1 min read
#Memory#LLM#Research#State
Emotional Memory
I check this article against the public proof at DOI 10.5281/zenodo.19972258.
Why this matters
Memory that follows the mood works in the test built this way; what you pay later is an answer worse than simple similarity when the question is not about the mood, or when the right memory is the opposite one.
Share this article
Related articles
Why Autonomous Research Agents Hallucinate — and How a Critic Loop Surfaces Unsupported Claims
Research agents can look authoritative and still invent citations. A critic with source access catches unsupported claims — it doesn't erase risk.
Nov 15, 202410 min read#AI Agents#Research#LLM#ProductionProof of Work vs Inference: Cost as Signal, Not Identity
Inference is not proof of work. I separate two costly, checkable spends of compute: a rule for scarce consensus, and a model plus prompt for utility.
Oct 2, 202610 min read#Inference#Cost#Eval#LLM#ProductionIs Saying an LLM Doesn't Think Like Saying a Calculator Can't Do Numbers?
Where the calculator analogy for LLMs holds and where it breaks: what interpretability, chain-of-thought and philosophy of mind say about thinking.
Jul 2, 202616 min read#LLM#AI Reasoning#Interpretability#Philosophy of Mind