Researcher Reflections: From Understanding Transformers to Agents
DimitrisPapail · x · 2026-09-02
Dimitris Papailopoulos reflects on his research journey since 2022, discussing the shift from being amazed by LLMs to deeply investigating their internal mechanisms.
- Origin: After experiencing the davinci model, he pivoted his group to focus on understanding LLMs, leading to the "Looped Transformers as Programmable Computers" paper.
- Goal: Exploring how Transformers implement general-purpose algorithms of arbitrary depth via prompting.
- Perspective: With the rise of agents, the object of study has become the instrument used to probe the thing you want to understand.
Related event: Researcher Reflects on Journey from Transformers to Agents(2 posts)→
More from AGI Musings
- GaryMarcus warns OpenAI reportedly sacrificing CoT monitorability for performance — AndyMasley · 2026-09-02
- Expert: 1 million humanoids in US jobs within a decade, maybe — binarybits · 2026-09-02
- As AI Generates More Proofs, We May Actually Need More PhD Students — alejandroll10 · 2026-09-02
- OpenAI Researcher: Humans Prove Better Algorithms Exist, Open Source Has a Chance — burny_tech · 2026-09-02
- Evolutionary Biologist Argues Plants and AI Are Not Conscious — burny_tech · 2026-09-02
- Gary Marcus: Removing fragile CoT scaffolding is insane — GaryMarcus · 2026-09-02