• Security incident: ISF was recently accessed by intruders. Please change your password, and change it anywhere else you used it. Read more

Merged Artificial Intelligence

Andy_Ross said:
Tech stocks tumble on concerns over AI spending

Financial markets received a sharp wake-up call on Tuesday following a sudden wave of selling in major technology shares, triggering widespread doubt over the sustainability of the AI boom.

The tech-focused Nasdaq index fell about 2% alongside international chipmakers, reigniting fears that dizzying market valuations have finally run out of momentum after a relentless three-month climb.

www.bbc.co.uk

Tech stocks tumble on concerns over AI spending

A sudden wave of selling in major tech shares triggers doubt over the sustainability of the AI boom.
www.bbc.co.uk www.bbc.co.uk
Click to expand...
Have a few "analysts" cracked open an old textbook and read about P&L and ROI?
 
In the May issue of Scientific American, there is an article called "How AI and Human Judgment Differs". In the online version, it is called "What we risk when we confuse AI and human intelligence". I don't know if it is behind a paywall.

In the article, human reasoning is compared to AI reasoning by subjecting human test persons and AIs to the same task. For instance, they should evaluate the credibility of a number of news sources, and justify their decision.
We observed that even when models reached conclusions similar to those of human participants, their justifications consistently reflected patterns drawn from language (such as how often a particular combination of words coincided and in what contexts) rather than references to external facts, prior events or experience, which were the factors that humans considered.
Click to expand...
I find it interesting that AIs can use combinations of words and context, and draw "similar" conclusions as humans, even though they do not have external facts, prior events, or experiences to draw upon.

Another test was reasoning about moral dilemmas. Humans reasoning about these draws on norms, social expectations, emotional responses, and culturally shaped intuitions about harm and fairness.
We found that a language model can reproduce this form of deliberation fairly well. The model provides statements that mirror the vocabulary of care, duty or rights. It will present causal language based on patterns in language, including “if-then” counterfactuals.
Click to expand...
Having established that AIs are good at moral reasoning, the authors hasten to say:
But it’s important to note that the model is not imagining anything or engaging in any deliberation; it is just reproducing patterns in people’s speech or writing about these counterfactuals. The result can sound like causal reasoning, but the process behind it is pattern completion, not an understanding of how events produce outcomes in the world.
Click to expand...
The authors seem to want to ensure that we should certainly not think that AIs can be intelligent, because the authors know how LLMs work, and that can only lead to a semblance of intelligence:
This gap between what models seem to be doing and what they are in fact doing is what my colleagues and I call epistemia: a situation when the simulation of knowledge becomes indistinguishable, to the observer, from knowledge itself. Epistemia is a flaw in people’s interpretation of these models in which linguistic plausibility is taken as a surrogate for truth. This error happens because the model is fluent, and fluency is something human readers are primed to trust.
Click to expand...
The only thing I agree about in this warning is the tendency for AIs to "hallucinate", but the authors use this to claim that AIs can never hold the "truth", whatever that means:
The danger here is subtle. It is not primarily that models are often wrong—people can be, too. The deeper issue is that the model cannot know when it is “hallucinating,” because it cannot represent truth in the first place. It cannot form beliefs, revise them or check its output against the world. It cannot distinguish a reliable claim from an unreliable one except by analogy to prior linguistic patterns. In short, it cannot do what judgment is fundamentally for.
Click to expand...
But rest assured, the authors do end with "None of this implies that LLMs should be rejected." We should just keep in mind that they although they reason just like us, but we should not trust their eloquence …
 
steenkh said:
In the May issue of Scientific American, there is an article called "How AI and Human Judgment Differs". In the online version, it is called "What we risk when we confuse AI and human intelligence". I don't know if it is behind a paywall.

In the article, human reasoning is compared to AI reasoning by subjecting human test persons and AIs to the same task. For instance, they should evaluate the credibility of a number of news sources, and justify their decision.

I find it interesting that AIs can use combinations of words and context, and draw "similar" conclusions as humans, even though they do not have external facts, prior events, or experiences to draw upon.

Another test was reasoning about moral dilemmas. Humans reasoning about these draws on norms, social expectations, emotional responses, and culturally shaped intuitions about harm and fairness.

Having established that AIs are good at moral reasoning, the authors hasten to say:

The authors seem to want to ensure that we should certainly not think that AIs can be intelligent, because the authors know how LLMs work, and that can only lead to a semblance of intelligence:

The only thing I agree about in this warning is the tendency for AIs to "hallucinate", but the authors use this to claim that AIs can never hold the "truth", whatever that means:

But rest assured, the authors do end with "None of this implies that LLMs should be rejected." We should just keep in mind that they although they reason just like us, but we should not trust their eloquence …
Click to expand...
Your first quote let the cat out of the bag, it admits that LLM "reasoning" gets things right by chance.
 
Gulliver Foyle said:
Your first quote let the cat out of the bag, it admits that LLM "reasoning" gets things right by chance.
Click to expand...
That is not how I see it. If it was by chance they would not have started poo-pooing the results, but immediately have said "this is what happens when you just pick the most likely word as the next word". But it is quite correct that LLMs do not have the experience that humans have, especially experience that is not based on writing, and hence the LLMs argue in another way than humans, and get similar results.

The moral questions are different, because physical or oral experience is not so important in deciding moral questions.
 
Over prior years, we didn’t pay as much attention as we should have to the experience of our most knowledgeable engineers that have been with us through many product cycles.​
​
“Mistakenly, we thought that by just introducing artificial intelligence and ingesting the design requirements that we had, that that would produce a high-quality product.”​
Ford hired AI and sacked humans. It backfired badly
https://www.independent.co.uk/tech/ford-ai-automation-human-workers-b3003787.html
 
This made me laugh out loud

Andy Wingo (@[email protected])

can't help getting astrology × audiophile vibes from llm enthusiasts: "oh no honey you can't use opus 4.8, it's so nerfed, try glm-5.2 or the codex 5.5, they have a warm gemini sound, very well rounded on details and tone. but don't use google's gemini models, they are really leos, they have...
mastodon.social mastodon.social
 
Ford rehires human engineers after AI fails to match quality checks

Ford says it has hired back some human engineers after AI failed to match their skills and experience.

In a bid to reap the benefits of the tech, which developers claim can cut costs and boost productivity, the US carmaker adopted it across some parts of its operations including for quality checks.

But, according to Bloomberg, external, its executives said the firm has rehired more than 300 "veteran" quality inspectors in recent years to make up for the pitfalls of automated systems.

www.bbc.co.uk

Ford rehires human engineers after AI fails to match quality checks

The car-maker found AI quality checks failed to match the skill of veteran technicians.
www.bbc.co.uk www.bbc.co.uk
 
I was making hotel reservations on line. All at places we've stayed before.
First one, no problem.
Second one's website defeated me. So I called their number and Dominic was really helpful, got it all done.
Third one's website defeated me, so I called their number. Answered by a very nice African-American sounding woman, I forget her name. But she did give one. It took a couple of minutes and a couple of weird questions and answers for me to realize that "she" was AI. Finally I just said "You are an AI, let me talk to a human being." And she did.
The human was somewhere on the other side of the Pacific Ocean and had quite limited English, but got the job done.
 
www.wheresyoured.at

Where Are All The Data Centers?

If you liked this piece, please subscribe to my premium newsletter. It’s $70 a year, or $7 a month, and in return you get a weekly newsletter that’s usually anywhere from 5,000 to 18,000 words, including vast, detailed analyses of NVIDIA, Anthropic and OpenAI’s finances,
www.wheresyoured.at www.wheresyoured.at

article on data center construction

So, I have two very simple questions to ask: how long does it take to build a data center, and how much data center capacity is actually coming online?

These simple questions are surprisingly difficult to answer. There exists very little reliable information about in-progress data centers, and what information exists is continually muddied by terrible reporting — claiming that incomplete projects are “operational” because some parts of them have turned on, for example — and a lack of any investor demand for the truth. Hyperscalers do not disclose how many data centers they’ve built, nor do they disclose how much capacity they have available.
Click to expand...
 
Darat said:
I've been following him since earlier in the year when I started to try and find information on the proposed and promised massive new datacentres and I couldn't find any that were finished and many had not even broken ground. At the moment he is talking a lot of sense (I know this because he agrees with me).
Click to expand...

must be if we all agree
 
Erin Brockovich is on the case!

"AI Data Centers Across the United States

A map of major AI data centers in the U.S. that are either operational or under construction, overlaid with locations where community members have emailed in concerns. Click any marker for details."

brockovichdatacenter.com

Brockovich Data Center Reporting – U.S. AI Data Center Awareness & Issue Map

Interactive map of major AI data centers across the United States — built or being built. Understand the community impact and report issues in your area.
brockovichdatacenter.com
 
You must log in or register to reply here.

Share this page

ISF - Join now!

Every member here is approved by hand. No bots, no spam, just people who care about evidence and honest debate.

Membership is free!

Create your free account

Back
Top Bottom