Solutions starting with "everybody in society must" have a long and celebrated tradition of going immediately (and often horrifically) pear-shaped, as it inevitably turns out that most of everybody doesn't want to, and therefore won't, and in many cases, can't.

For examples, see the Soviet Union's Communism, China's Great Leap Forward, the Khmer Rouge's agricultural collectivism, North Korea's juche, etc.

... this year. And next year, the same logic will have us getting rid of all of the entry-level employees, and the year after that, the mid-level employees, and some years from now, all employees. Yay, progress?

Comedians will do anything that works to get a laugh, but sourcing jokes from ChatGPT (or similar) is not an effective way to get a laugh. Comedy is based on surprise, and LLMs are based on summarizing old material, so there's a bit of a mismatch there.

Well, sure, in the same way a 747 is just a fancy mechanical bird. Which is to say, yes, but no.

Because people aren't solitary islands whose deaths harm only themselves. Anyone who has lived on Earth for more than a few days has formed relationships with other people, and suicide is the murder of a person those people have a relationship with. Suicide harms everyone who interacts with or depends on that suicider for anything, in much the same way that the murder of that person would.

Yes, that is precisely the problem.

Perl was way ahead of its time... they were doing "vibe coding" back in the 1990's :)

All true, which is why China's system isn't completely centralized like that. Centralization is a matter of degree, not a binary on/off switch. The Chinese government mandates the broad strokes, and leaves the detail work decentralized, to be handled by the market. They've got a lot more capitalism in their system at this point than they'd probably care to admit.

Waymo is currently at 250,000 autonomous rides per week, in six cities, and I mean autonomous in the strict sense of "there are no Waymo employees in the car". They seem to be competing pretty well.

As for how other companies will compete against Tesla in the future when Tesla finally makes good on their ambitious promises.... we'll find out, if and when Tesla finally makes good on their ambitious promises. You shouldn't count those chickens until they've hatched.

If you spend time with the higher-tier (paid) reasoning models, you’ll see they already operate in ways that are effectively deductive (i.e., behaviorally indistinguishable) within the bounds of where they operate well. So not novel theorem proving. But give them scheduling constraints, warranty/return policies, travel planning, or system troubleshooting, and they’ll parse the conditions, decompose the problem, and run through intermediate steps until they land on the right conclusion. That’s not "just chained prediction". It’s structured reasoning that, in practice, outperforms what a lot of humans can do effectively.

When the domain is checkable (e.g., dates, constraints, algebraic rewrites, SAT-style logic), the outputs are effectively indistinguishable from human deduction. Outside those domains, yes it drifts into probabilistic inference or “reading between the lines.” But to dismiss it all as “not deduction at all” ignores how far beyond surface-level token prediction the good models already are. If you want to dismiss all that by saying “but it’s just prediction,” you’re basically saying deduction doesn’t count unless it’s done by a human. That’s just redefining words to try and win an Internet argument.

They do quite a bit more than that. There's a good bit of reasoning that comes into play and newer models (really beginning with o3 on the ChatGPT side) can do multi-step reasoning where it'll first determine what the user is actually seeking, then determine what it needs to provide that, then begin the process of response generation based on all of that.

That reminds me of the Mark Twain quote: "Good judgement is the result of experience, and experience is the result of bad judgement".

i.e. you certainly can become an expert yourself, eventually, but you'll probably make some sub-optimal financial decisions in the process.

This is not a surprise, just one more data point that LLMs fundamentally suck and cannot be trusted.

Huh? LLMs are not perfect and are not expert-level in every single thing ever. But that doesn't mean they suck. Nothing does everything. A great LLM can fail to produce a perfect original proof but still be excellent at helping people adjust the tone of their writing or understanding interactions with others or developing communication skills, developing coping skills, or learning new subjects quickly. I've used ChatGPT for everything from landscaping to plumbing successfully. Right now it's helping to guide my diet, tracking macros and suggesting strategies and recipes to remain on target.

LLMs are a tool with use cases where they work well and use cases where they don't. They actually have a very wide set of use cases. A hammer doesn't suck just because I can't use it to cut my grass. That's not a use case where it excels. But a hammer is a perfect tool for hammering nails into wood and it's pretty decent at putting holes in drywall. Let's not throw out LLMs just because they don't do everything everywhere perfectly at all times. They're a brand new novel tool that's suddenly been put into millions of peoples' hands. And it's been massively improved over the past few years to expand its usefulness. But it's still just a tool.

AI could provide the push to finally popularize secure, authenticated email, if only because the phishing attempts will become so realistic that there will be no other way to distinguish your grandma from a foreign scammer :/

I concur with that (I'm 52, no more 10-hour marathons at the keyboard for me!).

Note however that the proper metric to measure isn't "time elapsed to create code", but "time elapsed to get that code sufficiently bug-free that you can ship it to customers without it causing a mess".

Any high-school student (or AI) can write code that looks reasonable and passes the basic acceptance tests. The real trick is getting that last 1% correct, so that the code can "just run" indefinitely, unsupervised, for all use-cases, without crashing or misbehaving or otherwise requiring a human being's time to manage it.

We'll have to do that also, and if the problem was just a single, one-time step-change (i.e. "we'll have to live in temperature range B from now on, instead of traditional temperature range A"), that would be sufficient.

But it's not that easy, is it? The climate changes, and as long as we keep increasing the CO2 content of the amosphere, the climate keeps changing more.

So we could adapt to the climate we expect to be living in 20 years from now, but without also solving the emissions problem, that won't be enough: we'll have to adapt again some years after that, to a climate that's even worse. And again, some years after that, and so on, likely until our only remaining way to "adapt" is through mass die-offs.