Forgot your password?
typodupeerror

Comment Re:Bye bye Wikipedia (Score 1) 30

Here's a case of a very experienced journalist getting caught by including made-up quotes that had been hallucinated by the AI he'd used to summarize research information: https://www.theguardian.com/te...

Vandermeersch added: “It is particularly painful that I made precisely the mistake I have repeatedly warned colleagues about: these language models are so good that they produce irresistible quotes you are tempted to use as an author. Of course, I should have verified them. The necessary ‘human oversight’, which I consistently advocate, fell short.”

When even experienced journalists fail to find AI hallucinations, you really can't expect unpaid volunteers to do better.

Comment Re:Bye bye Wikipedia (Score 4, Insightful) 30

Wikipedia is choosing to die. There is a lot wrong with a lot of what people are doing with GenAI but it is also super useful.

Unfortunately, even the best LLMs sometimes make up information ("hallucinate"), and the stuff they make up is deliberately crafted to appear exactly like real information. This is simply unacceptable for an encyclopedia.

If Wikipedia were written by paid professionals, you could plausibly put in place protocols to check and verify, and fire the ones who fail to check properly, but even paid professionals have been seen to let hallucinations through. As it is, as an encyclopedia that it is put together by volunteers, forbidding AI is pretty much a forced choice.

https://www.evidentlyai.com/bl...
  https://arize.com/llm-hallucin...
  https://thisweekinsciencenews....

Comment Re:So this is about AI slop spam from clankers (Score 0) 74

You aren't that interesting either and yet you found a way to look interesting with sock puppet accounts, maybe they'll just do the same.

The problem is that a sock puppet uses an actual human to operate it, while bots run with little or no need for human effort. So a human can turn out a handful of posts with sock puppets, while bots can be churn out thousands and thousands of posts, swamping any system with slop.

Comment Re:They're catching up to us! (Score 2) 23

Wikipedia was about half hallucination anyway.

Let's check that. Here is the link that takes you to a random Wikipedia article: https://en.wikipedia.org/wiki/...

How many times do you have to click that link to find an example of the hallucinations that you claim are "about half" of Wikipedia?

Comment Economy & safety are reasonable points to disc (Score 2) 76

SMRs will never be economical, and are no safer than current reactors. Maybe less safe, in fact.

Not sure why this is moderated troll; this may be opinions that may be debated, but are quite reasonable to bring up.

There's a pretty good argument that small reactors won't be as low cost as large ones, on a dollars per watt basis. Basically, the scaling laws for nuclear power favor high power-- it's not ten times cheaper to produce one tenth the power. The people arguing for small reactors suggest that they can manufacture them in large quantities, and get an economy of scale in that manufacture, but this is a projection unsupported by anything but hand-waving.

Safety is another issue. There exist much safer reactor designs. I don't know enough about the small reactor designs to comment here. Would have been good to have cited some links specifically about small reactor safety issues.

Comment Re:MacMann aka blindseer will no doubt be delighte (Score 2) 76

None of this, of course, addresses the original premise that the US should have built more of these reactors before engaging in any more oil-related conflicts. I mean, the US probably should not get involved in so many oil-related conflicts, but it is dubious for a number of reasons that earlier investment in these reactors would have negated the need.

At the moment, nuclear power plants produce energy for electrical applications, while oil produces energy for transportation applications. These are not interchangeable, so, no, building more reactors wouldn't have negated the need for oil.

There is a drive to move transportation to being powered by electrical energy, rather than oil (there's also a drive, funded by oil company profits, to delay or kill this shift), but it will still be quite a bit longer before that displaces a significant amount of oil use.

Oil is used for other things, of course, but while applications like heating can switch to electrical power, this also takes time.

Comment Missing text (Score 4, Informative) 62

The summary copied the text of the article, but left out the second half of the article, which is the important part:

It’s not clear, however, that the effort will get new supplies of electricity built quickly enough to ease pressure on grids, said Jon Gordon, who is a senior director at Advanced Energy United, a clean energy trade group that includes some datacenters. That’s in part due to Trump’s policy focus on increasing natural gas and other fossil fuel-fired power for datacenters, instead of quicker-build sources like solar and wind, he added. “The real problem is the inability to get generation online fast enough to meet the datacenter demand,” Gordon said. “Hyperscalers paying for the generation doesn’t get it online any faster.“

You can't build new power plants immediately. Companies saying they'll "pay for new power generation and grid upgrades" (even if they actually do this, instead of completely ignoring this non-binding "pledge") some time in the future won't help the fact that the data centers consume power now/

Slashdot Top Deals

I have a theory that it's impossible to prove anything, but I can't prove it.

Working...