Recent survey of experts: Current AI models are a 'dead end' for human-level intelligence

C C

C C

Consular Corps - "the backbone of diplomacy"
Valued Senior Member
Ethan Landes: "LLMs are not intelligent agents or advanced search engines. Modern LLMs just make predictions of what the next token will be, and choose one of the likelier tokens... [...] Because these LLMs are nothing more than predicting the likely next words, when they tell you what the capital of Canada is, whether or not they get the question right, they do not care about telling you the right answer. There’s nothing there to do the caring..." --Learning from AI’s bullshit
- - - - - - - - - - - - - - - -

Current AI models a 'dead end' for human-level intelligence, scientists agree
https://www.livescience.com/technol...human-level-intelligence-expert-survey-claims

INTRO: Current approaches to artificial intelligence (AI) are unlikely to create models that can match human intelligence, according to a recent survey of industry experts.

Out of the 475 AI researchers queried for the survey, 76% said the scaling up of large language models (LLMs) was "unlikely" or "very unlikely" to achieve artificial general intelligence (AGI), the hypothetical milestone where machine learning systems can learn as effectively, or better, than humans.

This is a noteworthy dismissal of tech industry predictions that, since the generative AI boom of 2022, has maintained that the current state-of-the-art AI models only need more data, hardware, energy and money to eclipse human intelligence.

Now, as recent model releases appear to stagnate, most of the researchers polled by the Association for the Advancement of Artificial Intelligence believe tech companies have arrived at a dead end — and money won’t get them out of it... (MORE - details)
_
 
I liked Emily Bender's coinage, "stochastic parrot." Agree heartily they are a dead end. Will join ELIZA in the dustbin of things that people can be fooled by.
 
Accepting that Coles Note compendium of human knowledge as an intelligence is like accepting Star Trek actors with little bumps on their foreheads as aliens.
Of course, that doesn't mean it won't put a lot of people out of their job - every advance in automation does.
 
  • Like
Reactions: C C
C C said:
Ethan Landes: "LLMs are not intelligent agents or advanced search engines. Modern LLMs just make predictions of what the next token will be, and choose one of the likelier tokens... [...] Because these LLMs are nothing more than predicting the likely next words, when they tell you what the capital of Canada is, whether or not they get the question right, they do not care about telling you the right answer. There’s nothing there to do the caring..." --Learning from AI’s bullshit
- - - - - - - - - - - - - - - -

Current AI models a 'dead end' for human-level intelligence, scientists agree
https://www.livescience.com/technol...human-level-intelligence-expert-survey-claims

INTRO: Current approaches to artificial intelligence (AI) are unlikely to create models that can match human intelligence, according to a recent survey of industry experts.

Out of the 475 AI researchers queried for the survey, 76% said the scaling up of large language models (LLMs) was "unlikely" or "very unlikely" to achieve artificial general intelligence (AGI), the hypothetical milestone where machine learning systems can learn as effectively, or better, than humans.

This is a noteworthy dismissal of tech industry predictions that, since the generative AI boom of 2022, has maintained that the current state-of-the-art AI models only need more data, hardware, energy and money to eclipse human intelligence.

Now, as recent model releases appear to stagnate, most of the researchers polled by the Association for the Advancement of Artificial Intelligence believe tech companies have arrived at a dead end — and money won’t get them out of it... (MORE - details)
_
Click to expand...
Good article. I note they mention, though only in passing, that these chatbots are increasingly subject to error due to reading the increasing amount of dud information put out on the web.......by chatbots! This is a point I have read about before. The errors are actually getting worse over time, because they are re-ingesting their own shit to create more shit. Flooding the zone with shit, in fact.

What was that again about enshittification?
 
I still stand by my claim that AI exceeded the intellectual capabilities of roughly half of Americans in all meaningful respects like nearly a decade ago.
 
  • Haha
Reactions: C C
TheVat said:
I liked Emily Bender's coinage, "stochastic parrot." Agree heartily they are a dead end. Will join ELIZA in the dustbin of things that people can be fooled by.
Click to expand...
Yeah, but doesn't that kinda... well.
 
exchemist said:
Yes but nobody treats the average guy as an encyclopaedia.
Click to expand...
Yeah, that, and what precisely are the parameters for General Intelligence? As Landes said, "There's nothing there to do the caring." Some might put that outside the confines and concerns of General Intelligence, but who knows? In consideration of that--and things like Gardner's multiple intelligences and the like--I even question my own claim. Still, half the time I think I'm being partly facetious, other times not so much.
 
I had a brief encounter (not that kind, if anybody' old enough to remember) with a chat bot today. It was articulate, it organized relevant material efficiently and effectively, and exhibited what in a human would be sensitivity to a person's feelings. It sure seems smarter than a lot of humans these days. But that's down to the culture, innit?
 
Jeeves said:
I had a brief encounter (not that kind, if anybody' old enough to remember) with a chat bot today. It was articulate, it organized relevant material efficiently and effectively, and exhibited what in a human would be sensitivity to a person's feelings. It sure seems smarter than a lot of humans these days. But that's down to the culture, innit?
Click to expand...
Except when it's wrong--and it's wrong a lot. A few weeks ago, AI told me that 20 percent of Americans are underweight. But, yeah, most people are wrong far more often than AI is, so there's that to consider.

Somewhat understandably, these assessments of the capabilities of AI (and LLMs generally, of course) often focus solely upon language contingent matters. AI does other stuff too. One of my go-tos for evaluating AI is the state of AI drumming, and AI drumming is far more prevalent than one might suppose (though not within actual good music, of course). AI drumming has gotten better, but it still sucks. I'm a crap drummer--from most to least skilled, my musical proficiency goes thusly: keyboards > reeds > tuned percussion > strings > drums--and I'm still a better drummer than even the best AI module. Maybe not technically, but as far as feel goes.
 
parmalee said:
Except when it's wrong--and it's wrong a lot.
Click to expand...
I've been hearing that from people who use it a lot. I don't ask it for factual information: I go to to original sources whenever possible. And I can't get excited about its artwork. But then, I didn't see much point in getting elephants to paint pictures, either.
Still, it will become a lot more useful after the DOE is dismantled.
 
You must log in or register to reply here.
Back
Top