[Discuss] LLMs and AI

[kentborg at borg.org (Kent Borg)]

On 1/29/26 12:28 PM, markw at mohawksoft.com wrote:
> Anyone want to discuss?

We are in a *very* interesting time.


These things can't "think", but what they can do anyway can still be 
very impressive.

These things are very dangerous; lonely, vulnerable, suggestible people 
are being destroyed.

These things are turning the economics of the web upside down; web sites 
with good content are losing ad revenue, new crappy web pages made by 
LLMs are flooding the web.

These things got better by training on more and more data, but they have 
run out of data, worse, coming models are going to be poisoned by 
training on their own terrible output.

These things will make us intellectually lazy unless we are extremely 
careful, latter-day Luddites are going to get their revenge as they will 
still be able to think.

These things are very well suited to programming, particularly in Rust, 
because the compiler is to strict about so much.

These things don't scale, the "transformer" innovation that made them 
possible does not scale.

These things got me fired, in fact, I am pretty sure ChatGPT devised how 
to decide who to dump when my employer did a massive layoff.

There things are turning the labor market upside down, junior level 
positions are are being eliminated, the pipeline for creating senior 
positions is broken.


And, there will be a crash, the hype is not justified. There is no way 
it could be.


-kb, the Kent who has learned a lot about these things, and is full of 
opinions, more everyday.