> The argument that training AI on scraped data is "fair use" and the resulting model outputs are "transformative works" has held up in courts.
Nope. Nope. Nope. That has explicitly not been ruled on yet. Transformative means that you don't need a fair use defense. Anthropic has only gotten away with their outputs being called transformative so far because they put a dubiously effective filter in front to block the most egregious infringing outputs. No one has actually challenged this afaik.
Under that logic is a free trial a bait and switch? How about a 1-month free deal? How about what Adobe (and many others) do where they license to the school and students get it free until they graduate?
It seems like a really weird point to make, when you could just as easily argue that Figma giving their services for free to students is a gift that levels the playing field, by allowing students without means to gain experience with industry standard tools they might not have been exposed to otherwise.
> Under that logic is a free trial a bait and switch? How about a 1-month free deal? How about what Adobe (and many others) do where they license to the school and students get it free until they graduate?
No. The key difference being transparency. You know when signing up for a free trial what the actual long term costs will be and can plan for it.
We might be talking about different things. I was mostly replying to this line from the OP:
> But I think it's part of a larger mistake Figma is making: they seem to have shifted to an extraction mindset too early
I’m not sure if this was just awkward wording that seems to condone these type of strategies.
All these loss leading, vendor lockin strategies have distorted markets heavily. Complex tools cost a lot of money to develop; and if another player is just going to burn piles of cash from elsewhere to undercut you, it becomes a game of capital allocation and not individual product quality/costs. It’s terrible for consumers and a big reason why even common chat apps are barely functional.
That's fair. I'm also heavily opposed to VC-funded, market-distorting behaviors and the later extraction-oriented outcomes they produce. In this case I was framing it in terms that might be more widely received by folks who aren't, and pointing out that, even if that was their mindset and goal, they were still making a mistake strategically.
But I appreciate the reminder to not cede ground in wording, thanks.
> Tell me you know nothing about web development without saying you know nothing about web dev
This Twitterism really bugs me.
You took the time to write a really detailed response (much appreciated, you convinced me). There’s no need to explicitly dunk on the OP. Though if you really want to be a little mean (a little bit is fair imo), I think it should be closer to level of creativity of the rest of your comment. Call them ignorant and say you can’t take them seriously or something. The twitterism wouldn’t really stand on its own as a comment.
It bugs me that the author is "dunking on" React without knowledge on the matter (React is the tool you use to enforce consistent UI on a site; it has almost nothing at all to do with a design decision to have inconsistent UI). So I guess I "dunked on him" in response.
But ... too wrongs don't make a right. I'd remove the un-needed smarminess, if it wasn't already too late to edit.
I’m a decade+ linux power user and I still do insane things like pipe outputs into vim so I can copy paste without having to remember tmux copy paste modes when I have vertical panes open.
Sounds promising honestly. One of the scariest parts of the big AI labs is all of the exclusive training data they get through their UIs. (It’s unclear whether distillation is a feasible way to close the gap).
If there were another party involved, that would (hopefully) diversify power that (potentially) comes with those streams of data.
It’s a bit ironic that the USA has mostly abandoned interoperability after being one of the pioneers with the American manufacturing method. [0]
> Across a thousand runs through our scaffold, the total cost was under $20,000
Lots of questions about the $20k. Is that raw electricity costs, subsidized user token costs? If so, the actual costs to run these sorts of tasks sustainably could be something like $200k. Even at $50k, a FreeBSD DoS is not an extremely competitive price. That's like 2-4mo of labor.
Don't get me wrong, I think this seems like a great use for LLMs. It intuitively feels like a much more powerful form of white box fuzzing that used techniques like symbolic execution to try to guide execution contexts to more important code paths.
I’m falling into the Socratic hole [0], but in a modern civil society there is a justice system through which people seek recourse. This has all sorts of desirable effects for societies.
Please educate yourself on the basics or at least put more effort in before participating in conversations.
[0]: It’s easy to abuse the Socratic method and devolve a discussion into one of first principles. It’s extremely tiresome and a huge waste of everyone’s time.
I'm a big fan of the justice system. Can't have a functioning civilization without it. And yes, violence that is used by a democratic society following regulations is generally speaking better for society than arbitrary vigilantism motivated by personal beliefs is. But I'm not arguing that it would necessarily be good to kill Sam Altman. I'm just arguing that it's ok to find the idea of his death pleasing. I find the idea of killing all sorts of people pleasing without necessarily thinking that actually doing it would be good for society overall.
I've worked in the system for decade now. and I cannot agree. I feel nothing but regret, shame, and guilt most days. It's a cruel and vindictive system. Lady Justice carries a sword for a reason, and she loves to swing it.
I commonly refer to our system as the legal system for there is little justice.
I concur. I think the true issue is that no system can solve these types of problems. There will always be people who benefit more than others, and there will always be people who slip through the cracks.
I think our system is not the worst system available by any means. I just wish there was a bit more focus on impartiality and rehabilitation. I am not so sure why there is an obsession with punishment when data suggests it does not really deter people.
No can do, this justice system actually protects war criminals rather than prosecuting them. The US threatened the international justice system by threatening to invade the Hague when it attempted to prosecute American war criminals. It's contradictory to respect the American “justice system” whilest it actively disrespects other justice systems both in other countries and in international law.
I intentionally said “modern civil society” instead of the USA to avoid talking about specifics.
Whether the USA has a sufficiently functional justice system is another topic. My intuition is also that, in the presence of a disfunctional social system, fixing (or replacing) the system will usually lead to better outcomes than side stepping it. Not that I really want to talk about the minutia and challenges of fixing the USA’s justice system.
Yeah a company causing mass death or other disasters is maybe the single clearest signal that they should go bankrupt and someone else should take over (if the tech is really that important).
“Very likely yes”, I reply to an account that <1yr old with mostly comments in AI topics many of which violate the HN guidelines (including the one I’m responding to).
Strange gatekeeping response. Yep i comment on topics i'm interested in. Forgive me for not being on the platform for more than a year yet. That's a cute attitude
Nope, there is no “The” Turing Test. Go read his original paper before parroting pop sci nonsense.
The Turing test paper proposes an adversarial game to deduce if the interviewee is human. It’s extremely well thought out. Seriously, read it. Turing mentions that he’d wager something like 70% of unprepared humans wouldn’t be able to correctly discern in the near future. He never claims there to be a definitive test that establishes sentience.
Turing may have won that wager (impressive), but there are clear tells similar to the “how many the r’s are in strawberries?” that an informed interrogator could reliably exploit.
Nope. Nope. Nope. That has explicitly not been ruled on yet. Transformative means that you don't need a fair use defense. Anthropic has only gotten away with their outputs being called transformative so far because they put a dubiously effective filter in front to block the most egregious infringing outputs. No one has actually challenged this afaik.
reply