Musing 88: GSM-Symbolic: Understanding the Limitations of Mathematical Reasoning in Large Language Models open.substack.com/pub/aiscientis…

Musing 100: What did I learn about LLM research from the previous 99 musings? open.substack.com/pub/aiscientis…

Musing 102: Agent Laboratory: Using LLM Agents as Research Assistants open.substack.com/pub/aiscientis…

Musing 103: UI-TARS: Pioneering Automated GUI Interaction with Native Agents open.substack.com/pub/aiscientis…

Fast Think-on-Graph: Wider, Deeper and Faster Reasoning of Large Language Model on Knowledge Graph open.substack.com/pub/aiscientis…

Musing 106: Can We Trust AI Benchmarks? An Interdisciplinary Review of Current Issues in AI Evaluation open.substack.com/pub/aiscientis…

Musing 107: A Survey on Large Language Models for Automated Planning open.substack.com/pub/aiscientis…

Musing 108: Towards Robust Legal Reasoning: Harnessing Logical LLMs in Law open.substack.com/pub/aiscientis…

Musing 110: The Value of Goal Commitment in Planning open.substack.com/pub/aiscientis…

Musing 111: An LLM-Based Approach for Insight Generation in Data Analysis open.substack.com/pub/aiscientis…

We hear a lot about #AIagents these days, but I was impressed with this survey on where, how, and why #AI agents may be useful in different subareas of #education. Musing 117: Evolution of AI in Education: Agentic Workflows open.substack.com/pub/aiscientis…

Musing 121: Decomposing Elements of Problem Solving: What "Math" Does RL Teach? open.substack.com/pub/aiscientis…

This is a very unique paper out of #Nvidia and #Gatech (and you can also correspond with the authors and be on the public record if you have thoughts!). There’s a lot to digest in it but I’ve distilled it down to the key basics in my musing here open.substack.com/pub/aiscientis…

Musing 125: The Singapore Consensus on Global AI Safety Research Priorities open.substack.com/pub/aiscientis…

Musing 126: Does Math Reasoning Improve General LLM Capabilities? open.substack.com/pub/aiscientis…

Musing 133: Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMs open.substack.com/pub/aiscientis…

Musing 136: Why Language Models Hallucinate open.substack.com/pub/aiscientis…

Musing 137: Defeating Nondeterminism in LLM Inference open.substack.com/pub/aiscientis…

Musing 138: Advances in Large Language Models for Medicine open.substack.com/pub/aiscientis…

Musing 139: GDPVAL: Evaluating AI Model Performance on Real-World Economically Valuable Tasks open.substack.com/pub/aiscientis…