“I believe that the conventional idea of ‘writing a program’ is headed for extinction, and indeed, for all but very specialized applications, most software, as we know it, will be replaced by AI systems that are trained rather than programmed.”

Who is going to create those AI systems? Programmer Analysts?

He's a blithering nincompoop, apparently.

A hand-coded program will always be more efficient and effective than an entire goddamn AI.

What the hell is he thinking?
Does he know anything about any of this?

“For example as people become smarter with the use of computers” - interesting, I have been thinking lately that people are getting dumber and dumber using computers

Yeah, i think i phrased that wrong. I wasn't referring to genetics of brains making people smarter, but rather the combination of brain with the aid of a computer makes a person smarter. But I do agree that growing dependence on computers could possibly result in reduced brain smartness, though I kinda doubt it. Who knows.

As interesting as the flynn effect studies are, I'm not sure you can depend too much on their results. Pretty small sample sizes over a relatively short period of time:

Flynn effect and its reversal are both environmentally caused
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6042097/

Flynn effect and its reversal are both environmentally caused

Using administrative register data with information on family relationships and cognitive ability for three decades of Norwegian male birth cohorts, we show that the increase, turning point, and decline of the Flynn effect can be recovered from withi…

^{PubMed Central (PMC)}

A human can do that better.

When people say "AI is making this possible" they really mean "This is stuff so worthless to us, that we won't hire a person to do it".

That is literally all it means.

And when coding jobs are axed because of AI, it is still the same message "This work is so worthless to us, that now that we can get not-a-person to do it, y'all are out of here".

They don't care about the quality. They foolishly believe the AI will improve, despite eating its own diarrhetic shit from here on out.

the combination of brain with the aid of a computer makes a person smarter

I'm not sure this is a notion that is generally applicable -- like in many things, real life scenarios are complex. "SQL" certainly made people vastly more productive, and the use of SQL instilled a good practical understanding of Set Theory logic. But, once you move up the ladder of abstraction from this already very abstract database interface, you are bound to loose a great deal of understanding of what is is you are doing. That's not smart.

So, the relational database idea is a good example of computer inspired abstraction.

This was my initial point. You are missing what followed. If we move upward from this level of abstraction, you loose all of this: set operations such as join, union, intersection, etc. So the SQL abstractions are: selected item, from sources, set operations on item sources, selection condition.

These are the basic operations, but it can get significantly more complicated from these basic operations as well. Who is going to be able to cognate what they need to do to get what they want when they have no basic understanding? They won't have the mental machinery in place, nor the symbols to employ.

But we can't afford to keep it powered right now, and our budget for energy use keeps shrinking.

If it is smothered in the crib, none of that will happen.

AI will design software, write software, and do all the maintenance and upkeep. But it’s foolish to try to predict when. AI seems to have been racing ahead in the last few years

AI is like an invasive weed in a plowed field. The "crop" is at a disadvantage. We have got to find our area of superiority, or we will get overgrown.

While “musician” is totally in the crosshairs of AI, as we see, that applies only to musicians who make recorded music – going “live” may be a way to escape the automation. No robots with the manual dexterity to play physical guitars, violins, etc, appear to be on the horizon. Maybe they can play drums?

I'm still not sure of this. The machines can "make" music, but can they make it meaningful to humans... you know, something like "Pinball Wizard".

Some people sneer at AI gobbling up resources. Ha, look at energy consumption around the world.

The human brain uses about 25 watts of electricity i'm told, and that's because it's very miniaturized. And there are a few billion people times 25 watts. So, people gobble up quite a bit of energy too. Won't be long before computers start miniaturizing too.

All this worry and sneering about computers is probably a bit misplaced. Like Wayne says, it's pretty hard to predict the future. I'm reminded of the story about the farmer driving his horse drawn wagon and passing a new fangled model A ford stranded by the side of the road with a flat tire, and the farmer laughs as he passes and says, "get a horse."

The human brain uses about 25 watts of electricity i’m told, and that’s because it’s very miniaturized.

Yet humans are extremely wasteful of "brainpower" or potential brainpower. Look what a crow can do, using about two orders of magnitude less.

“get a horse.”

Hey, horses can go anywhere.

The brain uses exactly 0W of electricity.’

If you fail to understand that hard fact, you’re going to make bad decisions, like this fella :

https://newrepublic.com/article/180487/balaji-srinivasan-network-state-plutocrat

The Tech Baron Seeking to Purge San Francisco of “Blues”

If Balaji Srinivasan is any guide, then the Silicon Valley plutocrats are definitely not OK.

^{The New Republic}

Like our muscles, our brains consume a lot of energy every day—approximately 20 watts to maintain spontaneous neuronal activity.

And exactly zero Watts is electricity.

Do you understand basic biology?

I was pointing out that you were spouting gibberish.

A supercomputer emulating a brain 100th the size of an ants is a completely different sort of problem than 1000 humans eating a sandwich.

AI is a wasteful and stupid technology, and we cannot afford to use any electricity on it, because we are already consuming more electricity than is sustainable.

Handwaving about "miniaturization" just underscores how poor your understanding of any of this is.

A brain has trillions of neurons.

The largest AI has the equivalent of maybe 10000 neurons, and uses MW of energy to run it all.

That has nothing to do with "miniaturization".

Actual neurons are just magnitudes more efficient than their emulations, and there is no reason to think this will ever change.

Intelligence can be thought of as information compression. Much of human intelligence is a result of culture having encoded a compressed scheme of reality in such a way that people can learn the shared encoding scheme. Similarly with AI. Moral of the story is that people who mock and sneer at AI for being overhyped and too resource intensive might not understand what the bigger picture is.

True, in some cases "get a horse" might be the answer, but evolution has many tricks up its sleeve, and one should not keep all their eggs in one basket of contempt. So to speak.

I'm not sure how we got on the relevance of miniaturization, but
looking at the numbers with a current "system on a chip", it seems processors are beginning to scale up to brain size. That sure does put a spin on what we think of as classical programming. Just imagine you are a programmer and you have a computer at your disposal the size of current mobile phone computers. Billions of transistors at Trillions of cycles per second! Kinda boggles the mind:

The A17 Pro boasts 19 billion transistors and a 6-core CPU, with two high-performance cores ..., and four high-efficiency cores.

The 16-core neural engine can process up to 35 trillion operations per second, .... There are also additional dedicated engines for [graphics].

The 16-core neural engine can process up to 35 trillion operations per second, … There are also additional dedicated engines for [graphics].

One of the fascinating thing about the brain is that it functions in a variety of modalities, some of which are happening in parallel. Then there is the hybrid digital/analog nature of the brain as well. We are still only understanding the surface of how all this actually works.

exactly zero Watts is electricity

Technically there is an amazing amount of actual electrical activity happening in the brain. Any movement of ions across a gradient can be termed “electrical activity”. More specifically, neuronal action potentials are formally considered “electrical activity”, at least by the definition that the NIH uses.

https://www.ncbi.nlm.nih.gov/books/NBK546639/

FFS, AI hype-jockeys are always trying to dishonestly exaggerate their bullshit.

Like, six months ago they were really desperately trying to compare Parameters to Neurons.

Now apparently it's comparing transistors to neurons.

Both are completely false comparisons.

The only reasonable comparison is between Layers and Neurons.

And even then, it's clear than one layer is less powerful than a neuron, although whether it's 7 times or 70 times, we don't really know.

Bottom line: As I said, the heaviest AI is not even a tenth the power of the brain of an ant.
In other news, no, LLMs do not understand the bullshit they spout.

And in yet other news, AI providers are downgrading expectations across the board, because this absolute bullshit technology is only generating profits for the hardware manufacturers, and that too will not keep up once the bubble bursts.

"LLMs do not understand the bullshit they spout."

LLM's have met their match here.

"the heaviest AI is not even a tenth the power of the brain of an ant."

Are you measuring ant power in watts? Lol.