Please create an account to participate in the Slashdot moderation system

 



Forgot your password?
typodupeerror

Comment Re:Unlikely. (Score 1) 43

I was thinking the same thing, but kind of hoping they get the reactors going before a sudden outbreak of common sense shows that throwing 6GW of electricity at GPUs so that people can get worse search results and shitty generated code is stupid.

They can shut down all the GPUs, but leave the reactors running - we need the energy.

Comment Re:Fine (Score 1) 130

Sorry, you're going to have to back that up with some kind of evidence.

Please show where Nikon, Sony, Fujifilm, Canon, or any serious company that makes cameras has had US manufacturing of their products in the last 40 years.

Here's a hint: some of their stuff was never manufactured in the US, because these are predominantly Japanese companies.

Comment Re: Garbage In / Garbage Out (Score 1) 77

That "people doing what they've been doing before" argument is a hot mess. It doesn't even make any sense. Using LLM tools to produce code is fundamentally nothing like programming yourself. If AI naysayers are trying out LLM tools, even if with a preformed opinion and expectation, they're not sticking to their learned processes. That's not even possible.

What good developers do as they did before is proper specification, clean architecture, considering edge cases, building robustly, recognising and mitigating security issues, ensuring consistency, maintaining extensibility, etc. etc. etc. And if you do any of these things, you'll very quickly realise how pathetically, laughably limited the actual usefulness of LLMs for programming is. LLMs help you with *none* of those things, but they're the core of what software engineering does and is required to do. "Coding" is, what, 10% of the job? LLM are a (sometimes very poor) tool to help with a part of the job that was never the core, never the bottleneck, never the cost driver, and never the main problem.

Comment Re: Consider the source (Score 1) 77

Except that on Stack Overflow, more often than not the shit responses are followed by comments explaining why they're shit. That's helpful information, and you might even learn something.

Crappy LLM slop code comes with no such disclaimers, so you can either trust it and end up with a crap application 95% of the time, or research whether the solution holds together, negating any efficiency gain you might have thought the LLM gives you. In addition, the LLM doesn't learn how to create less crap code in the future, and neither are you.

Comment Re:Fine (Score 5, Interesting) 130

Well, how about this?

When your arguments in appeals court are met with judges giving comments like "“IEEPA doesn’t even mention the word ‘tariffs’ anywhere” when you are specifically arguing that the authority for enacting the tariffs comes from the IEEPA, it's not so good for your argument.

There is no constitutional authority for the President to unilaterally enact tariffs. Period. Not under that law, and not under any other law Congress has passed, that they have named as what allows them to do this.

That's why I think the tariffs will disappear.

Also: Democrats are *not* fine with them. That's some baseless shit you made up.

Slashdot Top Deals

The IBM 2250 is impressive ... if you compare it with a system selling for a tenth its price. -- D. Cohen

Working...