Just imagine how much an authentic keyboard with a Clippy icon would fetch, to the right buyer that is.

We should be basing public policy on facts not marketing language.

Amazon CEO literally told US gov officials that it could be jailbroken and is dangerous. That was a fact (or from the pov of the gov it was). If they were going to ban based on marketing, all LLMs models would have been banned by now.

This assumes that the people in charge for public policy understand the technology beyond the marketing. Which I don’t think is this case based on the track record of policymakers when it comes to essentially any technology.

Why? You can’t yell fire in a crowded place, but “this is going to destroy the world, sign up now” is okay?

Most things should be fact based, but you also should have limits to your marketing

Did Anthropic say "this is going to destroy the world, sign up now"? The details matter quite a lot actually in free speech cases.

Should a nuclear energy startup that says "nuclear weapons are very dangerous and should be regulated" be liable to have its assets frozen?

22 year old single me would have enjoyed this little adventure.

It's one of the most physically though jobs there is, a great challenge for a young man wanting it.

NGOs are not lobbyists.

Some NGOs lobby/employ lobbyists.

That’s not the same thing.

I’m working on a DaaS startup (drone as a service) so maybe I’ll locate next to you and your customers can blow drones out of the sky and mine will keep having to replace theirs.

I like where you are going with this. Can you offer “hardened” drones for the advanced players?

Lots of AWS’s control surfaces are in us-east-1, and (not calling out any specific instances here) sometimes what’s call an “AWS outage” especially regarding us-east-1, is actually a limitation on accessing those control surfaces, ie, making changes to assets that are actually hosted elsewhere.

In such cases the services continue to operate as-is despite problems in us-east-1.

Not saying that’s not a problem, just, clarifying the scope.

The spa approach also achieves what’s probably the key ingredient to making this useful medically — consistent data over time. We know the “what’s this fuzzy bit!?!” hysteria with elective scans can be counterproductive, but if you truly had monthly scans going back 2 years, the fuzzy bit could in fact mean something.

The sci-fi writes itself. Someone hacks the spa’s schedule to find out when you’re getting your scan done, then hacks the machine to push the output to 11 when you’re in, and liquifies your insides.

The first Midjourney Medical murder.

It’s also possible to not really hold that strong an opinion on things. Not everything is a pitched battle where doing what your employer wants means you’ve sacrificed your integrity.

Aren’t human coders non-deterministic? There’s no guarantee two people with otherwise identical levels of experience will always write identical code.

Any software engineering practice that had enough review and feedback to work with humans should work more or less the same with AI coders.

It’s when someone tries replacing an entire team or an entire process with a single prompt that they get in trouble.

TBD on if the calculator can properly review and participate in the feedback loop with itself.

They also don't learn, so they never get less unpredictable. You can't give the senior robot the production keys and expect it won't delete prod.

>Aren’t human coders non-deterministic?

Sure, but LLMs are non-deterministic in ways that no sane human ever would be. See the "Is it better to drive or walk to the carwash" scenario from a few months ago as one of many, many examples. Or a personal example I encountered just a week ago: I asked Claude (Opus 4.8 in case any of the "you aren't using the latest model that totally fixes that issue" types are interested) to convert a bunch of DB calls that currently use raw ADO.NET calls to use Dapper instead.

The projects in this repo were on .NET 4.8.1 and were still using the older format for the .csproj file instead of the newer (and far better) "SDK-style" format that Microsoft introduced a few years ago. It tried to use the dotnet CLI to add references to Dapper, even though the older format of .csproj doesn't work with that. The dotnet CLI returned errors about trying to add the package references for Dapper, which Claude completely ignored while continuing to try and convert the ADO.NET calls to Dapper. And at the end it tried building the project, which of course failed, and then it confidently informed me that the conversion had been completed successfully and that the build completed successfully and all tests were passing successfully, even though the output from the build it had done immediately prior clearly told the LLM otherwise.

A real human, despite being non-deterministic, would have caught the issue at multiple stages. They would have seen the error when trying to add the reference. If they ignored that then they would have seen the red squiggly lines all over the (deterministic) IDE telling them there was something wrong, along with autocomplete for Dapper calls not working. And if they continued to ignore those and managed to keep going anyways, they would have clearly seen that the build failed, with tons of errors specifically about references to Dapper failing to resolve. An LLM keeps going on its merry way in ways that effectively 0 humans would.