Artificial Intelligence - what could go wrong?

Subject: RE: Artificial Intelligence - what could go wrong?
From: MaJoC the Filk
Date: 18 Jan 26 - 06:11 AM

Bingo: the RISKS Digest site at Newcastle is back up to date again. Herewith two consecutive articles from RISKS 34:83:

Capability Maturity Models and generative artificial intelligence

(see above)

The AI boom is based on a fundamental mistake (The Verge)

Large language mistake

Cutting-edge research shows language is not the same as intelligence. The entire AI bubble is built on ignoring it.

[ ... ]

Last year, three scientists published a commentary
https://archive.ph/o/BV16q/https://gwern.net/doc/psychology/linguistics/2024-fedorenko.pdf
in the journal Nature titled, with admirable clarity, “Language is
primarily a tool for communication rather than thought.” [ ... ]

Subject: RE: Artificial Intelligence - what could go wrong?
From: Sandra in Sydney
Date: 22 Jan 26 - 02:06 AM

well, you do get pretty pictures!!!

Tour website's AI sends visitors to Tasmanian sites that do not exist An AI-generated article on a travel booking website has sent tourists to a remote location in Tasmania's north-east, looking for hot springs that do not exist.

Australian Tours and Cruises has admitted the AI technology it uses to create content and articles to help drive bookings has "completely messed up".
What's next?

The company has said it will review all of its AI-generated content, which is produced by a third party ...

Mr Hennessy said that while all posts were normally reviewed before being posted, some had been made public by mistake while he was out of the country ...

Subject: RE: Artificial Intelligence - what could go wrong?
From: Donuel
Date: 22 Jan 26 - 06:05 AM

What's next?

people might work.

Subject: RE: Artificial Intelligence - what could go wrong?
From: MaJoC the Filk
Date: 23 Jan 26 - 10:03 AM

> well, you do get pretty pictures!!!

As it happens, Sandra, I saw an article some years ago about automated uglification using machine learning: teach the ML on pictures from one or more of the horror mags, then get them to "enhance" real-life photos to re-render them in that style. The most horrifying bit was that it was easy to tell that the image of a female English politician (Theresa May?) was processed this way, but that I found it difficult to tell that the one of Agent Orange wasn't the original.

Subject: RE: Artificial Intelligence - what could go wrong?
From: Stilly River Sage
Date: 30 Jan 26 - 12:44 PM

This interesting essay by James O'Sullivan was shared by an academic friend who writes about AI in the classroom.

I am so f**king sick of AI slop

Extracted from it:

What is alarming, however, is the sheer willingness of people to put their own names to any old slop the machine produces.

Take the rise of ‘vibe coding’, for example, which most tech bros represent as some kind of futuristic workflow. Am I the only one who thinks that there is a peculiar arrogance in a ‘developer’ who takes pride in their inability to write or understand the syntax that underpins their own creation? We are seeing software deployed into the wild that is little more than a patchwork of pasted suggestions, a digital Frankenstein’s monster stitched together by someone who has no concept of the structural integrity of the code base. They might feel as though they know what they have built, but they can’t comprehend the gaping security vulnerabilities they have introduced because they never actually engaged with the logic of the system; they simply prompted a black box until the error messages stopped appearing, creating a technological infrastructure that is terrifyingly insecure.

And further down:

The internet was, in its most idealistic (and yes, maybe naive) conception, a sprawling parlour for human conversation and the exchange of genuine thought. That vision is effectively dead. Open LinkedIn or Reddit (or X, if you really want to wind yourself up) and you will see streams of the same beige, hallucinatory text bearing the chirpy, predictive cadence of ChatGPT, generated by users who could not be bothered to read the content they are putting their name to. They enter a prompt and paste the result, engaging in a pantomime of interaction that benefits no one but the platform’s engagement metrics. It is a hall of mirrors where machines talk to machines while humans look on, increasingly alienated from the very networks built to connect them.

He concludes "I can appreciate the technology for what it is, but I am finding it increasingly difficult to forgive the laziness of the people using it."

Amen.

Subject: RE: Artificial Intelligence - what could go wrong?
From: Donuel
Date: 30 Jan 26 - 05:39 PM

The slop will still have better spelling than the flesh-and-blood moron activists.

A minor point, but I just saw a Trump commercial on CNN that uses AI copy of Trump's voice. I'm surprised the fine print informed people of this.

Subject: RE: Artificial Intelligence - what could go wrong?
From: The Sandman
Date: 01 Feb 26 - 05:17 AM

This interesting essay by James O'Sullivan was shared by an academic friend who writes about AI in the classroom." quote
I disagree, i find the article opinionated and subjective. but we are all entitled to different opinions.

Subject: RE: Artificial Intelligence - what could go wrong?
From: Stilly River Sage
Date: 01 Feb 26 - 12:47 PM

Artificial intelligence researchers hit by flood of ‘slop’
Conferences restrict use of LLMs after surge of low-quality AI-generated papers and reviews

Artificial intelligence researchers are grappling with a problem core to their field: how to stop so-called “AI slop” from damaging confidence in the industry’s scientific work. AI conferences have rushed to restrict the use of large language models for writing and reviewing papers in recent months after being flooded with a wave of poor AI-written content.

Scientists have warned that the surge of low-quality AI-generated material risks eroding trust and the integrity of the sector’s research by introducing false claims and made-up content.

“There is a little bit of irony to the fact that there’s so much enthusiasm for AI shaping other fields when, in reality, our field has gone through this chaotic experience because of the widespread use of AI,” said Inioluwa Deborah Raji, an AI researcher at the University of California, Berkeley. . . . In January AI detection start-up GPTZero published research that found there were over 100 AI-generated errors across 50 papers last year at the Neural Information Processing Systems (NeurIPS) conference, considered the most prestigious place to publish cutting-edge AI research.

Further down the article is "A tell-tale sign is when papers contain hallucinated references in the bibliography, or figures that are wrong, said Dietterich. These users are then banned from submitting papers to arXiv for a while, he added."

 

Subject: RE: Artificial Intelligence - what could go wrong?
From: Bill D
Date: 02 Feb 26 - 09:44 AM

Long but fascinating conversation with founder of AI

Subject: RE: Artificial Intelligence - what could go wrong?
From: Bill D
Date: 02 Feb 26 - 09:46 AM

That conversation is with Jon Stewart being serious.