bwfan123

15 ms·

331.
by bwfan123 1y ago
"Might this goodness be due to the resonance that math reveals between mind and cosmos? In Greek, cosmos means “order,” and suggests also “beauty,” as in our word “cosmetic.” Doing math leads one to entertain the idea that the cosmos i
332.
by bwfan123 1y ago
> Imagine a world where machine learning models are written in Rust and can run on both Nvidia and AMD Not likely in the next decade if ever. Unfortunately, the entire ecosystems of jax and torch are python based. Imagine retraining all
333.
by bwfan123 1y ago
faang has sucked the oxygen in the air. For experienced engineers, the one upside to startups is if you are a founder and you are creating a change you want to see in the world. But this delusion is shattered as time goes by and you realize
334.
by bwfan123 1y ago
The risk-reward equation for startups vs boring-big-tech has turned completely upside-down. Back in the day, startups were the only lottery ticket to riches, not to mention do interesting work. But now, the tables are turned. For experience
335.
by bwfan123 1y ago
The popular science title was funnier with a pun on "mathed" [1] "Human teens beat AI at an international math competition Google and OpenAI earned gold medals, but were still out-mathed by students." [1] https:/&#
336.
by bwfan123 1y ago
nice comment, a reason why ML approaches may struggle in trading markets where other agents are also competing with you possibly using similar algos. or self-driving which involves other agents who could be adversarial. just training on pas
337.
by bwfan123 1y ago
thanks for articulating so nicely what needs to be said in this debate. Pushing errors leftward vs rightwards is such a nice metaphor, not to mention the metaphor on mental models. Also, your comment on why natural language is unable to des
338.
by bwfan123 1y ago
We may enter a vicious loop where writing is increasingly generated by LLMs. Then, LLMs have to train on their own output leading to model collapse. Hence, the models depend on human writing.
339.
by bwfan123 1y ago
lovely app, beautiful idea. might save relationships.
340.
by bwfan123 1y ago
[flagged]
341.
by bwfan123 1y ago
This paper [1] shows that gemini pro 2.5 without data-contamination and some minimal prompting (llm orchestration) can solve problems 1-5 on imo 2025. [1] https://arxiv.org/pdf/2507.15855 For problems with 2 page solut
342.
by bwfan123 1y ago
Problem 6 is puzzling. Neither openai nor deepmind answered it. Humans would put out partial answers - but here we saw no answer which is odd. Does that mean that the llms realized they could not solve it. I thought that was one of the limi
343.
by bwfan123 1y ago
generally, these class of constraint satisfaction problems fall under the "zebra puzzle" (or einstein puzzle) umbrella [1]. They are interesting because they posit a world with some axioms, and inference procedures, and ask if a c
344.
by bwfan123 1y ago
Would you mind sharing a sketch of one problem from the test you mention ? I am interested in how it looks.
345.
by bwfan123 1y ago
Loved Dave Barry's writings over the years. Specifically his quote on humor struck me as itself deep. "a measurement of the extent to which we realize that we are trapped in a world almost totally devoid of reason. Laughter is how
346.
by bwfan123 1y ago
Humans build theories of how things work. llms dont. Theories are deterministic and symbolic. Take the turing machine for example as a theory of computation in general, euclidean geometry as a theory for space, and newtonian mechanics as a
347.
by bwfan123 1y ago
I fed the problem 1 solution into gemini and asked if it was generated by a human or llm. It said: Conclusion: It is overwhelmingly likely that this document was generated by a human. ---- Self-Correction/Refinement and Explicit Goals:
348.
by bwfan123 1y ago
To me, this is a tell of human-involvement in the model solution. There is no reason why machines would do badly on exactly the problem which humans do badly as well - without humans prodding the machine towards a solution. Also, there is n
349.
by bwfan123 1y ago
Without sharing their methodology, how can we trust the claim ? questions like: 1) did humans formalize the input 2) did humans prompt the llm towards the solution etc.. I am excited to hear about it, but I remain skeptical.
350.
by bwfan123 1y ago
Did humans formalize the inputs ? or was the exact natural language input provided to the llm. A lot of detail is missing on the methodology used. Not to mention of any independent validation. My skepticism stems from the past frontier math
351.
by bwfan123 1y ago
Based on the past history with frontier-math & AIME 2025 [1],[2] I would not trust announcements which cant be independently verified. I am excited to try it out though. Also, the performance of LLMs on imo 2025 was not even bronze [3].
352.
by bwfan123 1y ago
> At this point, I think it can only be explained by ignorance, bad faith, or fear of becoming irrelevant. Based on the past history with frontier-math & AIME 2025 [1],[2] I would not trust announcements which cant be independently v
353.
by bwfan123 1y ago
on the surface this is a great achievement - if it holds . alpha-geometry required 1) human formalization of the question and 2) a solver for geometry If the questions were given as-is (without a human formalizing it) and the llm didnt ne
354.
by bwfan123 1y ago
> I'm always surprised by the number of people posting here that are dismissive of AI and the obvious unstoppable progress Many of us have been through previous hype-cycles like the dot-com boom, and have learned to be skeptical. So
355.
by bwfan123 1y ago
A genuine new insight is a "theory" which is a deterministic symbolic representation of the chaotic worlds of meaning. A theory is a formalism which can explain and predict phenomena. A theory consists of symbols to represent conc
356.
by bwfan123 1y ago
> Cue the "you're doing it wrong" crowd. or the "humans make mistakes too" crowd or the "just wait, we are at an inflection point in the sigmoid curve" crowd
357.
by bwfan123 1y ago
Similar to dot-com, part of the reason is the multiplier effect of all the AI investments. If these investments prove to be uneconomic, which I strongly suspect, the backend of this investment cycle is going to be brutal.
358.
by bwfan123 1y ago
same, - a better summarizing google for some queries - snippet generator but has not changed any workflows.
359.
by bwfan123 1y ago
- internet & ecommerce & online changed the way we shopped. - smartphones in every pocket changed behaviors around entertainment, communication, and commerce what behaviors of people will gen-ai change ? perhaps the way we learn (in
360.
by bwfan123 1y ago
> We’re navigating a tightrope as Superintelligence nears. If the West slows down unilaterally, China could dominate the 21st century I stopped reading after this. First, there is no evidence of Superintelligence nearing or even any clea

More ›