7/
I'm not convinced yet how! but I believe the answer is in the Markovian/Attention model somehow to eradicate the hallucination chains. Something like states summarizations. This is an open discussion. Maybe you guys have an opinion @sama@elonmusk
What if Markovian is the key to fixing the reasoning chaos of AI? like how it solves Nekrasov's. My silly recap of AI history + open proposal for whoever build AI.
6/
Agentic AI (2023+):
AI becomes active, it takes actions to validate its reasoning (search, API calls, tools).
Cool! But here’s the new chaos:
The Structural Confidence Loop
→ The AI believes its own flawed reasoning
→ Then takes misaligned action
→ Snowballs hallucination
@irwndfrry Quality education for free
Pengen bangun virtual school untuk AI, LIVE teaching, dosen Top Univ dunia. Tapi gratis. ISA agreement. Rasio student:dosen 7:1, jaga quality.
Kebetulan ada link kampus di US. Pengalaman 4 tahun peneliti robotika. YSEALI AFP. Mawapres Indo 2021.