> It is hard to predict exactly how much this will accelerate AI progress, but there is a real risk that capability development rapidly accelerates beyond our ability to understand or control the resulting systems.
This is just the standard AI "doomer" talking point. The dangers of AI just used as a compliment sandwich. "AI is so cool and awesome and that's why we need to slow down."
The real risks are much more boring and much less considered. For just one example:
AI is set to completely upset the Software Development industry. There are 1 to 3 million software developers in the US, depending on how you draw the line exactly. A low-few hundred thousand of those are visa workers.
With AI set to significantly cut into tech employment, what's going to happen to those visa workers? ... The Trump admin is not particularly subtle, even openly talking about retroactively revoking birthright citizenship.
AI & Tech companies love to talk about how their AI is going to hack everything if they're not careful. But not a peep about these ethics concerns.
Even if you [the reader] consider deporting visa workers the ethical move if the amount of jobs shrink, that's not something to be done unprepared or out of nowhere.
> so from my perspective I don't see why we should be skeptical that frontier LLMs guided by experienced devs can translate whole libraries like that
There's a laundry list of reasons to be skeptical about LLM results in general. The biggest one is that LLMs are excellent at creating output that looks right but is nevertheless still wrong.
In this particular case, there's also a specific reason: This is obviously a marketing stunt regardless of whether the claimed results hold up. Anthropic has a long history of just lying and making shit up, so any results are subject to extra scrutiny.
> but why isn't it a regulation or requirements issue?
It's America. They don't do regulations there.
The more serious answer is that this political movement is leading to more strict regulation being enacted, it just wasn't in place because of reason #1 and the problems being much less severe.
Additionally, the tech industry is notoriously hostile to any kind of regulation. Especially now with AI. Just look at how many people in this thread will insist that it's "Chinese Propaganda", despite clear evidence that this is just a response to the actual problems datacenters cause. One news crew interviewing people with jet turbine noise will do more than a million of Beijing's finest trolls.
> my company is spending thousands on Anthropic a month
The problem is that this is simply not enough. They need you to spend tens of thousands, probably closer to hundreds of thousands, before the numbers start making sense.
> At minimum, my company is not part of this circular thing.
You're in the blast radius. And if you don't have a plan for "what if Anthropic hikes the API rates by 10x or worse", you're in the kill zone.
Anthropic and OAI are both looking for 1T valuations. Anthropic is projected* to make $500mil operating profit off 10bil annualized revenue. They need to grow these numbers 10-50x more for the valuation they're seeking to make sense.
Tough pill to swallow given they lack a moat and their compute is being subsidized by the companies they lease it from, all of which are INVESTED in Anthropic and have a desire for their growth story to look good because when they IPO it gives those same investors a better shot at making their money back.
That's the circular aspect of this whole scheme. Nobody makes their money back if the LLM company valuations get more realistic.
This is just the standard AI "doomer" talking point. The dangers of AI just used as a compliment sandwich. "AI is so cool and awesome and that's why we need to slow down."
The real risks are much more boring and much less considered. For just one example:
AI is set to completely upset the Software Development industry. There are 1 to 3 million software developers in the US, depending on how you draw the line exactly. A low-few hundred thousand of those are visa workers.
With AI set to significantly cut into tech employment, what's going to happen to those visa workers? ... The Trump admin is not particularly subtle, even openly talking about retroactively revoking birthright citizenship.
AI & Tech companies love to talk about how their AI is going to hack everything if they're not careful. But not a peep about these ethics concerns.
Even if you [the reader] consider deporting visa workers the ethical move if the amount of jobs shrink, that's not something to be done unprepared or out of nowhere.