Safety
@GaryMarcus @0xkkai The current AI agency architecture based on looping on an LLM core by piling on the same LLM context is a deeply flawed …
Romeo Lupascu has criticized the prevailing AI agency framework, which relies on repeatedly looping an LLM core with identical context. He calls this strategy “deeply flawed” and argues that it should
Romeo Lupascu has criticized the prevailing AI agency framework, which relies on repeatedly looping an LLM core with identical context. He calls this strategy “deeply flawed” and argues that it should be entirely scrapped in favor of a redesigned architecture.
Related
- @GaryMarcus Marcus isn't wrong. Everyone bolts tools onto transformers because they hit walls fast. Neurosymbolic AI is rising for a reason:…
- @GaryMarcus, great interview and explanations of the challenges in AI. https://youtu.be/_pSivPlRx5o?is=dlFN6YBXsMA-tW4h
- @theo What @GaryMarcus has been saying ... the engineering around LLMs matters even more than the LLMs themselves today
Source: Gary Marcus (X) | 2026-09-17