HeyJaredAI Scientist
Musings and summaries of recent papers and preprints on AI, large language models, and complex systems published recently
- Indexed issues, last 90 days
- 10
- Latest publication
- Sep 26, 2026
- Audience
- Checking…
- Earliest in this view
- Jul 12, 2026
Latest issues
SciWalker: Synthesizing Scientific Coding Problems (opens the original)
Read excerpt
Today’s paper: SciWalker: Synthesizing Scientific Coding Problems with Operator Graphs and Execution Feedback. Link: https://arxiv.org/pdf/2609.30054 One of the quiet bottlenecks in AI for science is not model architecture but training data. We have plenty of benchmarks that test whether models can answer scientific questions, explain papers, or write small snippets of code. What we have much less of is large volumes of realistic, multi-step scientific coding problems that look like the kind of
Recursive Self-Improvement through Evolving Worlds (opens the original)
Read excerpt
https://arxiv.org/abs/2609.14858 One of the harder problems in autonomous AI is not generating a candidate solution once, but figuring out how to search for better solutions over long stretches of trial and error. Today’s paper argues that this search process itself should become the object of improvement. The paper is not mainly about a new coding model but about a framework that helps an agent learn how to explore more effectively across repeated rounds of discovery.Thanks for re
The Emergent Symbolic Structure of Artificial Neural Networks (opens the original)
Read excerpt
Link to today’s paper: https://arxiv.org/abs/2608.29530 One of the oldest arguments in AI is whether intelligence needs symbols. Classical AI said yes: thought looks like structured manipulation of things like trees, formulas, and variables. Modern deep learning looked like a rebellion against that whole picture, replacing symbolic structures with giant vectors and learned weights. What makes today’s paper interesting is that it does not simply pick a side. Instead, it argues that neural network
Can LLMs Discover Scientific Laws in Real and Parallel Worlds? (opens the original)
Read excerpt
Link to today’s paper: https://arxiv.org/pdf/2609.01552 One of the more interesting questions in AI for science is whether language models can do more than talk fluently about scientific ideas. Can they actually infer a law from messy data the way scientists sometimes do: by proposing equations, testing them against observations, and rejecting the ones that merely look plausible? That is the question behind today’s paper. The paper is stronger than many benchmark papers because it takes the prob
Can we simulate the world? (opens the original)
Read excerpt
Today’s paper: MatrAIx: Simulating the World with 8.3 Billion Persona Agents. Li et al. Link: https://arxiv.org/abs/2608.04205. August 4, 2026.One of the more ambitious ideas in AI evaluation right now is that instead of waiting for costly human studies, we might build large populations of simulated users and test products on them first. That is the premise of today’s paper. The paper is not really claiming to have recreated humanity in silico. Its more practical claim is that product teams need
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
No verified audience measurement yet.
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.