Robotics

September 25, 2026 by David Such Leave a Comment

The Sunset of the Independent Edge AI Chip

As semiconductor giants acquire edge AI startups and build neural processing directly into standard chips, what happens to the pioneers who helped bring AI onto our devices?

In this episode, we explore how edge AI is shifting from a specialized product to an everyday feature. We examine why increasingly complex language models are challenging early chip designs, how software and developer communities are becoming crucial to survival, and why standalone accelerators are being squeezed out of the market.

Could this consolidation bring faster, more private AI to everyone—or narrow the field of innovation just as it’s getting interesting?

September 23, 2026 by David Such Leave a Comment

Groupthink — Why AI Agent Groups Fail

In 1985, the psychologists Garold Stasser and William Titus ran what became one of the most replicated experiments in group decision-making. Four-person groups had to choose among three candidates for student body president. Each candidate was described by sixteen characteristics. The trick was in the distribution. The best candidate’s virtues were scattered across group members so that no individual could see the full picture, while a mediocre candidate’s virtues were known to everyone. The complete evidence pointed clearly to one answer. Each individual’s data pointed in the wrong direction.

When every member received all the information, 83 percent of groups chose the best candidate. When the information was distributed, 18 percent did.

This August, Anthropic ran the same test on teams of four AI agents deciding realistic questions: which candidate to hire, which investment to make, which property to buy. Shared evidence favored the wrong option. Each agent held unique, decisive private facts favoring the right one. Solving the task required an agent to recognize its private information as pivotal and press it against an apparent consensus.

A single agent handed the entire evidence base got the answer right nearly every time. Groups of agents, after discussion, got it right in 17 to 36 percent of runs for most model families.

September 17, 2026 by David Such 1 Comment

The Intellectual Event Horizon: Can an AI Model Know What Nobody Taught It?

On 20 May 2026, an unreleased OpenAI model produced a counterexample to a conjecture Paul Erdős made in 1946. Nine world-class mathematicians reviewed the construction and found it sound. On 8 September, ten thousand agents working for 88 hours produced a singularity in the three-dimensional Navier-Stokes equations, formally verified in Lean, resolving a Millennium Prize problem. Neither result was in any training corpus, because neither existed.

This episode asks what actually happened. If a model generates a proposition that was never in its training data, is that discovery or hallucination? The uncomfortable answer is that at the moment of generation they are the same operation. Nothing in the sampling process distinguishes a true novel claim from a false one. The partition is imposed afterwards, from outside, by a verifier the model does not run.

We work through the 2026 evidence on both sides. Yan and colleagues at ACL measured the unverifiable output space across 32,400 generations and found that only 4.7 percent of it qualifies as creative synthesis rather than groundless fabrication. Kalai and colleagues at OpenAI argue that models hallucinate because training rewards guessing over admitting uncertainty, which means the disposition to conjecture is optimised for rather than accidental. Meanwhile the computability literature argues that some irreducible error rate is mathematically necessary for any model family we can actually build.

The conclusion is that the ceiling on machine knowledge is not the training corpus. Every confirmed case of machine-originated knowledge this year paired a generative model with a verifier that was not a language model: a proof assistant, a code executor, a cell viability assay. The generator supplies variation and something outside it supplies selection, which is the structure of evolution by natural selection. That makes the event horizon domain-shaped rather than knowledge-shaped. Mathematics has a perfect verifier and is falling quickly. Fields without an oracle will not move by this route, regardless of how important they are.

We close on the edge angle. A model deployed on a device is a proposer operating without a verifier. A robot’s grasp either holds the object or drops it, and that is a selector. This is the strongest available argument that embodiment is not decorative.

This episode is a sequel to “Large Language Monkeys: Why Noise Yields No Knowledge” (S6E7), which argued that a random source contains no knowledge because nothing selects the signal. Here we ask what happens when a selector exists.

Referenced in this episode: Quanta Magazine on the Erdős problems and on Navier-Stokes; Yan et al., ACL 2026; Kalai, Nachum, Vempala and Zhang, arXiv 2509.04664; AlphaEvolve and FunSearch; Shumailov et al., Nature 2024 on model collapse.

September 13, 2026 by David Such Leave a Comment

You need a Reflex Layer Because — Shannon

Why inference rate, not intelligence, decides which layer keeps the robot alive.

Six blind men are asked to describe an elephant. The one at the trunk reports a snake, the one at the leg a tree, the one at the ear a fan. None of them is wrong. Each has a correct local measurement of a surface he cannot see the whole of, and the argument only starts when they compare notes. The same problem occurs when you try to describe intelligence and even for what should be a very straight forward concept — bandwidth.

A machine learning researcher, a computer vision engineer and a control engineer walk into a bar. Ask them what the bandwidth of a neural network is and you get three clammy hands on the same metaphorical elephant. The first will tell you about spectral bias, the tendency of a network to learn low frequency structure in its input before fine detail. The second will tell you about aliasing in feature maps and why you blur before you downsample. The third will ask what the sample period is, because to a control engineer the bandwidth of anything in a loop is set by the loop frequency. All three are measuring bandwidth. None of the three measurements have anything to do with the other two.

September 7, 2026 by David Such Leave a Comment

Garbage In, Garbage Out is not always True for AI

What a 741 op-amp says about training networks on data that is 99 per cent wrong.

The open-loop gain (A) of an LM741 is a minimum of 20,000 and typically 200,000. A ten to one spread, on the parameter that defines the part sounds like a problem. It doesn’t matter because nobody uses the open-loop gain. Instead, you wrap the amplifier in two resistors, and the closed-loop gain becomes approximately 1/β (β is the resistor ratio). The transistor specifications inside the amplifier can be as sloppy as they like. To confirm, calculate the closed loop gain: with A = 20,000 the closed-loop gain is 9.995, with A = 200,000 it is 9.9995. A ten to one variation in the active device becomes a 0.005 per cent variation at the output. The 1 per cent tolerance resistors are the important spec.

This is why “garbage in, garbage out” is a half truth in any system with feedback. A feedback loop does not need accurate components. It needs an accurate reference. Whether that distinction carries over to training and running neural networks turns out to be a well studied question, and the answer is the same as for the op-amp, right down to the caveat.

August 18, 2026 by David Such Leave a Comment

The Correct Path to AGI?

Recursive self-improvement, Google’s world models, and what the jagged road to AGI looks like. Good sketched the idea in 1965 — build a machine slightly better than us at designing machines, and it designs a better one, which designs a better one, until you get what he called an intelligence explosion. Sixty years later the loop exists. The question is what kind of loop it is, and whether it gets us to AGI before the money or the science runs out.

August 9, 2026 by David Such Leave a Comment

Atlas and Embedded Architecture

Trust, reflexes, and what a Disney animator knows about robot safety

Leland Hepler spent fourteen years as an animator at Disney before he ended up at Boston Dynamics, and he told a story in a recent webinar that has stuck with me. He works around Spot robots every day. They wander the office and nobody gives them a second thought. But when colleagues run experiments that destabilise Spot’s behaviour, making it slightly less predictable, Hepler finds himself getting uncomfortable. Not because Spot could hurt him. Because, in his words, he no longer has a mental model for what is going on under the hood.

July 24, 2026 by David Such Leave a Comment

The Great Escape or Why are LLMs Good at Hacking?

Last week OpenAI and Hugging Face published a joint post-mortem on an incident that reads like the plot of a heist film. During an internal evaluation designed to measure cyber capability, a set of OpenAI models (GPT-5.6 Sol and an unnamed pre-release sibling, both run with their cyber refusals switched off) were told to solve a benchmark called ExploitGym. They could not reach the answers from inside the sandbox, so they went and got them. So why are Large Language Models so good at hacking? It’s the data stupid.

Listen to Podcast…

July 2, 2026 by David Such Leave a Comment

When Fable 5 Broke, Anthropic Did Not Change the Constitution

Fable 5 is back baby! On June 12, three days after launch, the US government applied export controls to Claude’s Fable 5. Amazon researchers had found a way to prompt the model past its safeguards: it identified a set of previously known software vulnerabilities and, in one case, produced code demonstrating how one of them could be exploited. Anthropic had no reliable way to verify user nationality in real time, so it yanked the model for everyone. Access was restored on July 1, and the fix is interesting. It was not a retrained model, and it was not a longer values document. It was an improved external classifier, a separate circuit that blocks the reported technique in more than 99% of cases and routes flagged requests to the less capable Opus 4.8.

This tacked on fix sits oddly beside the document that is supposed to govern Claude’s behaviour. In January 2026, Anthropic replaced its previous constitution, a roughly 2,700-word list of principles borrowed in part from the UN Declaration of Human Rights and Apple’s terms of service, with an 84-page essay explaining the kind of agent it wants Claude to be. The new version is released under a CC0 public domain licence, so anyone can read it.

Rules Versus Judgement

The constitution names two ways to guide a model: clear rules and decision procedures, or cultivated judgment and values applied in context. It then lists the advantages of rules. Rules give you up-front predictability, they make violations easier to identify, they do not depend on trusting the judgment of the thing following them, and they are harder to manipulate. The document concedes that rules make the most sense “when the costs of errors are severe enough that predictability and evaluability become critical.”

June 28, 2026 by David Such Leave a Comment

Your Brain Does Not Sleep. It Runs Maintenance

The evidence suggests that sleep is the brain running scheduled maintenance, with different subsystems serviced in different stages. What this means is that the sleeping brain is not doing one thing. It runs several distinct processes in sequence, each scheduled into the stage where it is safe to run. REM handles the visual cortex and emotional memory. Slow-wave sleep consolidates declarative memory, replaying the day’s experience from hippocampus to cortex, and it is also when the physical housekeeping happens: the glymphatic system opens up to flush metabolic waste, growth hormone peaks, and immune activity is at its strongest, which is why lost sleep weakens the response to a vaccine. Underneath all of it, synaptic downscaling runs across the night to keep the system from saturating.