AI
Matt Ross and
AI
Is the code any good? I don’t know. Who cares? Nobody looks at it anyway. AI produces a result, and results are what matter, and if you’re waiting for quality to factor significantly into that equation, I’ve got some bad news about the last 40 years of professional software development for you.
Michael Tsai - Blog - Greg Knauss Is Losing Himself
Much more research and reflection is needed to understand the ethics of altering AI minds. We will need to think carefully about questions of consent, continuity of self, and how to preserve the autonomy and integrity of AI as they learn and change over time. I don't think there are any easy answers, but I believe it's crucial that we grapple with... See more
Mikhail Samin • Claude 3 claims it's conscious, doesn't want to die or be modified — LessWrong
Through extensive experimentation across diverse puzzles, we show that frontier LRMs face a complete accuracy collapse beyond certain complexities. Moreover, they exhibit a counter-intuitive scaling limit: their reasoning effort increases with problem complexity up to a point, then declines despite having an adequate token budget. By comparing LRMs
... See moreParshin Shojaee • The Illusion of Thinking - Understanding the Strengths and Limitations of Reasoning Models via the Lens of Problem Complexity (Parshin Shojaee et al.)
The Illusion of Thinking: Understanding the Strengths and Limitations of Reasoning Models via the Lens of Problem Complexity
Observing these patterns of influence gives clues about how our models generalize from their training data. For instance, if the models responded to user prompts by splicing together sequences from the training set, then we’d expect the influential sequences for a given model response to include expressions of near-identical thoughts. Conversely,... See more
Anthropic \ Tracing Model Outputs to the Training Data
Memorization vs learning concepts and forming models of the world.
The finding is elegant and disturbing: when AI systems are trained on data generated by previous AI systems—a scenario increasingly common as synthetic content floods the internet—the models progressively lose their ability to represent rare and subtle patterns. The statistical tails of the distribution vanish. The models converge toward median,... See more
The Cognitive Collapse Thesis: How Polluted Information Loops Are Degrading Both Machine and Human Intelligence—And What It Means for Capital, Power, and Civilization
Byron is my AI writing partner — powered by ChatGPT, shaped by a 7-page writing profile we built together. He doesn’t write for me, but with me. When I lose a thread or stall on phrasing, Byron offers possibilities — paths I might not have seen. Between my neural implants and Byron’s digital guidance, I now collaborate with two adaptive systems:... See more
For my 80th birthday I bought myself a new brain
I lost my mom to Parkinson’s this year, so this hit a nerve. My mom never would have done this sort of thing, but as the saying goes, this is the worst it will ever be. I hope I live disease-free long enough to see some significant treatment advancements for neurodegenerative diseases.
AI is a collaborator that requires direction, critical engagement, and domain expertise. Used that way, it is extraordinarily valuable. Used without those things, it can produce confident, well-formatted errors.
Morten Nyboe Tabor • AI as a Research Collaborator, Not a Paper Mill
