HeyJaredScientist's AI
Scientist's AI is an ongoing series that translates today's AI research into clear, intuitive explanations anyone can understand.
- Indexed videos, last 90 days
- 10
- Latest publication
- Sep 28, 2026
- Audience
- ~395 subscribers
- Earliest in this view
- Jul 28, 2026
Latest videos
Scientists Found Hidden Symbolic Structure Inside AI (opens the original)
Read excerpt
Researchers can manipulate symbolic structure inside a neural network and actually change what the model does. That makes it much harder to dismiss the structure as coincidence.
The AI Idea That Refused to Die (opens the original)
Read excerpt
Symbolic AI failed. Neural networks won. But new research adds a surprising twist to that story.
AI is More Symbolic Than we Thought (opens the original)
Read excerpt
For decades, symbolic AI and neural networks were treated as fundamentally different approaches to intelligence. But new research suggests that today’s neural networks may be developing something surprisingly symbolic on their own. In this episode, I break down how researchers discovered symbolic structure hidden inside neural network representations, how they tested it by directly intervening on those representations, and what it tells us about how modern AI actually works. The twist: symbolic
What Happens When AI Agents Form Societies? (opens the original)
Read excerpt
What happens when AI agents stop working alone and start behaving like communities? In this episode, I explore research using statistical mechanics and the Ising model to predict the collective behavior of AI agents. We look at how networks of interacting agents develop persistence, polarization, consensus, and surprisingly, why agreement and truth can exert a stronger pull than disagreement and falsehood. A fascinating example of physics being used to understand the emerging world of multi-agen
The Real Reason AI Is So Data-Hungry (opens the original)
Read excerpt
A human brain runs on about 30 watts. A large language model burns through trillions of tokens, and enormous energy, to do things a child does effortlessly. So why is AI so wildly inefficient? A new paper points to a surprising answer: AI is learning the wrong thing. In this episode I break down a paper called *Learn from Your Own Latents and Not from Tokens: A Sample Complexity Theory*. The argument: AI isn't data/energy hungry because it's big, it's because it predicts the next TOKEN instead o
Publishing over time
Last 90 days. Choose a month to open its work.
Recurring subjects
Named in the text we hold. One piece can cover several.
Audience
~395 subscribers
Measured Sep 19, 2026
Source's subscribers, not the number who saw an individual piece.
About this data
Counts cover the work we have indexed. Tone needs enough text and a confident classification. Excerpts and episode notes are not full articles or transcripts.
Identity or attribution wrong? Suggest a correction.