Indeed. Regexes are _regular_, which is non hierarchical by definition.
The two are useful for different layers of abstraction: regex is for lexing and PEG is for parsing.
Speaking of the Chomsky hierarchy, last I checked, it is still unproven whether or not PEGs can parse all context free languages. Intuitively, they're _probably_ weaker than CFGs, but no one has yet provided a counter example.
PEGs aren't contained within the context-free languages, so it's not really intuitive that they're weaker, particularly as nobody has yet come up with a context-free language that a PEG cannot parse. They're capable of parsing things context-free grammars cannot.
I think people are focusing on the training data issue too much. If the data was contaminated, I can still blame that on negligence.
But, at least with the Navier-Stokes solution, it's clear [^1] that they learned that Alpöge and Buckmaster were getting close to a solution and learned of the general approach they were taking. Only after learning the secret to cracking the problem did they send the first prompt.
What makes this worse to me is the intention. They intentionally threw $15 million in compute at the problem in order to scoop the result. They intentionally left Buckmaster and Alpöge out of the citations.
Data contamination should be enough to disqualify them from the prize, but I can believe it to be accidental. On the other hand, someone made an intentional decision to scoop the result by throwing money at the problem. That's so much worse.
[^1]: That's the timeline claimed by Buckmaster, and no one from OAI has disputed it.
I think you’re overlooking what I’m implying here. It’s not that they knew contamination was possible but they went ahead anyway. To spell it out just a little bit more: learning the answer might be in model X’s training data made them believe that model X specifically might be able to solve the question, and they were able to very quickly find enough certainty about the former to commit millions of dollars to the latter.
“ On Tuesday, September 1, we heard rumors that two Millennium Prize problems had been resolved. Inspired by these rumors and by the step change in performance of our internal model, we launched an effort to evaluate it on all open Millennium Prize problems and a few other high-impact problems.”
They do not explicitly admit to knowing about NS specifically, but are extremely explicit that they tried to scoop some potential millennium prize winners.
So then they DIDN'T "learn the secret to cracking the problem". They simply knew that part of the problem was solved. Knowing a problem can be solved and knowing the solution are not the same thing.
The claim that OpenAI somehow used the mathematicians' ideas to leapfrog them seems unsupported at this time and IMHO it was irresponsible to bring it up because credulous people will immediately believe that narrative.
And from my perspective, if some math folks typing in a few questions to OpenAI provides sufficient training data for OpenAI to solve a big problem... that's amazing! A few conversations/prompts out of the billions that OpenAI trains on lead to this result- that means there is an awful lot of low-hanging fruit that could be exploited cheaply.
I'm curious how many other 300 billion output tokens OpenAI has "paid for" that have resulted in no breakthroughs.
Either they had a pretty good idea that investing this type of money in that compute on a model in training would lead to these specific results, or they gambled with other people's money.
I want to hear about the gambles and expenditures they don't brag about. In America's energy economy, there's finite resources to expend.
The (unprovable, yes, without OpenAI being willingly transparent) argument is that openAI constructed a prompt to scoop them using some inside knowledge about the approach, which they allude to in the announcement.
In the transcripts, Brubeck is very cagey and evasive about the prompt, when it was supplied, and its contents.
> learning the answer might be in model X’s training data made them believe that model X specifically might be able to solve the question, and they were able to very quickly find enough certainty about the former to commit millions of dollars to the latter.
They don’t need to know, because their IP stealing machine knows for them. They just have to buy enough compute, and someone else’s work is theirs.
I said “very quickly find enough certainty” to suggest hypothetical situations like “someone searches the conversation logs, confirms for themselves the solution is present, then shares the confidence gained from this knowledge without explicitly sharing the knowledge itself”. That person could recuse themselves from the project so the project can still legally make claims like “conversation data was not used” in the announcement, while also knowing that they are guaranteed to get there if they just pull the lever enough.
(Naturally, I have far too much respect for OpenAI’s legal team to suggest this is what happened in their project.)
Their claim is that training on their solution is "unlikely but possible".
Consider this scenario.
Has a google crawler read my new novel, which I may or may not have posted on my blog, page by page, as I wrote it?
Can you, without knowledge of what I have actually done, claim that the google crawler has not seen the novel?
Without any evidence that I have posted the novel online, it might be tempting to say that the crawler has not seen the novel, but what if I were in an adversarial position against Google on this topic and were challenging them to make that claim. You would wonder if I were hoping Google to overreach by making a definitive claim without taking into account some action that they had no knowledge of. It becomes difficult to use the scientific expression "There is no evidence for this" when there is an accusation of malfeasance because it can be so easily be conflated as "You can't prove we did it". It seems like the best you could say would be 'Unlikely, but possible'
No, they asked one guy to do a joint publish conditioned on leaving the other collaborator out, with veiled threats. The joint publish part smells awfully like admission of guilt given there’s absolutely no reason to do it if you believe you independently arrived at the result using only public prior work. The leaving out collaborator part is outright academic malpractice. Disclosure: I was an academic once.
This would be a new publication with OpenAI's novel solution. Buckmaster and Alpöge did not arrive at the NS solution, they used an approach that also happened to be used in OpenAI's proof.
> there’s absolutely no reason to do it if you believe you independently arrived at the result using only public prior work
Unless the goal is to assuage the existential pain felt by many mathematicians with respect to what may feel like increasingly inconsequential efforts. It also seemed like a way to build good will amongst knowledge workers dealing with similar issues, with the goal of showing they are dedicated to easing the transition for them as society parades into the future. This has clearly backfired.
> The leaving out collaborator part is outright academic malpractice.
They would not be leaving a collaborator out. Again, this would be a new publication. Here's the quote from Buckmaster's post [1]:
"Two proposals were offered to me. The first was that we post our Euler
result, and that OpenAI post its Navier-Stokes result the next day. The second
was that, after posting Euler, I alone write a paper presenting the Navier-Stokes
result, acknowledging that an internal OpenAI model had resolved it."
These are utter bullshit that breaks every norm, but may be convincing enough for some non-academic OpenAI cheerleaders who post dozens of comments on these topics.
> Such intensive use of AI doesn't come cheap. In a post on X, LisanBench, an LLM benchmark evaluator, estimated that the output tokens alone would cost about $6.5 million at OpenAI's average consumer price. Including the far larger volume of input tokens, the post estimated the total could reach $10 million to $40 million.
When I worked at Google, we spent $100M in power on protein folding and drug discovery (this was long before AlphaFold). Never underestimate the willingness of smart rich people to invest in speculative science.
I am using LLMs more and more to help me write docs at work, simply because I do not have time to start these docs from scratch. And it pains me so much that my writing no longer sounds like me.
This is fine at work, I guess. I'm not being asked to sound like myself. I'm being asked to convey an idea to an audience as efficiently as possible, both in terms of my time and theirs.
But I can't shake the feeling that the LLMs are eating my soul.
I still don't use them often to write code. Not that I have issue with those that do, but that sounding like myself in code is a part of my soul that I'm having difficulty giving up.
LLMs are pretty bad at compressing ideas down to the level a smart human can. I think this is quite important for good documentation. Essentially the AI wont do a good job of highlighting the really important and essential details unless you go in and edit the writing by hand afterwards using your human intuition. I am not saying to not use LLMs to write docs, but I would definitely edit to expose the really important details as early as possible in the docs and leave the bulk boring stuff the LLM writes for later.
Yeah, I find it... not incomprehensible, but perhaps deeply misguided the way many people are using LLMs mostly to bulk things up.
1. They generate a lot of fluff that people are accustomed to thinking of as a "sign of work" or "sign of intelligence", but now that counterfeits are easy they are not (should not be) valuable anymore.
2. That fluff (and grammatical annoyances) make readers zone out instead of getting what you want them to get from the document.
3. Much of the document is often an extrapolation from a much smaller set of data like source-code or a spreadsheet or the prompt-stuff. If you really believe the tech will get better, then you should be preserving those higher-truth artifacts instead, and use them to generate an improved extrapolation as-needed.
For sure. But I can definitely produce something good enough to get the job done in a much shorter time by bootstrapping myself from an LLM output and editing from there. (I do edit. Raw LLM output is such a drag to read. Too low signal.)
I don't care if anyone else cares that my code has my voice. My voice doesn't add value to the code. But what other people think of my code is entirely beside the point that I'm trying to make.
To me, code is a creative endeavor. I know that this isn't a universal feeling, but to me the act of creating is more personally valuable than the creation. So when I read code that I wrote (with the help of a tool) that doesn't sound like me, I feel like I have robbed myself of something that I find personally important.
My employer, or anyone else, doesn't care and probably shouldn't care about my experience during the act of creating. The creation is what matters to the external world.
But like, I don't think I'm alone in valuing my own creativity and finding that creativity in code.
(And please don't misunderstand. I'm definitely not arguing that code produced should be "clever" or "creative" in the other sense of the word. I'm just talking about the experience of creativity.)
People who work in English but have a different mother tongue might be able to avoid that unsettling feeling.
For me, writing in Spanish is my natural voice and I would not dream of allowing the output of a LLM to replace it. It just would not be me.
However, writing professional communications in English does not feel the same way. Of course, I strive to be clear and polite, and even try to have my own style, but I remain aware that I am playing a particular role in a specific context in a way that does not happen when I am using my mother tongue.
My advice is to keep writing. Writing is thinking. Write the docs, let the LLM turn it into code.
People have started treating writing as a commodity because generating words is commoditized. But thinking has not been. If anything it’s become more valuable because people are turning off their brains.
As an EM, I know that it’s an investment worth making and encourage ICs accordingly.
I am pretty anti-AI, but for once I am reading someone's complete and balanced take on things. Despite it's usefulness on the surface, I do think the lost of human personalities is tragic. Efficiency aside, it's the only thing that makes anything worthwhile.
I know I'm in the minority, but WFO works great for me.
My office is in walking distance of my home. I get great office perks (lunch, coffee, quality meeting rooms, good desk and workspace). And having a second space for work that is not my home is great for my mental health and WLB.
All of my meetings are still video conferences. Doesn't move the needle for me.
I choose to be in office 5 days a week even though only 2 are required, and it's been great for my productivity.
WFO in a shitty office is still a shitty experience, but an actually good office space is a productivity boost IME.
I got incredibly lucky at my first job. The first office we had was a five minute walk from my apartment. I could literally walk home to grab lunch, or to enjoy a private bathroom break. The second office moved to a 10 minute bike-ride away (5 minutes by car, 20 walking - which wasn't pleasant in the Texas heat), and was on the top floor of an office building, and I got my very own office with a door.
That was nice.
Sitting at a shared desk, one row away from twenty sales people who were on their phone all day, that wasn't.
Gray is not the default color theme on macOS. The default is blue. So the criticism about using right/left to distinguish on/off doesn't really apply. In reality, the use of color disambiguates the on and off states. Just like the physical toggle example it's being compared against.
That's not to say there aren't problems with toggles, but the article's core argument falls apart if you use any color theme other than gray.
> Gray is not the default color theme on macOS. The default is blue. So the criticism about using right/left to distinguish on/off doesn't really apply.
Graphite is one of the official standard configuration settings.
reply