12 Comments
Dec 9, 2022Liked by coloradotravis

I ran your article by my daughter today. She has a Philosophy degree with an ethics minor and is pursuing a masters in Philosophy in AI. Right up her ally.

To sum up her points, she said your argument is based on the two premises that 1) AI is designed as a regression engine and 2) we taught them to mirror and not learn.

Her counter is that the same can be said about humans. While we humans hold ourselves up as a (unjustifiable) gold standard, we take our accumulated knowledge and try and extrapolate from there.

The argument that AI is too mechanical is a feature, not a bug.

Expand full comment
author

Her counter ignores the fact that AI does not ‘solve from there’

Expand full comment
author

Like let’s say you ask the AI the following question: “why do green ideas sleep furiously?”

- what does the AI do?

- what does a human do?

Expand full comment
Dec 9, 2022Liked by coloradotravis

I would assume that the AI would be programmed to go through the same sentence breakdown a human would.

What is meant by:

"Green", then "Green Ideas"

"Sleep", then "Sleep furiously"

then as a full sentence.

Continue to parse through the meanings until you have the highest likely understandably rating and present that.

BTW, my daughter is the student. I just sling precious metals for a living.

Expand full comment
author

Ha! Fair enough.

The thread to pull is the AI's lack of agency. It cannot reject the question, nor does it have the framing to do so, and that sort of 'decision not to play' is one important signpost of how proper intelligence is different.

Expand full comment

Don't know how to post a jpg here, but here is the proper link.

http://hilobrow.com/wp-content/uploads/2010/12/wargames.jpg

Expand full comment

I’ve played around with GPT chat bot and boy is the prose convincing. Very ‘Her’ type vibes and gives you what you want to hear. But still more practical advice instead of just platitude. I’d rather bet on the entire internet dataset than depend on what one expert therapist says for example.

Expand full comment

So your saying its not ai unless its skynet. Hmkay.

Expand full comment
author

Well I literally detailed ways in which it might be interesting and useful, it’s just not ‘intelligence’ — that part is overstated branding

Expand full comment

I fully agree, i just think it’s amusing that once a pattern recognition engine does in fact become „intelligent“ by a reasonable human definition like yours it always ends in our demise. Helpful ? Absolutely! Intelligent? Lets pray it’s not !

Expand full comment

Always great, think I learned some things but not sure what. How do i know? My head hurts from thinking.

Expand full comment

And, a lot of books didn't make it into the Library of Alexandria, just like a lot of university archives didn't make it into Google after all. These days, it's what isn't on the Internet that's really interesting.

Expand full comment