Posts

If LLMs Are a Dead End, Could They Still Accelerate AGI?

By -
Image
François Chollet has made one of the sharper criticisms of the current artificial-intelligence boom. In a 2024 interview with Dwarkesh Patel , the creator of Keras and ARC-AGI, and co-founder of the ARC Prize, argued that OpenAI had effectively set progress toward artificial general intelligence back by five to ten years. Chollet gave two reasons. First, he argued that frontier AI research had become much less open, with leading laboratories publishing fewer of the technical details that previously allowed ideas to spread across the research community. Second, he believed that the success of large language models had caused them to “suck the oxygen out of the room”: money, compute and researchers were flowing toward an approach he regarded as an off-ramp rather than a route to AGI. The first criticism is mainly institutional. The second creates a stranger counterfactual. Suppose Chollet is right that scaling transformers never produces artificial general intelligence. It still does...
Post a Comment
Read more

Who Gets to Know What? Narrative Point of View and the Control of Information

By -
Image
Editorial image generated by author. First person, second person and third person can look like surprisingly small grammatical choices. I opened the door. You opened the door. She opened the door. The action has not changed, and neither has the door. Yet change the narrative point of view and almost everything surrounding that action may change with it: what the reader knows, whose interpretation accompanies what they see, which information can plausibly be withheld and even what kind of surprise the story can produce. Narrative perspective is less like choosing a camera angle than designing an information system. A mystery changes if the narrator knows the solution. A tragedy changes if the reader knows something the protagonist does not. A character study changes when we hear someone's rationalisations from inside their own head. A plot twist may depend on the narrator being unable to tell us something, choosing not to tell us, or telling us everything necessa...
Post a Comment
Read more

Everything Is Becoming an Electricity Problem

By -
Image
Why climate policy, artificial intelligence, industrial strategy and energy security are increasingly converging on the same physical system. An electric car looks like a transport technology. A heat pump looks like a heating technology. A data centre belongs to computing. An electric arc furnace belongs to heavy industry. An electrolyser producing hydrogen seems to belong to yet another technological system. Increasingly, however, they are all asking the same question of the economy: can you supply enough electricity? Not merely enough over the course of a year. Enough in the right place, at the right time, at a price that makes the activity viable, through a grid capable of delivering it, without making the wider system unreliable. This is one reason debates about climate, artificial intelligence, industrial policy and national security are becoming harder to separate from debates about energy. Electricity is moving from being one important energy carrier among several toward...
Post a Comment
Read more

What Is Actually Happening in The Quantum Thief? Jean le Flambeur, the Oubliette and the Ending Explained

By -
Image
Editorial image generated by the author. Hannu Rajaniemi's posthuman science-fiction novel The Quantum Thief has a peculiar way of making an intelligent reader feel as though they have accidentally skipped fifty pages. Gogols appear before anyone explains what a gogol is. People exchange memories instead of messages. Privacy is something you negotiate almost physically. Citizens of Mars carry Watches that measure how long they can remain in ordinary human bodies. Meanwhile a detective investigates crimes in a society where even memory is governed by permissions, while the protagonist tries to recover information that an earlier version of himself deliberately hid. Rajaniemi rarely stops to explain any of this, which makes it tempting to assume that The Quantum Thief has an extraordinarily complicated plot. It does not, really. The simplest version is this: Jean le Flambeur is an incomplete version of a legendary posthuman thief. Mieli breaks him out of a virtua...
Post a Comment
Read more

Meditations by Marcus Aurelius Review: The Emperor Who Had to Remind Himself How to Live

By -
Image
Marcus Aurelius ruled one of the most powerful empires in history, commanded armies and could make decisions affecting millions of people. Yet in Meditations he can be found arguing with himself about whether he really has to get out of bed in the morning. In Book 5.1, he acknowledges the attraction of staying warm under the covers before reminding himself that he has work to do. He even appeals to the example of ants, birds and bees getting on with their tasks. I found that strangely endearing. We tend to imagine a philosopher emperor as someone who has already achieved a level of wisdom and self-discipline beyond the rest of us. Instead, much of Meditations consists of reminders not to be lazy, angry, vain, distracted or unnecessarily troubled by other people. Marcus clearly understood how he ought to behave. Putting that knowledge into practice was another matter. Having previously read William B. Irvine’s A Guide to the Good Life , and with a general interest in Stoicism, I ...
Post a Comment
Read more