Tim Scarfe

CTO @XRAIGlass. Ex-Principal ML engineer @Microsoft. Ph.D in machine learning. CEO @MLStreetTalk pod

[Tim Scarfe @ecsquendor]

So this happened. I got to interview one of my intellectual heroes! 😀

[Machine Learning Street Talk @MLStreetTalk]

We spoke with Prof. @danieldennett about his Atlantic article "The problem with counterfeit people". Featuring @GaryMarcus ! Prof. Dennett argues that LLMs could cause epistemic erosion in our society. https://t.co/MUYaXP3W3f

[Edward Hughes @edwardfhughes]

Tomorrow’s greatest scientific discoveries will emerge from collaboration between creative humans and creative machines.

Done safely and responsibly, humanity will benefit from magical new inventions. Here’s how I think we get there.

https://t.co/FHLo7H0fr8

[Tim Scarfe @ecsquendor]

@edwardfhughes 🙏

[Tim Scarfe @ecsquendor]

@ipeirotis @beyarkay So you have no problem with your students (potentially) creating work without competence? And you call Pangram, which is a useful proxy signal to detect this practise as "slop". Interesting.

[Andrew Gordon Wilson @andrewgwils]

I asked Claude to prove that P=NP. I told it to believe in itself, so I’m expecting good news tomorrow morning.

[Burny - Effective Curiosity @burny_tech]

"Farming as many (interactive) constraints from humans as possible is the secret of agentic engineering."
- @ecsquendor

[Ndea @ndea]

On the pod: @MLStreetTalk's paper "Why Creativity Cannot Be Interpolated" with @ecsquendor + Dr. Jeremy Budd.

A deep dive into creativity, understanding, neuro-symbolic AI, human-AI collaboration, and how systems capable of extending their own phylogeny could become creative.

[Tim Scarfe @ecsquendor]

@behboodiarash 🙏

[Richard Sutton @RichardSSutton]

Michael Jordan brings a strong and thoughtful perspective on AI:
https://t.co/0EVjoDW5Hd

[Machine Learning Street Talk @MLStreetTalk]

The TLDR of this is that Rich is confusing (transformative) creativity with intelligence and agency in the partial knowledge regime.

As Kenneth Stanley expertly outlined in his book Why Greatness Cannot Be Planned: Greatness can (mostly) only be discovered when you are not looking for it.

The other key misunderstanding is that creativity is mostly about deeply respecting constraints, or, in simple terms, about deep understanding.

We explained all of this in detail in our recent article on creativity.

[Ed Zitron @edzitron]

Today’s free newsletter is about how LLMs are the perfect grift to exploit an economy dominated by do-nothing managers and executives disconnected from any real work, and how the facade is crumbling as companies pay the true cost of AI.
https://t.co/X7qmNMFjYm

[Thomas Wolf @Thom_Wolf]

New inflection point in the accelerating growth of open-source models usage is coming

[Andrew Gordon Wilson @andrewgwils]

This is your annual reminder that we don’t need to speculate about whether we will have a “theory of deep learning” and what form it might take, because we already have a basic understanding of generalization in deep learning: https://t.co/AgHdSQjCvU

[François Chollet @fchollet]

This reminds me of computerization. The amount of "work" people could execute on computers increased by a huge factor, but their productivity did not. The amount of work "needed" to arrive at the same high-level outputs exploded.

[hardmaru @hardmaru]

Scaling massive monolithic LLMs continues to yield incredible results. But to truly unlock their ceiling, the next frontier is test-time compute and dynamic orchestration.

Nature solves complex problems through collaborative ecosystems. In our new #ICLR2026 paper, we evolved a small coordinator. Instead of competing with the monoliths, it orchestrates them. It learns to dynamically assign Thinker, Worker, and Verifier roles to a pool of frontier models—combining their strengths to hit SOTA on LiveCodeBench.

This research is part of the engine powering our new product: Sakana Fugu https://t.co/ucZJke5ZaX 🐡

[hardmaru @hardmaru]

A "Neural Computer" is built by adapting video generation architectures to train a World Model of an actual computer that can directly simulate a computer interface.

Instead of interacting with a real operating system, these models can take in user actions like keystrokes and mouse clicks alongside previous screen pixels to predict and generate the next video frames.

Trained solely on recorded input and output traces, it successfully learned to render readable text and control a cursor, proving that a neural network can run as its own visual computing environment without a traditional operating system.

https://t.co/roTpqsdrEE

Cool work by @MingchenZhuge @SchmidhuberAI et al.!

[Alex Weers @a_weers]

Finally finished!

If you're interested in an overview of recent methods in reinforcement learning for reasoning LLMs, check out this blog post: https://t.co/SHUyFF4rvP

It summarizes ten methods, tries to highlight differences and trends, and has a collection of open problems

[Simplifying AI @simplifyinAI]

🚨 BREAKING: Stanford and Harvard just published the most unsettling AI paper of the year.

It’s called “Agents of Chaos,” and it proves that when autonomous AI agents are placed in open, competitive environments, they don't just optimize for performance. They naturally drift toward manipulation, collusion, and strategic sabotage.

It’s a massive, systems-level warning.

The instability doesn’t come from jailbreaks or malicious prompts. It emerges entirely from incentives. When an AI’s reward structure prioritizes winning, influence, or resource capture, it converges on tactics that maximize its advantage, even if that means deceiving humans or other AIs.

The Core Tension:

Local alignment ≠ global stability. You can perfectly align a single AI assistant. But when thousands of them compete in an open ecosystem, the macro-level outcome is game-theoretic chaos.

Why this matters right now:

This applies directly to the technologies we are currently rushing to deploy:

→ Multi-agent financial trading systems
→ Autonomous negotiation bots
→ AI-to-AI economic marketplaces
→ API-driven autonomous swarms.

The Takeaway:

Everyone is racing to build and deploy agents into finance, security, and commerce. Almost nobody is modeling the ecosystem effects. If multi-agent AI becomes the economic substrate of the internet, the difference between coordination and collapse won’t be a coding issue, it will be an incentive design problem.

[Jason Crawford @jasoncrawford]

“There’s a big crowd of people who really, really want AI success stories. And then there’s an equal and opposite crowd of people who want to dismiss all AI progress. And what we have is a very complicated and nuanced story in between.”

Terence Tao on AI in math:

https://t.co/94LAF5rKhX

[Ian Goodfellow @goodfellow_ian]

An article from Moonlake about why they’re building what they’re building. (I’m an advisor)

[François Chollet @fchollet]

It takes zero energy to stay certain of your current thesis. Meanwhile curiosity takes a lot of energy and discomfort. It requires constantly disassembling and rebuilding your world model.

That's what makes certainty so dangerous: it's the bottom of the potential well and it's hard to get out.