Micah G. Allen

Why they matter

Researcher with public evidence across AI research.

AI signals
3
past 30d
Sources
3
distinct domains
Discussions
4
past 30d
Latest signal
14d ago
View every signal from Micah G. Allen →
Professor of Computational Neuroscience and Psychiatry, Aarhus University. PI @ the Embodied Computation Group. We study perception, interoception, & metacogniton. https://www.the-ecg.org

Articles & links

↻ Micah G. Allen reposted
@shimmermathlabs.com

how to add dreaming to your agent (theory) arxiv.org/abs/2606.03979

Language Models Need Sleep: Learning to Self-Modify and Consolidate Memories arxiv.org
AI Weekly's analysis →
  • The paper frames continual learning as a 'Sleep' paradigm with two stages, Memory Consolidation and Dreaming, inspired by human learning.
  • Memory Consolidation uses 'Knowledge Seeding' to distill a smaller model's memories into a larger network via on-policy distillation plus RL imitation.
  • Dreaming uses reinforcement learning to generate a curriculum of synthetic data for self-improvement without human supervision.
Read full analysis →
View on Bluesky →
↻ Micah G. Allen reposted
@cstrauch.bsky.social

How much of a risk are LLM-generated bots for online behavioral experiments? We present results of an entirely natural language prompt-generated agent, capable of absolving any of the 26 experiments we tested it on. Most tasks are passed with data that can pass as human osf.io…

OSF osf.io View on Bluesky →

This is still very much WIP but we have this “lab book” for in depth guidance on how to use agentic coding (e.g. Claude code) in the lab, largely adopted from @russpoldrack.org’s book chapter, a few other sources, and our own experience. I’m working on a new lab manual. www.th…

Agents in the Lab the-ecg.org
View on Bluesky · ♥ 7 ↻ 0 ↩ 2 · 14d ago

as many as 34% self report doing this in one study... journals.sagepub.com/doi/pdf/10.1... 9% detected here journals.sagepub.com/doi/10.1177/... it is a real problem to be sure. we were shocked that Sol could do our task as easily and convincingly as it did.

journals.sagepub.com
View on Bluesky · ♥ 3 ↻ 0 ↩ 1 · 59d ago

as many as 34% self report doing this in one study... journals.sagepub.com/doi/pdf/10.1... 9% detected here journals.sagepub.com/doi/10.1177/... it is a real problem to be sure. we were shocked that Sol could do our task as easily and convincingly as it did.

journals.sagepub.com
View on Bluesky · ♥ 3 ↻ 0 ↩ 1 · 59d ago

Recent commentary

So… we just found out that GPT sol can complete a complex online behavioral/cognitive study producing nearly indistinguishable behavior and subjective reports… right as we are about to launch. Is this doomsday for online cognitive testing?

View on Bluesky · ♥ 125 ↻ 68 ↩ 31 · 59d ago

Just reviewed PhD applications for an open call. 100% of the applicants I reviewed disclosed using AI to help craft their applications. Like it or hate it, we are living in that world.

View on Bluesky · ♥ 16 ↻ 1 ↩ 4 · 92d ago

As a scientist with ADHD AI has changed my life in so many positive ways. I’m deeply tempted to write about it.

View on Bluesky · ♥ 14 ↻ 0 ↩ 4 · 36d ago

Don’t worry guys I asked ChatGPT if this was the start of the autonomous AI take over and it said no so nothing to worry about.

View on Bluesky · ♥ 11 ↻ 0 ↩ 1 · 6d ago

In Micah G. Allen's orbit

Center = Micah G. Allen. Left = members they follow (green edges). Right = members who follow them (blue edges). Top = mutual follows (orange edges, slightly larger). Drag any node to reposition; click to open that profile.

Are you Micah G. Allen? Show it.

Add the Who’s Who of AI badge to your site or bio. It links back to this profile.

Listed in AI Weekly's Who's Who of AI

Markdown: [![Listed in AI Weekly's Who's Who of AI](https://aiweekly.co/modules/custom/aiweekly_whoswho/images/whoswho-badge.svg)](https://aiweekly.co/whos-who/person/micahgallen-com)