“If an AI cannot apply an abstract lesson to a new situation, it is not truly reasoning or learning”, new study with further evidence backing up what I have been saying for 25 years. cc @dwarkesh_sp
AGI
-
Google DeepMind paper maps the path from AGI to ASI
By
–
From AGI → ASI Google just mapped what comes after human-level AI. there's a brilliant new paper from Google DeepMind called "From AGI to ASI," and it skips the fight everyone else is having. almost every AI debate today is about reaching human-level intelligence. this one
-
Gary Marcus notes rediscovery of his 2022 AI insights
By
–
So funny how so many people are rediscovering so many things I pointed out in late 2022. (In this case in my essay “AI’s Jurassic Park Moment”)
-
Gary Marcus questions if Karpathy was hired for recursive self improvement
By
–
is that actually true that @karpathy was hired to “run recursive self improvement”? https://t.co/nDmB4P99Z3
— Gary Marcus (@GaryMarcus) 13 juin 2026is that actually true that @karpathy was hired to “run recursive self improvement”?
-
Exponential accumulation of AI capabilities and need for open weights
By
–
We believe that capabilities will most likely continue to accumulate exponentially. We believe that a world where a handful of companies provide all AI is worse than one where many can develop it. We believe that open weights and the
-
From AGI to ASI: definition by Google DeepMind
By
–
This article by Google DeepMind defines how AGI is a unique human-level general system, and ASI is a system or collective that surpasses large expert human organizations in almost everything. They argue that the leap could
-
AI models fail simple long tasks despite passing exams and olympiads
By
–
8/ It fits a pattern researchers keep hitting. The models that pass medical exams and win math olympiads still fall apart on simple tasks once those tasks run long. Source: Patel, Wang, Fan. PNAS Nexus, 2026.
-
AI failed Stroop task despite knowing the rule
By
–
5/ One result stood out. In one test the model correctly named the experiment and explained the Stroop task in detail. Then it failed the task anyway. Knowing the rule did not help it follow the rule.
-
The smartest AIs fail a simple color test
By
–
The smartest AI models on Earth just failed a color test that a 6-year-old can pass. The longer the test went, the worse they performed. One went from 91% to 15%. Scientists have discovered a flaw in the way AI
-
Training a Mythos model requires noticeable amounts of compute
By
–
And by regulatable amounts of compute, I mean training a Mythos class model uses enough power and chips that national governments will obviously notice. No ine is training a model of that size without permission