Fedus: Intelligence is not scalar; AI shows extreme, non-intuitive spikiness
Liam Fedus · AI for Atoms: How Periodic Labs is Revolutionizing Materials Engineering with Co-Founder Liam Fedus · Apr 3, 2026 · at 22:11
Periodic Labs co-founder Liam Fedus explains why models that excel at specific benchmark domains can collapse on adjacent variations.
0:00 / 0:23exact quote · 23.6s
▶ Watch the full episode on YouTube → 720p mp4 · rendered on demand · StarZero watermark
“I think one fallacy is thinking about intelligence as a scalar. We've consistently seen these systems have a very odd spikiness, and it's actually possible to architect a system that is world-class on some math domain, but then you could do some perturbations to the questions and actually degrade it substantially, so it's like a bad high school student.”
quote is from the automated transcript, cleaned for reading: filler sounds and stutters are removed, nothing is rephrased. names can be misheard (the analysis reads context, assessments check outside sources). how →
More from Liam Fedus
Fedus: Physical sciences and engineering will follow machine learning scaling laws
“And I think the physical sciences, physical engineering, Will have a very similar property where we establish these scaling properties and Bring that mindset.”
Fedus: Self-improving software AI will not automatically generalize to domains like biology
“So rolling forward that software engineering self-improvement, I think you're going to have a system that can write complete repositories, identify bugs, refactor code, But it doesn't suddenly understand biology. Right? It's just like there's a domain gap ther…”
Fedus: Recursive self-improvement is happening in software and entering AI research
“So, In that domain, I think it's happening now-ish. And I think we'll see the same thing too for AI research. That's a slower outer loop because now the experiment isn't just checking some unit tests passing, but it's checking what was the scaling property? Di…”
Fedus: John Schulman steered OpenAI toward a general chatbot over narrow tools
“And we're all spitballing ideas, like writing bot coding bot. You know, very natural at the time. Some of our least interesting ideas were a meeting bot, so it would just sit in a Google Meet, take notes, and then send out, like, to-dos after. But John Schulma…”
Fedus: AI must connect to the physical world to accelerate science
“The opinion that I and others held as periodic was, you're not going to see the same kind of acceleration in science and technology unless you start connecting these things to the physical world.”
Fedus: Training materials AI on academic literature fails to find ground truth
“One of the engineers on our team was looking at a reported material property and It was just sort of extracted values from literature, and it was really interesting to see the reported value spanned many orders of magnitude. And so you train an ML system on th…”