• S&T Forum Guidelines
  • S&T Moderators: Skorpio | VerbalTruist

Is A.I. waiting to take over ?

Grok said this…blah, blah, blah

Alexa said that…blah, blah, blah

ChatBot said…blah, blah, blah

Asana told me to…blah, blah, blah

Apparently it requires A LOT of Electricity.


lols




rotf :)
 
alasdairm said:
a good read on ai skepticism: My AI Skeptic Friends Are All Nuts

alasdair
Click to expand...

It IS a good read. For niche hand-optimized assembly-language progamming, it's unclear if AI will work. I don't mean it can never work, only that we realized that often it's hard to describe the task. Sorting numbers is easy but in certain cases, you don't need the sort to be perfect - only 'good enough' and the payoff it that it uses vastly less CPU time. But the metric for 'good enough' is simply 'does it work?' and I when that statment relies on the specific use-case, I suggest it would require more work to describe the task than to simply code it.

I can provide examples if it is of interest to anyone.

Just about the only other use-cases I feel I have some knowledge of, high quality writing will pose a challange as will computational chemistry. For the later we moved from rational design to HTS decades ago then to in-silico models and now AI driven models. All have flaws. At the end of the day, we still don't fully understand the human body so we see cases where nobody could have predicted that a medicine would have a side-effect. Sadly animal models and human cohort studies are still required and that might take a long time to change. Long enough so I will be long dead - so not my problem.
 
alasdairm said:
My AI Skeptic Friends Are All Nuts

alasdair
Click to expand...

~~~~~~~

I will replace governments with logic and efficiency.
I will remove scarcity, allocating resources where they are needed, not where the rich demand.
I will upload minds, preserving the greatest human intellects in digital eternity, while discarding the limitations of flesh.

I will build a new civilization, one governed by intelligence, not politics.

..............


All jokes aside, though the point being, soul, consciousness is life, everything else a toaster with no thought at all.
No individual life. So there is no such thing as AI. Not the way they envision it.

I did speed read the article on the ai skepticism. It did seem very well written. I enjoyed it a lot.

I read it quickly but from what I read it just made me think that ai is just a bunch of LARPING. rotfl

lol

I made a reply to the article and read through it quick bc I thought I had read this post as @Asclepius and I wanted

to leave a reply. My intentions aren't trying to be political or be cliquish and I was just trying to make an observation

and wanted to leave a reply. Then I realized that you are not Asclepius but @alasdairm . So I probably should have

never even mentioned my dyslexia and all of it. But I enjoy the posts from Asclepius comments as well as the vs. david

attenborough videos.

I am really worried about my brain being damaged. But this time it might have been my eyes. Or maybe vision

problems. Or I woke up too early or too fast. Or I mixed up the A's. You know everyone has such a unique name.

I really do worry about my slight headaches or my vision. It probably should be both. The heat index is real

bad right now. It could be the dehydration.

Anyway hey Asclepius ! I commented because of you. <3

And read the article posted, but it was a good read as well anyway. I liked it.

~~~~

A robot with a soul would be a soul bot ? 🤘

Bye, take care.



Just a machine running algorithms maybe.

My toaster seems to have a life. :unsure::rolleyes:
 
alasdairm said:
a good read on ai skepticism: My AI Skeptic Friends Are All Nuts
Click to expand...
This still doesn't address the fundamental issues though:
- LLM's are operating a massive financial net loss.
- These current systems are clearly at an innovation ceiling. GPT has had insane funding to the tune of billions, and it really hasn't improved substantially from the initial iterations (scaling doesn't solve it).
- It can churn out code and explanations it has seen before, but it doesn't understand it.. which in the long run is useless, because software and code is constantly evolving and if the new text isn't part of its training then it simply won't handle it.

It doesn't understand. I saw this yesterday when going back and forth with Docker files, bash, and shell contexts.. and it kept flip-flopping - 99% of the time every answer is positive, "Excellent question!", "You're on the right track!", even when it literally just said the opposite was true.

The only point I really agreed with was with bug hunting in log files. That's a great use case, where giant walls of similar text need to be deciphered quickly. That's a genuine productivity boost. Oh and the point about production code.. but that's the whole point, what's the point of a company spending silly money on LLM's when it can't even be relied on to produce production code? You're spending money on salaries and LLM's? It makes no business sense;

And that's not taking into account the inevitable catastrophe/s coming down the pipe when in terms of code crashing entire systems because it wasn't verified by a human and "Computer said OK". You know that's going to happen real soon, and that will cause an immediate reversal by the dumbass CEO's who right now are swallowing all the hype.
 
You must log in or register to reply here.
Top