AIs can spit out a thousand articles a minute. But are they worth reading? Are they located on a site where people actually go?

A few years ago, we started to see SEO-driven spam websites, with nothing but regurgitations of other websites' contents. These sites existed just to harvest clicks. But everybody knew they had made a mistake when they happened to land on one of these sites, and clicked away.

This is what I suspect is happening with this "vast number" of AI-written articles. You're certainly not going to see them on sites where people actually go intentionally.

Let's not get ahead of ourselves.

A few years ago, more than half of all websites were SEO-driven automatically-generated spam. The point is, quantity means nothing. Nobody actually read those SEO pages, and clicked away as soon as they mistakenly entered them.

The headline implies that half of all articles *people read* are AI-generated. I highly doubt that. Just because the articles are generated, doesn't mean they are on sites that anybody will actually read.

I agree your illustration is good, if your audience is familiar with "open world" games. Many of us are not, but just about everybody has seen a movie.

It's true that some things have increased in cost faster than inflation. But you don't get to just pick those things, leaving out all the other things that have *decreased* in cost, and claim that life is worse now. Inflation and real wages take into account *all* the things a family needs, not just the ones that have experienced the highest price increases.

This is a gross oversimplification. No, the difference between weather and a simulation is *not* just that the weather can get you wet. If you go into one of those carnival "4-D" rides, the movie can "get you wet" and many of them do. But that fact doesn't make the movie "weather." It's still just a simulation.

I won't respond to all your points, but there are two that I feel are deserving.

Yes, we do know how LLMs work. Maybe you don't, but the engineers at OpenAI and Microsoft and Google etc., absolutely do. The evidence of this is that over the past three years, they have been able to repeatedly and steadily improve the quality of the chatbots' responses, and to correct incorrect responses of the past. One notable example was an early Gemini image generator, which, when asked to render drawings of historic figures like Lincoln, would render an image with the wrong gender or race. The designers had built this into the LLM, but didn't fully anticipate the ultimate fallout. So they fixed the image generator to be more historically accurate, before they brought it back online. This kind of correction would not be possible, if the engineers didn't understand thoroughly how it works.

This statement misunderstands how LLMs work. At one level, you're right, intelligence can't be "simulated." But the responses a chatbot gives, certainly do simulate the responses an actual intelligence would give. But it's an illusion. The LLM simply digests a myriad human responses to the question provided, and synthesizes and adapts the response to *your* question / prompt, based on the patterns it saw in its training data.

We may not fully understand how humans think, but we do know how LLMs work. LLMs are essentially a sophisticated pattern recognition algorithm. Based on their training, they compose sequences of tokens that approximate what would be expected in response to a prompt.

AI is to intelligence, as a movie is to motion. When watching a movie, there is a very convincing appearance of motion, but in fact, nothing on the screen is actually moving. It can be so convincing that viewers using 3D glasses might instinctively recoil when an object appears to fly towards them. But there is no actual motion. The characters have no intent, though humans assign intent to what the "characters" are saying and doing. The point is, it's an illusion. And in the same way, AI is an illusion, a fancy (and very useful) parlor trick.

It tries to do too many things.

If you've ever written software that hooks into the Explorer API (such as to show contents of Zip files), you find out that Explorer calls your API *incessantly*, often hundreds of times just in the process of showing a list of files. It wants to know about icons and descriptions and content previews and such. You have to implement some serious caching to make sure you don't reduce it to a crawl.

It's no wonder it takes a long time to load.

Models like DeepSeek have demonstrated a significantly lower requirement for GPU power. Other models will no doubt be following in its footsteps, one way or the other. The current massive buildout of AI data centers, isn't sustainable. People *will* figure out how to make AI processing more efficient, probably by orders of magnitude.

Yeah, I used to have that problem with my toenails, but luckily there is now a pill to cure that.

Just about all of us are "low value" from this perspective. High value targets are always subject to spying and stalking, technological or otherwise. The paparazzi is not new.

It's true, metro areas are different from cities. A "city" is a political entity with a government and specific geographic boundaries. A metro area is not about politics, it's about concentrations of people who live as a single community. Both are valid ways of looking at population, the important thing is to be consistent in such comparisons. https://en.wikipedia.org/wiki/...

This is kind of like how hospitals are now required to make their prices "public." So they make a web page listing all the prices, but don't make it indexable by Google. They're technically complying with the law, but because Google can't find it, neither can potential patients.

In the rent case, landlords can simply create "public" pages (not indexable by Google) listing their rents. Since they are now "public" the algorithm can use the prices, even if regular people can't find the pages.

Both outcomes are about equally likely.

Yes, AI *could* replace 3 million UK jobs in 9 years. But those who suggest this could happen, greatly underestimate the complexities of automation. AI is not a special case, when it comes to overcoming automation complexity. It can quickly handle the "happy path." But it sucks at handling edge cases.

Consider the Taco Bell AI that complied when somebody ordered 18,000 waters. A human would have known that the order wasn't real. This is just one of a million one-off scenarios that AI will encounter, and it won't be able to handle them until somebody does the work to make it handle them.

Another example. I recently used GitHub Copilot to help me embed a Vue.js page within an ASP.NET Core app. There are several ways to do this that are mutually exclusive. AI couldn't keep them straight. Once I picked an option, it kept suggesting steps that didn't apply, because they pertained to a different toolchain.

These are the kinds of complexities that will prevent wholesale elimination of human jobs.

There are a whole lot of people, some of whom frequently comment on Slashdot, who apparently think AI is actually becoming intelligent, and will soon replace all human thinking, and especially, jobs that require thinking. You don't have to read many posts to run across these guys!