FaceDeer

It’s almost as if this were a community where we discussed new technologies that are just emerging and haven’t become well-established yet.

An equally-true headline: “At last, a promising use for AI agents: debugging smart contract code.” The availability of this tool should make smart contracts more secure in the future, and cryptocurrency more reliable as a result.

Really, you think all existing uses of data centers stopped now that there’s AI in the mix? There may be specific facilities under construction that are intended primarily or solely for AI use, but all the existing demand is still there.

PJM has lost more than 5.6 net gigawatts in the last decade as power plants shut faster than new ones enter service, according to a PJM presentation filed with regulators this year. PJM added about 5 gigawatts of power-generating capacity in 2024, fewer than smaller grids in California and Texas. Meanwhile, data center demand is surging. By 2030, PJM expects 32 gigawatts of increased demand on its system, with all but two of those gigawatts coming from data centers.

So this is a combination of utter mismanagement by the power companies, combined with growth in data center demand. Data centers are not purely AI. And I would expect that if PJM continues to be a basket case with exceptionally high prices those data centers will move elsewhere, or at least not get set up so more in those locations. Data centers generally don’t have to be located in specific places, by their nature. AI-specific ones in particular since the bandwidth required is a lot smaller than their processing power.

Clearly what we need to do is ban plastic straws some more.

The image’s caption:

Bret Adee, one of the largest US beekeepers, has 55,000 hives used to pollinate crops. As monoculture farming spreads, bees have to be moved across the country, especially for almonds, blueberries and cherries. He lost 75% of his bees over the past year. Photograph: K Brinson/New York Times/Redux/eyevine

A quick Google search reveals that his operation, Adee Honey Farms, is in South Dakota. here’s another article on this subject with a different photograph of the same place. Their web page doesn’t have many photos of the field in question, but there’s a couple.

This is not an AI image.

Some years ago now, a bunch of bike lanes got added to the streets in my city. The city did a big project of adding them and afterwards proudly declared that X number of kilometers of bike lanes had been made.

When an investigation was done into how the decision process had gone for where to add them it turned out that the only consideration had been “how cheap is it to add bike lanes in these locations?” Not “would bike lanes actually be used in these locations?” They were solely trying to maximize the kilometers-of-lane-per-dollar-spent so that they could put out that headline with as big a number as possible.

Subsequent studies showed that a lot of those lanes weren’t being used by bikes in any significant number. Bike lanes had been added on streets that ran alongside sidewalks that were already designated bike paths. I’m a bike rider myself, some lanes were added in my neighborhood but they somehow managed to put them everywhere except the routes I usually took. The city wound up spending a bunch more money to remove a bunch of the bike lanes that were doing nothing but increasing congestion.

It may be that this was a similar situation, where someone wanted to proudly show off headlines of how they’d pushed for bike access and got X numbers of kilometers installed and those were the only real metrics that mattered.

I can’t say anything about your math, but I can say that you didn’t read the article.

I’m not trying to argue for or against this position. As I said all I’m doing is explaining a misrepresentation of the position that people are holding, namely that “a machine that can’t think straight will do it for us.”

I don’t see what it’s going to conclude that we haven’t already.

Well, that’s the point of trying to build ASI. To have it think of things that we haven’t been able to think of.

I really, really, can’t picture a scenario where we actually listen.

Of course not, you’re not an ASI.

That’s fine, I’m just correcting the misrepresentation of the view that was in the headline.

Nobody’s expecting a “machine that can’t think straight” to do it. Some people are hoping that a more competent machine will be developed.

Maybe this is because China just said they needed Russia to win? Trump’s got an obsession with opposing China.

Or maybe it’s just the way the mush inside Trump’s skull happened to flow this particular day.

Now make the exact same meme but substitute “AI training” for “piracy” and watch the downvotes flow in.

Sure, not disputing that. I’m more annoyed by the double standard regarding his successful decisions.

No, just surprised about how uninformed and knee-jerk those opinions are.

In my experience, it’s likely that some of those downvotes come from reflexive “AI bad! How dare you say AI good!” Reactions, not anything specific to mental health. For a community called “technology” there’s a pretty strong anti-AI bubble going on here.

What I mean is that when Musk-owned companies have successes people are very often quick to accuse him of “just hiring smart people” or “just buying a successful company.” It’s only when those companies have failures that he gets credit for being hands-on in their design decisions.

Don’t get me wrong, I think Elon Musk is a pretty terrible person both in terms of his personality and his politics. But pretty terrible people can nevertheless be smart and make good engineering decisions. Just look at von Braun as a prime example.

Always interesting to see the view of the degree of Elon Musk’s involvement in his companies’ decisions swing depending on whether the outcome is good or bad.

They are using them, however. They’re visiting websites with them, using apps with them, and so forth.