Time to talk about the AI revolution

[Lowell]

Friends, over the next five years, many of our lives will be very dramatically affected by the advent of Artificial Intelligence: AI.

I have been following AI closely for the past two years and, with my paid subscription to ChatGPT, have been using AI on a nearly daily basis, as, for example, web coding for an art project I am creating.

I no longer write the code or use apps to enable it. Instead, I tell the AI what I want, and it writes the HTML, CSS, and JS coding for me. I simply copy and paste the code, push it to the web, review it, and tell the AI what tweaks I need until I get what I want. I've saved hundreds of hours.

The implications of this are obvious. If I, in my little world, am having these efficiencies, why can't a coder in a company do the work of 10, 20, 30 other coders? In fact, they can and increasingly are.

30-50% of white-collar jobs are vulnerable to replacement by AI agents by 2030, a disruption already being dubbed as "The Coming Job Apocalypse". [See my post about Torc Robotics for blue-collar effects]

As of now, all major AIs can pass the bar exam in the 98th percentile, measure over 130 in the Mensa Norway test, and code on par with the best coders in the world. Not bad for three-year-olds, right? Imagine them as grown-ups.

Thoughts?

[401don]

The concerns about all the potential evil AI can cause aside from job loss makes me glad I'm old and won't be around for the worst of it.

[Lowell]

The concerns about all the potential evil AI can cause aside from job loss makes me glad I'm old and won't be around for the worst of it.

If I had a nickel for every time I've heard that from my age cohort, I'd be a millionaire.

[401don]

If I had a nickel for every time I've heard that from my age cohort, I'd be a millionaire.

Mostly going from what I've read and seen on tv. Cybersecurity threats, deepfakes, etc. we're seeing now are nothing compared to what's in store. Those who are running the AI companies, who know the most about it, appear to be the most concerned.

[mwilbert]

AI is going to be very disruptive and we aren't really prepared for it. I think Lowell is overstating the case a bit in the near term. Coding is really what it is best at right now, and while there are some kinds of code it's quite good at, it still has a lot of issues. But usually with code, you can quickly tell when it's making stuff up. In other contexts, not so easy.

That being said, it's already plenty good enough to displace a bunch of workers, and also to flood the internet with junk, and apparently also to lead people into kind of self-destructive behavior. This article from a week ago was kind of worrying:

https://www.nytimes.com/2025/06/13/t...smid=url-share

[Jimaz]

... This article from a week ago was kind of worrying:

https://www.nytimes.com/2025/06/13/t...smid=url-share

Well, thanks for posting that. That was creepy as all Hell. It's as if the programmers and the programmees were swapping roles.

The first problem I have with AI is that it's so poorly defined. It means different things to different people. That makes it difficult to discuss at all.

The second problem is that it seems to be an alluring way to persuade people to accept inferior quality that's cheaper to produce. There's a whole lot of profit-driven gaslighting going on here.

I would really like to see AI invent something that humans couldn't invent. I wouldn't be surprised if that's already been done.

[DetroiterOnTheWestCoast]

the concerns about all the potential evil ai can cause aside from job loss makes me glad i'm old and won't be around for the worst of it.

same here!

[canuck]

same here!

Ditto à la power of ten. Lol

[Zacha341]

I'd like to think like that but have adult children and grandchildren...

The concerns about all the potential evil AI can cause aside from job loss makes me glad I'm old and won't be around for the worst of it.

[Dan Wesson]

AI doesn’t get its hands dirty.

[mwilbert]

AI doesn’t get its hands dirty.

Not very often, for the moment. But it's not going to be that long, I think.

https://www.techspot.com/news/108425...ing-truck.html

[Jimaz]

A linguistic professor's opinion from a couple of years ago about large language models:

Noam Chomsky on Artificial Intelligence, ChatGPT

Watch the full interview with Chomsky on China, AI, and the 2024 election now: Noam Chomsky: On China, Artificial Intelligence, & The 2024 Presidential Election

[JBMcB]

Remember about 15 years ago when IBM built a computer that could play Jeopardy? It did pretty well. IBM then positioned it as being able to replace doctors and CPAs, as it could learn and do just about anything.

It went on to not do any of those things.

Modern Large Language Models are a bit more advanced than Watson, mainly because of the huge amount of speed modern computer hardware can throw at the AI problem. It isn't significantly more capable, though. The algorithms are the same. The problems with reliably training the models are the same. The need for humans to constantly tweak the output to get something useful is the same.

The only real revolution over the last decade and a half is you can run one of these things on a midrange home computer instead of needing a $500,000 mini-super-computer.

[mwilbert]

Remember about 15 years ago when IBM built a computer that could play Jeopardy? It did pretty well. IBM then positioned it as being able to replace doctors and CPAs, as it could learn and do just about anything.

It went on to not do any of those things.

Modern Large Language Models are a bit more advanced than Watson, mainly because of the huge amount of speed modern computer hardware can throw at the AI problem. It isn't significantly more capable, though. The algorithms are the same. The problems with reliably training the models are the same. The need for humans to constantly tweak the output to get something useful is the same.

The only real revolution over the last decade and a half is you can run one of these things on a midrange home computer instead of needing a $500,000 mini-super-computer.

This is simply not true. There are a huge number of things that can be done now that were impossible 15 years ago.

[Henry Whalley]

There are a huge number of things that can be done ...

The Machinery of Death
https://youtu.be/VJUY4pgd3WQ

[JBMcB]

This is simply not true. There are a huge number of things that can be done now that were impossible 15 years ago.

Not impossible, just slower. The main advancement between Watson and current LLMs are that we have a lot more RAM available, so we can train on larger sets of text. So, instead of a network with general knowledge, it also stores recipes and code and detailed biographical and sports data and ad copy.

Watson could have been trained to create recipes, but it wouldn't have been good at Jeopardy or recommending movies to watch or chatting with you. You would have to have dozens of language models set up for specific tasks loading in and out all the time, which would have taken ages. Now, with hundreds of gigabytes sitting on pooled GPUs, everything can be loaded at the same time.

[Lowell]

NY Times had an article a few days back [Link here] trumpeting

A.I. Might Take Your Job. Here Are 22 New Ones It Could Give You.
In a few key areas, humans will be more essential than ever.

However, after reading it, I think it might require an asterisk after New Ones* with the footnote reading *Temporary.

Almost all of the jobs are related to humans validating what the AIs create for CYA legal, ethical, and error-checking for an organization’s purposes or bringing bosses up to speed on how to use AI to eliminate more jobs.

I could see where most of the jobs cited could, in turn, be greatly reduced or ultimately eliminated by more advanced AI models.

This is certain to be a political issue. How will an economy where a vast section of the middle class and its consumption is excluded, function? How will the abundance created by AI be shared?

[Jimaz]

... The need for humans to constantly tweak the output to get something useful is the same....

And this belies the "artificial" part of "artificial intelligence." If it still requires a human to decide whether the output is useful or tweak it until it is useful, how is that different from a human debugging a program? I would say there is no difference. That part of the intelligence, the natural human part, is being falsely included with the artificial part.

True artificial intelligence would be able to invent something useful without being corrected {selected or tweaked by a human} or even "trained."

I'm not saying that's impossible, I'm just saying that the common definition of "artificial intelligence" is far too broad to be impressive.

The intelligence of AI is inversely proportional to the intelligence of the observer.

[casscorridor65]

Yes, I agree with Lowell that AI will make dramatic changes, most significant in white collar jobs. Trump wants manufacturing back to offset those job losses. As an investor, I follow several Wall Street companies that are all going to be more profitable because they are quickly replacing workers with AI. Salesforce [[CRM), one of my stocks, is up today after announcing 30 - 50% of its internal work is now done by AI, eliminating over 1000 jobs. With companies being forced to return to the United States and immigrants being sent away, there will be jobs "putting tiny screws in I-phones," according to Peter Navarro, t-shirt and sewing jobs, and jobs picking fruits and vegetables, cleaning hotel rooms, construction, and lawn services. So, all is not lost!

"Earlier this year, a World Economic Forum report found that 48% of U.S. employers plan to reduce their workforce because of AI."

AI's impact on the job market is ‘inevitable,’ says workforce expert: 'It's going to hurt for certain parts of the population'

[canuck]

Yes, I agree with Lowell that AI will make dramatic changes, most significant in white collar jobs. Trump wants manufacturing back to offset those job losses. As an investor, I follow several Wall Street companies that are all going to be more profitable because they are quickly replacing workers with AI. Salesforce [[CRM), one of my stocks, is up today after announcing 30 - 50% of its internal work is now done by AI, eliminating over 1000 jobs. With companies being forced to return to the United States and immigrants being sent away, there will be jobs "putting tiny screws in I-phones," according to Peter Navarro, t-shirt and sewing jobs, and jobs picking fruits and vegetables, cleaning hotel rooms, construction, and lawn services. So, all is not lost!

"Earlier this year, a World Economic Forum report found that 48% of U.S. employers plan to reduce their workforce because of AI."

AI's impact on the job market is ‘inevitable,’ says workforce expert: 'It's going to hurt for certain parts of the population'

I don't worry too much about people, it's the economy I worry about. lol

[Cincinnati_Kid]

I don't worry too much about people, it's the economy I worry about. lol

Hear, hear!

[Lowell]

Give them time Jimaz. They're just three-year-olds.

True artificial intelligence would be able to invent something useful without being corrected {selected or tweaked by a human} or even "trained."

I'm not saying that's impossible, I'm just saying that the common definition of "artificial intelligence" is far too broad to be impressive.

The intelligence of AI is inversely proportional to the intelligence of the observer.

[Henry Whalley]

I don't worry too much about people, it's the economy I worry about. lol

Capitalists will inherit the earth as prophesized.

[Henry Whalley]

How will the abundance created by AI be shared?

It won't be shared. Some things never change.

[Henry Whalley]

The concerns about all the potential evil AI can cause aside from job loss makes me glad I'm old and won't be around for the worst of it.

Ditto. Now to cheer up, I'll read Schopenhauer.