hatwd

So is the idea here that this'll make it cheaper to fine-tune models than by using reinforcement learning?

Per the post, articulation of the desired results is the only thing AI-assisted development has in common with "management". This is only about 20-30% of what constitutes "management" though.

Even if you're running fleets of agents coordinating other agents, you're still most certainly not a "manager". You're just constraining and launching more complex, long-running software generation trajectories.

> What were you doing with the computer prior to managing it?

In the most literal sense, "instructing it".

This would only be acceptable if the ebooks cost 5 to 10% of what the physical book costs.

Interesting language - its syntax looks like a mix of Rust and Go, with a few of its own idiosyncrasies to distinguish it from those languages.

> Carbon is a concentrated experimental effort to develop tooling that will facilitate automated large-scale long-term migrations of existing C++ code to a modern, well-annotated programming language with a modern, transparent process of evolution and governance model.

This is probably where Go and Rust fail to be C/C++ "successor" languages, as interop between those languages doesn't seem to be as seamless as Carbon aims to be.

Will keep an eye on its development!

I totally get the spy agencies' "moral flexibility" requirements, as I've heard it put.

From what I understand, the spy agencies have ways of obtaining your private information that don't necessarily involve blanket requirements to access all users' data (e.g. creative ways of injecting malware into specific people's devices). But those approaches don't scale, of course. And they shouldn't need to.

Is it any different with Android phones? From what I've read it doesn't seem so.

My comment applies just as much to the people working at Apple and Google as to the folks in the UK government.

It's baffling to me that any sane, healthy person would advocate for invasion of not just one person's privacy (in the case of known or highly suspected criminal activity), but a whole country's people's privacy. (In this case, at least, the privacy of all Apple users in the UK.)

Where does this problem start? Is it a basic education thing that valuing one's own and others' privacy needs to be taught to kids from a young age?

For instance, in the meetings in which these ideas are proposed, why are they not considered a serious, fireable offence, like bringing up racist or sexist comments?

On sites that serve user-generated content, how would one technically detect if one's ads are being served alongside CSAM? Wouldn't the ad source need to be able to fully read the contents of the page itself?

Are there good technical solutions to that problem? And whose responsibility is it to solve - the ad source, or the platform through which CSAM is being served?

Definitely something that needs solving, but sounds really hard to solve - unless you just don't serve ads on that site at all.

Thanks for the recommendation! What was the coldest temperature it was able to withstand?