Whether covered under fair use or not, the laws around copyright today did not anticipate this use case. Congress should pass laws that clarify how data is and isn’t allowed to be used in training AI models, how creators should or shouldn’t be compensated, etc - rather than speculating whether this usage technically does or doesn’t comply with the law as-is.
I think what really sizzles me is that some of these same companies helped develop such strict enhancements to copyright to begin with in the realm of software. So I'm not falling for the crocodile tears when they get caught in the very snare they used to litigate thousands of other companies for and bully potentially millions more with just because now it's more profitable to tear it down. Made your bed...
And yes, regardless of results I agree there should be new laws made. But we know Congress in the US this year has been a roller coaster, to put it lightly. And I don't even think this is top 5 of what congress needed to codify into law properly. So all the short term work will be the judicial branch interpreting what few laws we do have.
> I think what really sizzles me is that some of these same companies helped develop such strict enhancements to copyright to begin with in the realm of software.
Create a private right of action. If creator A can show that AI trainer B used their works (e.g. like how we've seen Getty watermarks show up in AI generated pics), then they can sue for $X dollars.
I make about $4,000 per million streams on Spotify for the tracks I’ve released independently. For label releases I make less, but the label promotes them so that sometimes results in more net revenue. I have a bit over 10M Spotify streams over the last 3 years.
Also, Spotify promotes my music via editorial playlists and algorithmic (eg Radio or Discover Weekly), so I’m probably making a lot more total revenue than I would have on iTunes.
The richest person in the world, Elon Musk, is a climate entrepreneur who got there in part due to climate-driven government subsidies. Just because someone made money off it though does not mean that climate change is a fake concern.
> richest person in the world, Elon Musk, is a climate entrepreneur who got there in part due to climate-driven government subsidies. Just because someone made money off it though does not mean that climate change is a fake concern.
Nobody said this. If the only person arguing the dangers of climate change was Elon Musk, there would be room for reasonable skepticism. That's the difference between the AI debate and "any expert, scientist, or company executive who says 'this stuff could be dangerous'."
And similarly to climate, many people who signed this letter are academics who do not appear to have any financial incentive to push for government regulation.
I don't think that's a difference. The open letter the source article is criticizing is signed by a pretty wide variety of experts, scientists, and company executives.
(Airplane founder here) Airplane isn't YC backed. Though interestingly my prior startup, Heap, went through YC and has tons of YC-backed competitors (Amplitude, Mixpanel, Posthog, etc).
Personally I like that YC remains agnostic to the ideas and is willing to back competitors because it ultimately means more great startups get funded. Later-stage investors care more about conflicts because being involved at the level of taking a board seat matters a lot more for conflicts.
At this point they've backed 1000s of companies; if they had to vet that entire list for conflicts to back their next batch it would be incredibly difficult. Also, given the stage they're investing at, tons of companies pivot and end up competing even if they didn't start out that way.
Not sure if this is what you're implying, but I think it's a mistake to think of YC as a monolithic organization that makes decisions by saying, "idea X is good, we should fund teams doing it."
More likely, each of the teams doing each of these startups interviewed with completely different partners who had no idea of the other startups even existing, and in that interview, they thought the founders seemed solid and had thought through their idea well, and chose to fund it. It's even possible some of the people doing these ideas came up with the idea after they got into YC (i.e. they pivoted) - some of the most successful YC startups were companies that pivoted mid-batch (e.g. Brex).
In general YC doesn't want multiple shots on goal in a specific market area. They want as many shots on goal as possible among great founders in general.
LangChain is especially slapped together. Not something I would use in production (especially with the eval bug that was left unpatched for weeks despite an open PR)
It feels at some point being too slapped together isn’t correlated with success
Wishing the LangChain team well! But there are already alternatives cropping up that are far more polished (Microsoft’s Semantic Kernel being one)
Yes, Cowen argues against the idea we can anticipate consequences of technological change, and the specific consequence he focuses on is the idea of existential risk stemming from AI. He says because technology is unpredictable, we shouldn’t try to predict the type of risk imposed by AI, and we should mostly just accept that this change will happen and cope with it afterward. This stance is what I was arguing against in the post.
> But to wish to halt AI advancement requires an unhealthy mix of pessimism and overconfidence in your predictive powers.
That’s perfectly valid! No one is obligated to respond or engage with any specific argument. But my point was that if you do choose to engage, saying “the world didn’t end before so it will be fine now” is invalid.
It was valid in every other instance so what makes this one different?
It’s almost like humans have some built-in tendency to step back from the edge and not cause their own extinction.
And no, AI isn’t the same as an asteroid barreling towards the earth where you can point to it and also reference the dinosaurs and say things might not turn out so good this time.
Thanks for taking the time to read the article and comment. Appreciate your feedback. As you point out, my last couple paragraphs were somewhat speculative and handwav-y. Do you have an alternative viewpoint on what allows LLMs to be able to somewhat accurately answer complicated math questions, despite lacking an explicitly programmed math solver? It sounds like you may be better informed than me–would love to hear your thoughts.
> that the author clearly didn't read. I guess there's too many scary maths for a "layman".
No need for the personal attack. I did read the paper and the math in the paper is not particularly complicated.
Well, that's awkward. I didn't realise you were on HN. I'm sorry for the
personal tone of my comment. You are right that it was uncalled for.
The paper you linked is clear on the scope of its proofs and in any case it's a
very big assumption to say that "neural nets are Turing complete", when there
are scant few such proofs, compared with the large number of different
architectures (for most of which, no careful investigation of their
computational capabilities is ever done anyway).
You could add a clarification to your article.
>> Do you have an alternative viewpoint on what allows LLMs to be able to
somewhat accurately answer complicated math questions, despite lacking an
explicitly programmed math solver?
Yes, it's because they're language models. In particular, they're very powerful,
very smooth (in the statistical sense) language models trained to represent
gigantic text corpora. Their ability to produce correct answers once in a while
is not a surprise and does not need any other explanation.
Predicting what a language model (big or small) will output is another matter,
so one particular instance of generated output might be surprising in the sense
that the user won't expect it - not in the sense that the model shouldn't be
able to produce it.
In any case, it's clear that the performance of those models depends on the
prompts. Change the prompt slightly and you get a different answer, to any
question. That suggests retrieval from memory (modulo stochasticity) much more
than it suggests computation. And we know that these models are not models of
computation, so there's no question what's really going on.
When I say "retrieval from memory" I don't mean that these models memorise whole
sequences of tokens verbatim. To make a very big fudge about it, it's as if
they've memorised templates that they can then apply to questions to generate
the right answers.
I guess that still sounds magickal and mysterious if one hasn't worked with
language models before, so all I can say is, if you are really curious, and
really want to understand the specifics, you should try to learn more about
language models.
Those are rather "wax-on, wax-off", but if you want to learn Karate, that's where
to begin. Then you can go on to beat up the Transformers and win the girl.
The Charniak book in particular is small and sweet and easy to read. Start
there.
