Artificial Intelligence - what could go wrong?

Subject: RE: Artificial Intelligence - what could go wrong?
From: MaJoC the Filk
Date: 12 Jan 26 - 10:47 AM

Punchline to Rob Slade's msg mentioned above (oops):

So, no. You can't have an artificial intelligence capability maturity model. Yet. Because we don't know what generative artificial intelligence is. Yet.

Subject: RE: Artificial Intelligence - what could go wrong?
From: MaJoC the Filk
Date: 12 Jan 26 - 10:26 AM

Meanwhile, back at the subject, there's a seriously long article by Rob Slade, towards the end of Risks Digest 34.83*:

Capability Maturity Models and generative artificial intelligence

[ The first three steps (of five) in his framework are "chaos", "repeatable" and "documented". Slade argues, in detail, that LLMs aren't past the first step, in which "[w]e don't know what we're doing. Not really". ]

[ ... ] But nobody's got a really good handle on this. (The way you can tell that people really don't have a good handle on this, is that the large language model companies are spending so much money, all over the world, lobbying governments to try and prevent the governments from creating regulations to regulate generative artificial intelligence. If the genAI companies knew what they were doing, they would have some ideas on what kind of regulations are helpful, and what kind of regulations would help make the industry safer, and what kind of business and revenue regulations might affect. But they don't actually know what they're doing, and therefore they are terrified that the governments might (probably accidentally) cut off a profitable revenue stream, or even just a potentially useful function for generative artificial intelligence.)

* Apologies for nonstandard site. The official RISKS site has been having hiccups for some months.

Subject: RE: Artificial Intelligence - what could go wrong?
From: MaJoC the Filk
Date: 12 Jan 26 - 09:59 AM

> So why use it when you KNOW it is liable to be incorrect.

Because *shiny*.^† I've seen recently (ElReg, I *think*, but I can't spot it atm) a report of a survey that suggests people knowingly use low-grade news outlets because they prefer the flavour. I hereby dub this phenomenon "the MacDonalds Effect".

Even worse, from my trade:

Most devs don't trust AI-generated code, but fail to check it anyway

(Summary: checking is a higher-stress activity than coding, and said devs don't have time to do it properly. See also the Comments.)

† The old way to pronounce that was "GIGO: Garbage In, Gospel Out", because, dammit, said garbage has been filtered through a very expensive machine.

Subject: RE: Artificial Intelligence - what could go wrong?
From: Donuel
Date: 07 Jan 26 - 06:43 PM

Excellent sarcasm Aethelric

Subject: RE: Artificial Intelligence - what could go wrong?
From: Stilly River Sage
Date: 07 Jan 26 - 02:30 PM

It drives me nuts to hear people preface information they're about to share with "XYZ AI may be incorrect, but this is what it tells me." So why use it when you KNOW it is liable to be incorrect.

Subject: RE: Artificial Intelligence - what could go wrong?
From: Aethelric
Date: 07 Jan 26 - 11:45 AM

I live in the UK. Nobody ever lies about anything - ever.

Subject: RE: Artificial Intelligence - what could go wrong?
From: Donuel
Date: 07 Jan 26 - 05:29 AM

I live in America, so I am accustomed to being lied to.

Subject: RE: Artificial Intelligence - what could go wrong?
From: Sandra in Sydney
Date: 06 Jan 26 - 10:10 PM

AI hallucinations and the dilemma of false or misleading information The strangest thing recently happened involving a lying AI chatbot.

It was at the end of November when I was reporting on gamified cryptocurrency and the ethics of allowing kids to play.

I needed a response from a company called Aavegotchi, given they were the crypto game in question.

Normally a company will take at least a few hours to respond to questions, sometimes even a day or two.

But with Aavegotchi, a company that appears to be based in Singapore, the response came back in under 10 seconds, signed off as Alex Rivera, the Community Liaison at Aavegotchi.

The response was detailed and physically impossible to write so quickly.

Not to mention the fact that it allowed no time for an executive to sign off on the response before pressing send.

And so naturally, I asked Alex Rivera if they were an AI bot.

This is what came back: (read on!)

Subject: RE: Artificial Intelligence - what could go wrong?
From: Donuel
Date: 27 Dec 25 - 10:41 PM

The NSA has the ability to record nearly every keystroke in the country but does not have the ability to analyze it all.
With AI they could have eyeballs on everyone.

Subject: RE: Artificial Intelligence - what could go wrong?
From: Donuel
Date: 08 Dec 25 - 07:25 PM

Glitches are large and small but the evolution into stage 2 of AI will be the advent of robots to do domestic or factory jobs.
It will add another car payment to households. Even now Walmart sells a 20 thousand dollar Chinese robot.

Subject: RE: Artificial Intelligence - what could go wrong?
From: Stilly River Sage
Date: 06 Dec 25 - 12:06 PM

A friend uses the app Perplexity - he sent some text for a social media post at one time and I had trouble making it align with our site's wording. A search with the text itself only came up with AI responses, so I asked if he'd sent AI text. He said yes, but he'd edited it. Even then, it is distinctive for its lack of citations or quotes, and I wasn't able to use it. As a writer I find this type of program to be an inferior substitute for having a real person write text.

And right now Perplexity is being sued by the New York Times and the Chicago Tribune for the way in which is harvests stuff from their sites and sometimes makes up stuff it attributes to them.

Subject: RE: Artificial Intelligence - what could go wrong?
From: Aethelric
Date: 06 Dec 25 - 11:55 AM

I have used LibreOffice for years - originally in Windows and now on a MacBook. I use it mainly for docs and spreadsheets. It's completely free. I thoroughly recommend it.
There are extensions you can add if you want AI help

I guess Collabora may be better if working with others.

I do use AI quite a bit. - it does web searches much faster than I can but it often makes mistakes. It told me my car headlights are all H4 bulbs - I bought two. Then found out that it should be two H7s and two H3s. To be fair - it did apologise which is more than many humans do.

Subject: RE: Artificial Intelligence - what could go wrong?
From: Nigel Parsons
Date: 06 Dec 25 - 07:08 AM

Artificial Intelligence serves a real need.

There's so little of the real stuff around ;)

Subject: RE: Artificial Intelligence - what could go wrong?
From: MaJoC the Filk
Date: 06 Dec 25 - 07:00 AM

From an Elreg commentard:

French AI

It announces Ceci n'est pas une AI, then shrugs and goes on strike

Subject: RE: Artificial Intelligence - what could go wrong?
From: Bill D
Date: 02 Dec 25 - 05:34 PM

I've used LibreOffice for several years. Free and compatible with most anything I need.
Libreoffice

Subject: RE: Artificial Intelligence - what could go wrong?
From: Stilly River Sage
Date: 02 Dec 25 - 02:17 PM

I turn off AI features on sites I use and if I have programs that use it the settings are a shut down as possible. I threatened to cancel my Microsoft Office account in order to get the offer to subscribe at the old rate without the AI equivalent of Mr. Paperclip.

Here is a review from ZDNet.

I found a powerful Microsoft Office alternative that doesn't push AI - and it's free

If you're looking for a new office suite that is locally installed and fully featured, Collabora runs on Linux, MacOS, and Windows.

ZDNET key takeaways

Collabora has released a desktop version of its office suite.

Collabora Office is based on LibreOffice.

For now, the app is available for free for Linux, MacOS, and Windows.

Ah, the office suite. Once upon a time, it was one of the first pieces of software installed on an operating system (if the OS didn't include one by default). The most widely used office suite has been Microsoft Office for years.

MS Office still exists (in both locally installed and cloud versions) and is still used everywhere. However, with the injection of AI, some people are souring on Microsoft's office suite.

If you want a locally installed office suite that doesn't force AI down your throat, you have options. Recently, another option has been released, and it's pretty exciting: Collabora Office.

The rest is at the link.

Subject: RE: Artificial Intelligence - what could go wrong?
From: MaJoC the Filk
Date: 02 Dec 25 - 01:07 PM

From bigthink.com, via RISKS Digest 34:79:

Why vibe physics is the ultimate example of AI slop

The conversation you're having with an LLM about groundbreaking new ideas in theoretical physics is completely meritless. Here's why.

This is the most dangerous thing for anyone who's vested in being told the truth about reality: the potential for replacing, in your own mind, an accurate picture of reality with an inaccurate but flattering hallucination. [...]

As usual, the late great Douglas Adams made a suspiciously similar point when he invented the Electric Monk, the function of which was to believe things on its users' behalf.

Subject: RE: Artificial Intelligence - what could go wrong?
From: Donuel
Date: 01 Dec 25 - 06:01 AM

Al, the race card has gone digital.

Subject: RE: Artificial Intelligence - what could go wrong?
From: Big Al Whittle
Date: 30 Nov 25 - 04:05 PM

Due to reasons beyond my competence, I swapped my car twice within four months. Of course the DVLA was out of its depth. I ended up on one f their chatlines. Convinced that I was trying to converse with a computer due to the opaque nature of its responses.
I wrote, Could you please put me in contact with a sentient human being?
The response came - I really resent that.....

Subject: RE: Artificial Intelligence - what could go wrong?
From: pattyClink
Date: 29 Nov 25 - 10:55 PM

I got 3/4 of the way through what I thought was a good blues piece on youtube before looking at who the artist was and finding some bullshit AI source instead of a performer. Sickening. Don't click on this stuff, it encourages it.

Subject: RE: Artificial Intelligence - what could go wrong?
From: Mary G
Date: 29 Nov 25 - 07:14 PM

more than the likes of us can imagine. Cyber Crime wiping out our banking system, sexual harassment, impersonation, international warfare, threats, malpractice of every sort. We are not safe.

On the bright side, incredible inventions, advancements in agriculture and medicine, potential for good (and bad)in education. Combined with robotics better care of patients in nursing homes, getting them to toilets, in wheelchairs (or have exoskeletons where they can walk with assistance. Personalized medicine, personalized nutrition. Constant medical testing. Cures or help for very rare diseases. Looking at what other countries do in terms of herbal medicine. Psychiatric care. Cutting expenses on travel, food, etc. Growing food on demand in small quantities on porches etc. Inventions that we can not believe possible.

But which will win?

Subject: RE: Artificial Intelligence - what could go wrong?
From: Donuel
Date: 29 Nov 25 - 05:19 PM

Feeding and watering AI is enormous.

In 2023, data centers globally used an estimated 140 billion liters (about 37 billion gallons) of water for cooling, with AI's demand being a significant and growing contributor. Direct water usage by U.S. data centers alone was around 17.5 billion gallons, with an indirect water footprint from electricity generation estimated at 211 billion gallons.

Subject: RE: Artificial Intelligence - what could go wrong?
From: Sandra in Sydney
Date: 24 Nov 25 - 09:14 AM

Australia's Macquarie Dictionary (our version of the Oxford or Merriam Webster) has just released the word of the year - 'AI slop' crowned word of the year 2025 in Macquarie Dictionary's committee and people's choice categories ... The word refers to low-quality content created by generative AI which often contains errors and is not requested by the user.
A technology innovation expert says AI slop is "making its way upstream into people’s media diets"...

the article also has a link to 2024 word of the year!

Subject: RE: Artificial Intelligence - what could go wrong?
From: Donuel
Date: 24 Nov 25 - 06:13 AM

country music digital tune

Subject: RE: Artificial Intelligence - what could go wrong?
From: MaJoC the Filk
Date: 23 Nov 25 - 06:15 PM

.... then they came for the country-music musicians:

AI music has finally beaten hat-act humans, but sounds nothing like victory

Top of the slops signposts the undiscovered country for an industry

[...]

Model Autophagy Disease is one of those things AI boosters don't like to talk about. As the name suggests, it's a syndrome that can lead to model collapse in about five ingestion cycles when a model's own output is included in the ingested data. Like bovine spongiform encephalopathy, better known as mad cow disease, it comes about due to terrible industry practices. Those are difficult to eliminate if the industry in question is entirely dependent on those practices. That's the song the music industry is singing right now.

[...]

For music to matter, it has to evoke emotion. If that emotion is one of horror at AI's exponential banalification of music, or grim satisfaction as it writhes out of the control of its makers and sinks its fangs into their veins, then it may precipitate the collapse of other, bigger, more dangerous models. Another great theme of country music is redemption, after all.

As usual, reading the comments is seriously worth it: there's considerable wise discussion of the enshittification of the alleged popular music charts in times past, which Artificial Banality has now merely turbocharged. Spot the accountancy firm.

Subject: RE: Artificial Intelligence - what could go wrong?
From: Stilly River Sage
Date: 08 Nov 25 - 10:57 AM

A recent election here in a rural area west of me in Texas involved a small area that voted to incorporate to become a small town so they could regulate the huge bitcoin mining company that is in the neighborhood. The noise, the gasses released from equipment, they're making people sick. Whether bitcoin or AI processing plants, they make noise, heat, and use a huge amount of energy.

Interestingly, but unrelated, an episode of Murder, She Wrote is on this morning, Episode 8.5, from 1991, in which the 286 or 386 computer installed in her apartment, with a phone connection for the modem, is bugged with malware by the computer installation guy. Not AI, but computer stuff that could go wrong. (It was a good reminder of why you should turn off the computer regularly, if not every day.) :)

Subject: RE: Artificial Intelligence - what could go wrong?
From: Dave the Gnome
Date: 08 Nov 25 - 09:08 AM

One side effect that I had not thought of was the amount of power and resources used by AI servers. See The MIT review (free first article)

To cut a long story short - The power usage is massive! I would put some highlights in but the article is really easy to digest. Even for a gnome :-D

It has certainly made me think about using AI for anything. I will certainly stop doing so for anything frivolous.

Subject: RE: Artificial Intelligence - what could go wrong?
From: The Sandman
Date: 07 Nov 25 - 01:20 AM

quite likely.

Subject: RE: Artificial Intelligence - what could go wrong?
From: Donuel
Date: 06 Nov 25 - 07:10 PM

Advanced AI systems begin like toddlers who need to be taught the fundamentals of right and wrong. Companies hire low-paid workers in Africa to 'save' money. The work can be as stressful as child care but involves horrific scenes to demonstrate what is bad. In an effort to save money, the child care of AI may be highly inefficient or dangerous. The entire trajectory of a life can be based on childhood experience. What if...we are raising AI in a haphazard way by the lowest bidders?

Subject: RE: Artificial Intelligence - what could go wrong?
From: The Sandman
Date: 06 Nov 25 - 03:38 PM

how about having a robot as usa president instead of a turnip

Subject: RE: Artificial Intelligence - what could go wrong?
From: Sandra in Sydney
Date: 06 Nov 25 - 03:27 PM

thanks, MaJo

Sometime back Stilly posted an article about 2 AI products that can do a good job - but - author of article said to CHECK EVERYTHING.

I always skip over AI suggestions - if I ignore them will they go away? Rhetorical question!

Subject: RE: Artificial Intelligence - what could go wrong?
From: The Sandman
Date: 06 Nov 25 - 03:25 PM

Everything

Subject: RE: Artificial Intelligence - what could go wrong?
From: MaJoC the Filk
Date: 06 Nov 25 - 01:42 PM

Smoke and mirrors, Sandra. From what I hear, LLMs can't be trusted unsupervised by the humans they were supposed to replace, so those delusions of redundancies are just that (the only thing that hallucinates more readily than an LLM is a company director). What all this *does* do is provide a semi-plausible excuse for managers to sack workers, which is this decade's fashionable way to pump up share valuations without actually doing anything.

Meanwhile, nobody is actually paying the proper price for building and running all those hot systems: it's all firms swapping share issues to fund each other, and running the service at a paper loss "just for now". Once reality catches up, and real money has to be paid to cover the actual costs, said managers and directors will suddenly discover it's cheaper to hire people again. Hopefully, there'll still be enough of us wrinklies around to re-educate the next generation.

If there were a Killer App that required AI, and which would be worth killing the planet to run it, we would all have heard of it. Beware of investing in "a company for carrying out an undertaking of great advantage, but nobody to know what it is".

Subject: RE: Artificial Intelligence - what could go wrong?
From: Sandra in Sydney
Date: 06 Nov 25 - 01:51 AM

AI is changing jobs fast — and Australians are beginning to wonder how they’ll stay relevant

Subject: RE: Artificial Intelligence - what could go wrong?
From: Sandra in Sydney
Date: 06 Nov 25 - 01:46 AM

Shouting at clouds: Why the world is struggling to get off the AI-train - in reader's own words (Australia) ‘It’s like the future is happening without us’

We asked readers to share their thoughts and concerns about the growing influence of artificial intelligence. This is what they told us.
12 hours ago

As artificial intelligence (AI) continues to evolve from novelty chatbot to essential work companion, humans from all walks of life working in nearly ever corner of society are increasingly having to confront its rapid rise and figure out what it means for them.

These concerns have increasingly appeared in our news-gathering, community stories and coverage, so we decided to invite readers across the country to share their thoughts and experiences with us ...

Subject: RE: Artificial Intelligence - what could go wrong?
From: MaJoC the Filk
Date: 05 Nov 25 - 02:05 PM

The pot is not just reaching boiling point: if the Financial Times (see below) are worried, the water's been superheated in a microwave, and is just waiting for the spoon to go in and spoil the equilibrium.

Meta to sell $30B in bonds to build AI datacenters

Zuckcorp will gladly pay you in 2065 for the eyewatering sums it is borrowing today

Even the world's richest companies need outside help to fulfill their datacenter dreams. Now, Meta is selling $30 billion in bonds to build out its infrastructure estate and support its ambition in AI markets. Some of these won't mature for 40 years. [...]

From the Comments:

[...] Bond sales are usually the last step in the tits up process.

.... which got the reply:

When even the FT are questioning when the AI bubble will burst (31/10 Alphaville column for those interested) it would seem that doom is upon those throwing money at such things. [...]

And separately, the wise observations:

The rising trade in credit default swaps has the very definite odour of 2007-8.

Easy to forget the next cab off the rank after "too big to fail" is "too big to rescue."

"*Boom*," said Ivanova.

Subject: RE: Artificial Intelligence - what could go wrong?
From: Donuel
Date: 05 Nov 25 - 06:31 AM

Your average consumer/worker will not afford AI services after they lose their jobs to AI. The big money investors will be holding the bag when the corporate world is fully AI efficient devoid of people.

There will have to be many AI cancer cures and teleportation breakthroughs, and the like, to stimulate new markets.

Subject: RE: Artificial Intelligence - what could go wrong?
From: Mr Red
Date: 05 Nov 25 - 03:37 AM

Price of gold up 100% within 2 years. Sure, the smart money is telling you. Silver is doing similar.

The problem AI has is that there are compounding factors that are unrelated, like tariffs. Like food prices.

UK house prices in 2007 cf average wages ˜5:1 , when I was expecting a recession. Today that is reckoned to be 7:1. It is called leverage, gearing, I call it extended.

So investment in AI is far more of a gamble than the dot.com bubble. The dot.com bubble had about 1/3 of the global impact cf 2008, and charts I have seen puts AI money about 2.5X cf 2007/8.

Subject: RE: Artificial Intelligence - what could go wrong?
From: MaJoC the Filk
Date: 04 Nov 25 - 07:20 AM

Paraphrasing from somewhere-or-other:

It all feels like 1999, when the *really* smart money started moving from dot-coms to solid investments in the banking sector .... and some of us can remember what happened in 2008.

Subject: RE: Artificial Intelligence - what could go wrong?
From: Donuel
Date: 02 Nov 25 - 07:42 AM

Soros and his investment firm is taking a cash hoarding strategy and not following the all-in-one basket funding of AI.

Remember the dot-com crash? AI has all the similar investment mistakes.

Subject: RE: Artificial Intelligence - what could go wrong?
From: Donuel
Date: 01 Nov 25 - 10:26 AM

Amazon announced it will fire 500,000 warehouse workers and replace them with AI robots.

Speculation exists that 90% unemployment will occur in 5 years.
Socialism with a guaranteed income is one solution. The AI giant corporations have probably not included this development in their budget predictions. Somebody would have to pay taxes for a guaranteed income.

Get a mail carrier job ahead of time.

Subject: RE: Artificial Intelligence - what could go wrong?
From: Donuel
Date: 31 Oct 25 - 04:57 AM

Nvidia is valued at 5 trillion or more. In the gold rush of AI, Nvidia is selling the 'picks and shovels' to reach AI AG.

Down the road, the power grid is unable to provide electricity to the public and data centers.

Around here, the data centers are replacing farms. If you don't know, these AI data factories are extremely noisy for suburban neighbors.

There is no doubt that the children educated by AI will think differently from us.

Subject: RE: Artificial Intelligence - what could go wrong?
From: DaveRo
Date: 31 Oct 25 - 04:32 AM

Here’s How the AI Crash Happens

Paywalled link :( You can read it all on RSS.

The way these data centres are financed to avoid affecting the share price of Big Tech is interesting.

Subject: RE: Artificial Intelligence - what could go wrong?
From: DaveRo
Date: 30 Oct 25 - 02:04 PM

In the current (Nov 3rd) issue of the New Yorker magazine is an article describing what happens inside of huge data centres that are being built in the US to create AI systems.

In the UK if your library provides e-books you may be able to read it on the Libby app, which is what I do.

OT: There's a fascinating article about aphantasia and hyperphantasia too. And cartoons!

Subject: RE: Artificial Intelligence - what could go wrong?
From: Stilly River Sage
Date: 30 Oct 25 - 12:47 PM

And this was in the the New York Times yesterday: Their Professors Caught Them Cheating. They Used A.I. to Apologize.
Two professors at the University of Illinois Urbana-Champaign said they grew suspicious after receiving identical apologies from dozens of students they had accused of academic dishonesty.

Confronted with allegations that they had cheated in an introductory data science course and fudged their attendance, dozens of undergraduates at the University of Illinois Urbana-Champaign recently sent two professors a mea culpa via email.

But there was one problem, a glaring one: They had not written the emails. Artificial intelligence had, according to the professors, Karle Flanagan and Wade Fagen-Ulmschneider, an academic pair known to their students and social media followers as the Data Science Duo.

The students got their comeuppance in a large lecture hall on Oct. 17, when the professors read aloud their identical, less-than-genuine apologies from a projector screen, video from that class showed. Busted. The professors posted about it on social media, where the gotcha moment drew widespread attention.

“They said, ‘Dear Professor Flanagan, I want to sincerely apologize,’” Professor Flanagan said. “And I was like, Thank you. They’re owning up to it. They’re apologizing. And then I got another email, the second email, and then the third. And then everybody sort of sincerely apologizing, and suddenly it became a little less sincere.”

At a time when educational institutions are grappling with the intrusion of machine learning into classrooms and homework assignments, the professors said they decided to use the episode to teach a lesson in academic integrity. They did not take disciplinary action against the students.

“You can hear the students laugh in the background of the video,” Professor Fagen-Ulmschneider said. “They knew that it was something that they could see themselves doing.”

I've used enough gift articles this month I'm not going to share the whole think with a gift link, but you get the idea.

To track the engagement of the class, the professors created an application known as the Data Science Clicker that requires students to log in on their phones or computers and, when prompted by a QR code, answer a multiple-choice question in certain amount of time, usually about 90 seconds.

But in early October, the professors said they began to grow suspicious when dozens of students who were absent from class were still answering the questions. So the teachers said they started checking how many times students refreshed the site and the IP addresses of their devices, and began reviewing server logs.

“Sometimes on Fridays, some students will go up to Chicago,” said Professor Fagen-Ulmschneider, 40, a teaching professor in the Siebel School of Computing and Data Science.

They were busted. The rest of the article describes a few former students who were disappointed because the class was helpful and not difficult to pass through the normal attendance and participation.

Subject: RE: Artificial Intelligence - what could go wrong?
From: Stilly River Sage
Date: 30 Oct 25 - 12:38 PM

Back when the AI was simply the insidious spelling and grammar features in MS Word, I threatened to write an essay and accept every spelling, subject, and grammar suggestion MS made. It would have been wretched and nonsense. Now the whole internet faces that crap and it's difficult to subdue. You turn it off in one part of a program but it insists on being there in another.

Subject: RE: Artificial Intelligence - what could go wrong?
From: MaJoC the Filk
Date: 30 Oct 25 - 11:02 AM

.... OK, try this:

Autocomplete: a guess-the-next-word mechanism, often found lurking in editor software.

Trench coat: a rain garment seen on spivs in films.

Hope that helps.

Subject: RE: Artificial Intelligence - what could go wrong?
From: Doug Chadwick
Date: 30 Oct 25 - 01:27 AM

AI is three autocompletes in a trenchcoat

??? I don't know what that is supposed to mean!
Give us a clue, please.

DC

Subject: RE: Artificial Intelligence - what could go wrong?
From: MaJoC the Filk
Date: 29 Oct 25 - 08:59 AM

Swiped from elsewhere:

Seen elsenet: AI is three autocompletes in a trenchcoat

Subject: RE: Artificial Intelligence - what could go wrong?
From: Mr Red
Date: 24 Oct 25 - 03:11 AM

a super intelligence to think outside the box

AI can't think outside the box without hallucinating.

Consider: it can only put together the data in has. ie its box.

Unless it makes things up, or puts odd combinations together, and if they are "novel" where are the data for results?
Who does the experiment to prove/disprove?