems from this year's International Mathematical Olympiad (IMO), achieving a score equivalent to a silver medal. The tech giant claims this marks the first time an AI has reached this level of performance in the prestigious math competition—but as usual in AI, the claims aren't as clear-cut as they seem.”

Asides

• 🤯 📱 Sorting through old bookmarks I happened on my oldest ‘social bookmark.’ Made on August 18th 2004!! Considering rampant link rot, I’m impressed the article is still online. Little-known startup was behind iPod’s easy-to-use interface / Firm’s founder now working on the latest handhelds.
• 🐧 📚 😍 Note to readers, I’ll always post any gallery of Penguin book covers. In this case Greg Neville’s. The art of Penguin book covers.
• 😭 🎥 ✍🏼 Benedict Cumberbatch reads a letter of apology. “In 2021, Stuart Capstick, deputy director at the Centre for Climate Change & Social Transformations in Cardiff, wrote a letter to his children.”
• 😱 🥵 🌏 🌡️ Doom scrolling a beautiful NYT visual essay. How Close Are the Planet’s Climate Tipping Points? “Earth’s warming could trigger sweeping changes in the natural world that would be hard, if not impossible, to reverse.”
• 🤔 📉 🇰🇷 South Korea to establish birth rate ministry amid population decline. “Facing the lowest birth rate in the world, South Korea's population is projected to shrink by nearly 20 million by 2070. The government launched a new ministry dedicated to addressing the national crisis.”
• 🗺️ 🇮🇷 🤩 The Oldest World Map in the World. “Irving Finkel, a curator at the British Museum and an expert in cuneiform, takes a look at a 2900-year-old Mesopotamian tablet that contains a map of the world as it was known at the time.” As Jason says, “Finkel could not possibly look more like a British Museum curator than he does.”
• 🏢 🧱 Why architects appreciate brutalism, even if you don’t. “Gang says there is an environmental incentive not to demolish brutalist buildings. The process of curing cement into concrete causes a lot of carbon emissions. Generally, people are now trying to use more environmentally friendly building materials. But the brutalist buildings have already been built, and the carbon has already been released into the atmosphere.”