treszkai

At the same time, the original Phorion post says,

> Before appending, the script checks whether svc-telemetry is already in the file rather than checking for the key itself. When hunting for this activity, it’s worth checking the public-key fingerprint rather than relying solely on the comment string

Except customer demand doesn't follow the fixed supply of GPU compute that any company has, so they either a) rent compute on demand, b) sell their excess compute, c) let it sit idle and maybe an employee claims it.

Because it's a poorly written article.

uhm, spoiler alert?

Except in the case of nuclear weapons, humans have their hands on the launch codes, and the nukes are not going to launch themselves. In the case of extremely capable AI, nobody has their hands anywhere close, because nobody is able to – nor willing to – supervise a thousand agents running in parallel, when they usually get shit done. Except eventually they are so good at getting things done that they disregard the values of the human asking them the request, such as "do not hack other corporations". Interpreting human intent precisely is an extremely hard problem, especially in the face of something 1000x faster than you operating autonomously.

> the HF hack was partly the result of a training algorithm that incentivized goal completion as the highest priority, and let them run endlessly in an unmonitored sandbox with weak security

I don't mean to be a smart-ass, just emphasize the fatality of this: goal completion will always be the highest priority, and even if one day it takes second place on certain deployments to "human values" or whatever, there's no way to guarantee at the moment that it will be so on every deployment of highly capable models across the globe. Same goes for running AIs in an unmonitored sandbox with weak security.

The number of implementations and the lack of success suggest that it is a genuinely hard problem for humans, and "agents" are not going to magically make it simpler. Unlikely that humans abandoned their projects (or failed to band together) solely because of their desire for code quality.

It sounds like AA scraping Spotify was an unnecessarily risky move in terms of legal consequences and the toes they step on (who might be behind the downtimes). And it's not even fully in line with their aim of cataloging all the books in existence: while music could be argued is "information", one's sense of IP is much stronger with pirated music than with pirated books.

How on earth are they able to cash in on that? Surely they are not ordering neither air fryers nor their servers from Amazon.. Or anything for that matter, given the illegality of AA. Do they resell them at 90% on eBay?

Sorry I'm out of the loop: how did Spotify use AA or other pirated data?

But you're in the air fryer production business, not in the food video production biz (which is really saturated). So you likely don't have the skills to launch a food porn channel organically, so you have to pay a youtuber, which is not the worst, but the yield would be unpredictable (and more costly to test). Sounds overall better than Amazon ads, though.

> Your access to this resource has been blocked.

Do you have VPN enabled?

To roll with this idea: soon the credits will include more names like Sol and Opus 5 and various pointy-haired bosses than names like Mike Little. That's not very encouraging either.

And similarly, humans are not Turing-complete either: none of us can even in theory sit down and store a trillion bits of information in their mind (or paper, whatever) without loss and without a computer.

So Turing-completeness is not even a necessary condition for intelligence.