bwfan123

31 ms·

451.
by bwfan123 1y ago
LLMs have mesmerized us, because, they are able to communicate meaning to us. Fundamentally, we (the recipient of llm output) are generating the meaning from the words given. ie, llms are great when the recipient of their output is a human.
452.
by bwfan123 2y ago
an analogy i find useful is the search-engine (google). yea the search-engine improved productivity of almost everyone, but didnt change any workflows.
453.
by bwfan123 2y ago
Many areas of humanities devolve into a bag of words where the evaluation of it is subjective depending on ones tastes - and LLMs excel here, and academics are rightfully scared. As an analogy and contrast, take the case of euclidean geomet
454.
by bwfan123 2y ago
why request code.. when all of pytorch2 is open, and this is built on top of it with some enhancements, why not put this out in the open
455.
by bwfan123 2y ago
play adversary, and seed it with dubious content. to your point though, i can bet there are departments whose job is to apply data-fixes for these problems - since each prompt that stumps an AI is equivalent to a "bug"
456.
by bwfan123 2y ago
I thought ceos are more like coaches who motivate, and inspire. But not dictate employees on how they should execute. Since, engineers are expected to be far more capable than the ceo on their daily work (if not, the ceo needs to evaluate h
457.
by bwfan123 2y ago
I use cursor, and these tools dont encode a causal representation of how something works. Hence they cannot root cause and fix bugs. Fixing issues with 99% correctness snowballs into a hot mess very soon due to compounding errors. Somehow t
458.
by bwfan123 2y ago
if you looked hard, you might even find the offices of the American Institute of Math at a fry's location... https://en.wikipedia.org/wiki/American_Institute_of_Mathemat...
459.
by bwfan123 2y ago
It is good to see both sides of the debate. Thus far, media was feeding into the hype, now media is being skeptical. This is a nature evolution path of any new tech. why be salty ?
460.
by bwfan123 2y ago
just curious: why dont the aibots play by the rules ? what incentivized them to misbehave.. is it just sloppy crawler code (ie, the misbehavior is unintentional) ?
461.
by bwfan123 2y ago
It is the golden rule. Whoever has the gold rules. Laws are for mortals, while big-tech can do whatever they please while channeling their masculine energy. Basically, behave in a self-serving manner violating basic norms such as dont do un
462.
by bwfan123 2y ago
Around 2011 or so, when facebook went IPO, there was a sudden step-jump in engineer total comp in silicon valley. Prior to that, a tech-job was just another job, and the path to riches was via startups and risk taking. Then, because of FOMO
463.
by bwfan123 2y ago
It was the first boom-and-bust experience for me. I got laid off from a startup that folded after the bust. That made me risk-averse and a skeptic. The housing bust of 2008 had a completely different flavor and character unlike dot-com. And
464.
by bwfan123 2y ago
please provide context when making statements, thats a better way to communicate your ideas. what, when, why, how, where - so a reader takes your map, and can relate to the territory.
465.
by bwfan123 2y ago
I had an aha moment recently. An excited AI researcher claimed, wow: claude could solve this IMO problem. Then, a mathematician pointed out a flaw which the AI researcher overlooked. The AI researcher then prompted the AI with the error and
466.
by bwfan123 2y ago
whaa ? have you read the article ?
467.
by bwfan123 2y ago
so, the narrative from semianalysis is shifting from "agi is imminent" to "robotics will take over" - what does that mean for semi sales ?
468.
by bwfan123 2y ago
> Most natural language arguments are not sound because the argument is not deductive logic. Most natural language arguments are persuasive, not formal reasoning related notes that there is some evidence that "Language is primarily
469.
by bwfan123 2y ago
With llms and theorem provers doing the formalism, what is left is intuition - love the intuitionist focus in this book, also tying into what Bessis said in [1] I intend to read this book in depth. Author: please fix the html rendering. [1]
470.
by bwfan123 2y ago
I am interested in this problem as well. Please share any notes. I am attempting to create parameterized "logic" problems (similar to the zebra puzzle) which cannot be solved by LLMs even when they are trained on it, or even when
471.
by bwfan123 2y ago
Mathematician David Bessis describes similar exercises in his book [ref] - wherein he explores a room blindfolded to awaken other senses, or visualize the places he has been to. [ref] https://www.amazon.com/Mathematica-Secre
472.
by bwfan123 2y ago
very nicely articulated.. so, you are suggesting to be "sensitive" to "feelings" - such the one of being "off" - since feelings primarily provoke thought. Another feeling that i have is one of "inconsisten
473.
by bwfan123 2y ago
Physicist Bohm has written about this. That, the beginning of new perception is in mindful listening or mindful awareness.. But most of us are stuck in our thought-rut preventing this. Perception itself carves out categories almost unconsci
474.
by bwfan123 2y ago
Problem with startups like Devin (AI sw engineer) and Sakana (AI research scientist) is that they are full of hot-air. They get caught up in the hype, and focus on the marketing and not the essential engineering.
475.
by bwfan123 2y ago
I found the gpumode lectures, videos and code right on the money. check them out.
476.
by bwfan123 2y ago
lovely from-scratch implementation, without the magic and complexity of lex and yacc. brings back memories from 30+ years ago when I implemented a subset of C in C using lex and yacc in a compiler course. in your blog, i would have liked to
477.
by bwfan123 2y ago
imo, The idea of a cause is a logical concept of containment when used in theories. A causes B means the phenomena represented by A implies the phenomena represented by B. So, causation is a device of our symbolization and understanding of
478.
by bwfan123 2y ago
We dont know what the human brain does, but we know it can produce symbolic theories or models of abstract worlds (in the case of math) or real worlds (in the case of science). It can also produce the "symbolic" turing machine whi
479.
by bwfan123 2y ago
0) causal model == symbolic representation with associated rules for generating statements 1) understanding anything == building a causal model of it 2) intelligence == ability to build causal models 3) reasoning == proving or disproving st
480.
by bwfan123 2y ago
A "causal model" is needed to fix bugs ie, to "root-cause" a bug. LLMs yet dont have the idea of a causal-model of how something works built-in. What they do have is pattern matching from a large index and generation of

More ›