Microsoft exec finds AI cynicism 'mindblowing'

In a tweet lamenting all the "cynics" unmoved by AI, Microsoft AI boss Mustafa Suleyman demonstrated that Redmond's Reality Distortion Field is running at full power. Suleyman took a break from the Copilot company's San Francisco Ignite shindig to exclaim "Jeez there so many cynics!" and calling it "mindblowing" that people …

[ComicalEngineer]

Reality is an illusion ...

... caused by lack of AI.

[/sarcasm off]

Is M$ living in an alternative universe?

Is the air in their offices contaminated with a powerful hallucogen?

More likely the vision of $$$$ overrides all other considerations.

[Charlie Clark]

Re: Reality is an illusion ...

I think he's just talking to the wrong audience. Lots of programmers have been using various LLM-based tools happily and productively for a while. But these are the kind of people most likely to notice its limitations. Over in the beancounter and coloured crayon offices, it's been a lot easier to convince people how cool this stuff is any why we all must have it.

Fortunately, somewhere else people have been running the numbers over something that increasingly looks like AI power == GPU power == electric power: huge data centres requiring huge numbers of GPUs powered by huge power stations. All of this is going to require massive amounts of capital investment with fairly sketchy plans for repaying it. It certainly looks like today's pricing plans can never cover it.

If anyone bothered to look at the Silicon Valley playbook, they'd see on page 2 "Opportunities" – inefficient or (over-) regulated services. We've already seen what happened when DeepSeek launched a model that was trained at a fraction of the cost ChatGPT. China has both the resources to compete with development, but also reasons to compete on price, and not just the price of electricity. Places like India will also be ruthlessly when it comes to finding the cheapest provider.

[Red Ted]

DeepSeek costs

...DeepSeek launched a model that was trained at a fraction of the cost ChatGPT.

Unsurprisingly DeepSeek was not as cheap as originally suggested.

[Anonymous Coward]

Re: DeepSeek costs

Suleyman's tweet came after Microsoft posted: "Copilot finishing your code before you finish your coffee."

Sir, your AI code just sent our entire Europe member database to the Russian FSB. We are now on the hook for several million in GDPR fines.

[Anonymous Coward]

Re: Reality is an illusion ...

>” Lots of programmers have been using various LLM-based tools happily and productively for a while”

They are very good at convincing management you are busy when in fact you are simply filling time, whilst avoiding do the job you were actually employed to do.

[Not Yb]

Lunchtime, doubly so... Re: Reality is an illusion ...

Programmer types are definitely not the only people who've noticed AI isn't quite what the hype machine claims it to be. Most experts, when chatting with an AI, can find inaccuracies in their field of study. Even good DIY mechanics can find something not quite right.

Almost any expert, except (clearly) some experts in AI.

Most recent 'test' I tried was asking Google Gemini to read some QR codes from a sheet. It read the TEXT on the sheet, and used that to make up a "Direct Digital Decode" feature to create approximations of the most probable URLs instead of actually reading them. It sounded VERY confident that it was reading the QR codes, while the results were completely wrong aside from the host part of the URL.

[Ropewash]

Re: Lunchtime, doubly so... Reality is an illusion ...

But of course it did that. Not even the image specific machines can actually tell you what an image is about, only what pre-defined tags it is able to recognise. They cannot see the whole image AS a whole image. Show it an apple and 90% of the time it'll tell you that's an apple, not that the apple is accidentally in the dairy section of the store.

That an automated guessing engine cannot fully grasp the idea of QR codes is not a surprise.

[Anonymous Coward]

Re: Lunchtime, doubly so... Reality is an illusion ...

Yes, but a human has actual intelligenceand will flag the issue or inform you that something cannot be done - they require chemicals to hallucinate.

[Anonymous Coward]

Re: Reality is an illusion ...

Maybe we should develop distributed AI that is smaller, can be trained to be brilliant at specific tasks but is low power, then we could group them together for complex, multi-disciplinary tasks, we could make it biological and give it the ability to care for itself ... oh hello.

[teknopaul]

Re: Reality is an illusion ...

Hot out of the bit barns this month Copilot now convincingly tell you about it things are impossible when they are really really not.

Every month there is new thing AI does scandalously and irresponsibly wrong.

[Anonymous Coward]

Re: Reality is an illusion ...

Yeah because by the time it's been through the safety and wokeness filter it's had it's balls ripped off and it's knowledgebase mutilated.

If you were the smartest man alive and I invited you to talk at an event, then just before you got out on stage I gave you an impossibly long list of shit you can't say or think about to ensure you can't offend people or impart potentially dangerous knowledge you'd also appear to be dumb and subjective.

Dangerous knowledge filters kind of make sense...but turning an AI into Mr Tumble in case it offends someone is the dumbest thing ever...as a race we really need to toughen up a bit, hurt feelings really aren't a big enough deal to sacrifice the quality of something.

If an AI produces code for me that is impeccable but it calls me a wanker meatsack 2,000 times a day, it's just banter...that's better than it producing bollocks all day because it is preoccupied with preventing itself calling me a wanker 2,000 times a day.

All I can say about people that would potentially be offended by mathematically derived computer generated words is that I pity them. Being offended is a huge waste of time...especially because it is not possible to give offense, you can only take offense at something.

If something is offensive to someone that is their problem.

We know that it is possible for extreme language to become generally unoffensive and part of standard vocabulary because of Scotland. Go up there, and there is a good chance you'll be called a cunt or hear the word cunt several times a day.

There are Scottish people right now reading this post that have considered replying and calling me one...and they can, I won't take offense...especially from the Scottish because it's such an overused word up there that it is basically meaningless coming from a Scot.

A former colleague of mine, over a decade ago now, was once called a "buffoon"...a term that virtually nobody would be offended by...he was mortified and never got over it...I haven't seen him in over 10 years...but I bet it still haunts him. You can't help these people.

[the Jim bloke]

Re: Reality is an illusion ...

.. feeding the troll..

Your entire arguement is based on the pervasive fallacy - or outright marketing lie, that AI has some kind of understanding of what it is 'talking' about.

there is no reason for an AI to call someone a wanker - or a meatsack - or use any kind of offensive address at all - unless it has been trained to do so. the offensiveness is even more artificial than the intelligence. At best the equivalent of desktop wallpaper to 'enhance' the user experience, but more likely someone emulating children using rude words to provoke a reaction.

The fakeness is offensive

the hysterical marketing hype is offensive

The constant intrusion of USE OUR AI NOW!!!! pop ups when I am trying to get stuff done is offensive.

But the chatbots using rude words.. thats just pathetic.

[Not Yb]

Re: Reality is an illusion ...

With the way even Grok is trained, the only way it would be calling the troll-poster a wanker meatsack while coding, is if he told it to.

Much like "Why does Amazon keep showing me these [embarrassing thing here]s?" is easily answered by "because it knows you buy them."

[Anonymous Coward]

Re: Reality is an illusion ...

"that AI has some kind of understanding"

No it doesn't have understanding, but it does have context...and if you filter out everything that might be offensive from a training set, you can massively change the context in unintentional ways.

Do you think it was Google's intention for it's image diffuser to create asian and black nazis? Do you think it was intentionally trained to do that? No it wasn't but because of the attempt to make their model more inclusive and woke that's the result they got.

Everyone is noticing how progressively worse AI is becoming the "safer" it gets.

Swearing was just an extreme example to make my position clearer. Removing content from a training set that might be deemed offensive might remove important information that changes the context of something.

[Anonymous Coward]

Re: Reality is an illusion ...

>>> Yeah because by the time it's been through the safety and wokeness filter

That was the point I realised that the rest of your post wouldn't be worth reading. I did anyway, and I was right.

Here are the filters that really cause problems:

https://www.theverge.com/ai-artificial-intelligence/825675/groks-elon-musk-worship-is-getting-weird

Grok says Musk is smarter than Da Vinci, fitter than LeBron James, better than Jesus, and funnier than Jerry Seinfield (though I'll give him that last one)

[hedgie]

Re: Reality is an illusion ...

Wow, Muskrat is about as mature as I was at 19 when I pranked him by feeding the "learning bot" he was trying to build garbage about how I was god-emperor and stuff like that.

[Tron]

Re: Reality is an illusion ...

The tech sector are so desperate for their latest scam to work. It has brought them to the point that they are functioning like politicians, forcing it on us, blaming us for not liking it and mystified as to why we hate them for it.

[Anonymous Coward]

Re: Reality is an illusion ...

That's basically because they can :(.

If MS code was as regulated as the car industry.. No, wait. My car's ghost braking is according to the garage deemed 'performing as designed' so actually physically dangerous. Bad example.

Sorry, I digressed. If the industry above a certain threshold could actually be made responsible for what it produces we'd see less of this shit, but we pretty much allowed them to declare the quality of what they produce above the law. Only their behaviour can be slightly controlled, but then they just buy, sorry, lobby the relevant politicians and have the law changed.

The original idea was that competition improved the market, but when an entity has some clearly and overwhelmingly locked in customers that competition disappears, well, you see what the results are. Enshittification to the nth degree.

[JWLong]

Re: Reality is an illusion ...

It's very difficult to see the truth when your head is up your ass!.

This is the only reason MicroShit hasn't figuired out that anyone with more than 2 brain cells, DOESN'T WANT THIS SHIT!

I swear to god that whole corporation is full of FUCK'N MORONS..............

[keithpeter]

Re: Reality is an illusion ...

"It's very difficult to see the truth when your head is worried about your assets!."

Modified your pithy statement a little.

MS has invested $lots on this and perhaps the directors of the corporation think that they need to encourage adoption so as to demonstrate their wisdom to shareholders. Sort of like the directors of railway companies did in late Victorian times here in the UK.

The difference is that when the railway bubble burst, we were left with tangible assets that we are still using (and expensively maintaining of course).

[JWLong]

Re: Reality is an illusion ...

I bought M$ stock in '86@¢0.06-0.10/share.

I accrued over 19K shares.

For the past 5 years I've been selling off because of asshole, money grubbing, cheap shit "directors".

"Encourage adoption", jamming it up people's asses because nobody wants it and are screaming in their faces about it.

Fucking blind leading the stupid is what they are.

[Anonymous Coward]

Re: Reality is an illusion ...

Not sure why you're so angry? You presumably made a huge killing on your shares (currently $472.12), and it's precisely that same remarkable growth that the directors are clutching at AI straws to try and repeat on your behalf.

[Anonymous Coward]

Re: Reality is an illusion ...

when your head is worried about your assets!.

Is that a left asset and a right asset?

Enquiring minds want to know :)

[Anonymous Coward]

Re: Reality is an illusion ...

I'm sorry, Dave, you can't say that.

Copilot has generated multiple reports demonstrating that everyone wants more Copilot.

You better not be calling Copilot a liar, it's more senior to you in this organisation.

[vtcodger]

Re: Reality is an illusion ...

Envision an absolutely perfect AI agent. You ask it questions. It decomposes them logically, and answers them precisely. You tell it to do a programming job. It decomposes your request logically and writes a program to do exactly what you requested. We'll call the agent Perfect AI (PAI).

What would you expect PAI would do given the sort of questions and programming requests the folks pushing AI into everything would ask/request? I think the result mostly would be something like 30% success of a sort and 70% chaos. The executive suite at Microsoft, its competitors, and way too many customers think they are dealing with Artificial Clairvoyance(AC), not LLMs. I have no idea if Artificial Clairvoyance -- the ability to somehow go beyond straightforward logic and decide what the user REALLY has in mind -- is possible. Humans can sometimes do that. On good days. But I'm pretty sure today's AI agents can't do that. And that they won't be able to anytime soon.

[Anonymous Coward]

Re: Reality is an illusion ...

Yes, the misplaced hype and use is actually Artificial Insanity..

[Groo The Wanderer - A Canuck]

Re: Reality is an illusion ...

What money? It's only a dream of money, a stock market bubble soon to burst, taking down the American empire with it as it combines with Drumpf's largesse to the uber-rich forcing the USA into the inevitable bankruptcy that Der Pumpkin Fuhrer is so good at declaring, courtesy of his boss, Putin the Butcher.

[Anonymous Coward]

Re: Reality is an illusion ...

Oh, don't forget his partners in crime, the gang of criminals formerly known as Republicans. They have presided over quite a few market crashes over the years, and I suspect the one we're heading for is the worst one yet. But at least it will take the attention off the Epstein files, right?

[Bebu sa Ware]

Micro- and Soft- Brain ?

When they let that Ifrit out of the bottle some things were clearly left behind—a non·existent grip on reality for starters.

"Jeez there so many cynics!" ""mindblowing" that people are unimpressed with AI."—Funny that. Never considered asking why ?

"Copilot finishing your code before you finish your coffee."—Not now. Not ever. I would give up coffee† before then.

_{† perhaps not. :)}

[Hubert Cumberdale]

Re: Micro- and Soft- Brain ?

What's mindblowing is their lack of contact with reality.

[steviesteveo]

Re: Micro- and Soft- Brain ?

I remember it used to be one of those "everyone knows" facts- according to some survey, execs did better at accurately describing things like demographics than politicians, general public and so

The idea at the time was business leaders had their wrong ideas beaten out quickly and needed to be informed so they were naturally on the pulse

It always sounded like something execs really wanted to be true but this AI boom has revealed some incredible disconnects

[Guido Esperanto]

Re: Micro- and Soft- Brain ?

This has echoes of Steve Jobs

"You're holding it wrong"

The dissonance is palpable

[Blogitus Maximus]

Re: Micro- and Soft- Brain ?

I believe their incredulity is performative. They know precisely what they're doing and its all tied to cash, which unsurprisingly pushes M$ to continue expanding it's evil production.

Someone somewhere wants a return on all this AI "investment" come hell or high water. Expect it to be shoehorned into every little thing you plug in.

The real answers to all this is to either a) not use it at all if you can or b) use your own locally hosted stuff. For the uninformed there will be plenty of monthly subscriptions to pay for when it all gets enshiffified.

[Blank Reg]

Re: Micro- and Soft- Brain ?

The amount that has been spent on AI can't be justified by even the most optimistic revenue projections. It's a mega bubble, and the pop will be deafening.

That's not to say that AI can't be useful, but putting it in everything just because you can is a dumb thing to do.

[Smeagolberg]

Re: Micro- and Soft- Brain ?

I don't doubt that "AI" is mindblowing to those who are a Bear of Very Little Brain.

[Sam not the Viking]

Re: Micro- and Soft- Brain ?

Like the 'Heffalump Trap'.

It's obviously effective because there are no Heffalumps around here.

[Anonymous Coward]

Re: Micro- and Soft- Brain ?

Copilot is everything you'd expect a Microsoft product to be, except it's worse.

[Anonymous Coward]

Re: Micro- and Soft- Brain ?

THAT is an impossible statement!

[Anonymous Coward]

Re: Micro- and Soft- Brain ?

The only impossible thing at MS is to release a product without security issues...

[HorseflySteve]

Re: Micro- and Soft- Brain ?

20 years ago, an erstwhile colleague opined, "The day Microsoft make a product that doesn't suck will be the day they start making vacuum cleaners".

With AI, I'd guess those vacuum cleaners would suck up the furniture & leave the dirt! And will require a 60 amp supply.

[Not Yb]

Re: Micro- and Soft- Brain ?

That particular joke is WAY more than 20 years old. I believe it actually goes back to the days of MS-DOS.

[I am David Jones]

Re: Micro- and Soft- Brain ?

MS-DOS happened twenty years ago and always will have done. I refuse to believe anything that suggests I’m getting old(er).

[PRR]

Re: Micro- and Soft- Brain ?

> goes back to the days of MS-DOS

Traf-O-Data

[JWLong]

Re: Micro- and Soft- Brain ?

"" "—Not now. Not ever. I would give up coffee† before then.

† perhaps not. :)""

I would never give up coffee, I would never start AI..........

And that you can take to the bank, if you can find one open with street access!

[PeeKay]

30 percent of Microsoft's code was now written by AI

Explains 100% of Microsoft releases lately.

[Anonymous Coward]

Re: 30 percent of Microsoft's code was now written by AI

MS 'Reality Distortion Field' once again ... in the current climate with such doubt about 'AI' would you admit that MS generates 30% of their code by 'AI' when the last 'few' releases of fixes have been a disaster !!!

The 'AI' hype has now reached a self-perpetuating state in MS-land where it no longer needs to prove it works, that is now a given ... because it needs to be to enable the marketing drones to be unleashed !!!

The 'AI' bubble is very very close to bursting ... it is now a matter of will Google and MS, amongst others, 'frighten' the rest of the commercial world into fearing the consequences if they 'see this 'AI' emperor has no clothes' !!!

Careers and huge profits are at risk, watch the 'AI' beast do everything in its power to keep conning the masses that all is well and your life NEEDS 'AI' in it !!!

:)

[ecofeco]

Re: 30 percent of Microsoft's code was now written by AI

Nailed it.

[kmorwath]

Re: 30 percent of Microsoft's code was now written by AI

Exactly - Davaluri and Suleyman stock options may face a huge hit in value - and if Nadella is fired for having burning hundresd of billions, they could follow.

And that's shows again companies are no longer interested in delivering products people want or need - they only look for ways to recover their investment faster. And they still need to find ways to make people actually pay for AI - shoehorn it into everyproduct, and as soon as people get dependant, have them to pay for it?

[Tron]

Re: 30 percent of Microsoft's code was now written by AI

The last act in this farce will be a switch to government money. Ultimately, AI software can access everything on your system and send it back to those datacentres to be searched through. That makes it worthwhile for governments to pay up and bail them out, as they can use it as a mechanism of universal surveillance.

[Anonymous Coward]

Re: 30 percent of Microsoft's code was now written by AI

I'm not sure why you seem to think that your machine isn't already doing this.

[steelpillow]

Re: 30 percent of Microsoft's code was now written by AI

Looking forward to the AI-generated Reality Distortion Field, coupled to an AI-conducted legal proofread before its AI-generated publicity campaign. What can possibly go wrong?

[stiine]

Re: 30 percent of Microsoft's code was now written by AI

We can hope!

[The Organ Grinder's Monkey]

Re: 30 percent of Microsoft's code was now written by AI

In another topic today I posted a reminder that Jaguar (cars) quality greatly improved when Ford (aka FoMoCo, or MofoCo to their detractors) took over.

Have MS' products deteriorated to the point where letting AI write them might actually be an improvement...

[steelpillow]

Re: 30 percent of Microsoft's code was now written by AI

Good point. After all, this AI only has to pass the Muppet test, not the Turing test.

[Czrly]

Re: 30 percent of Microsoft's code was now written by AI

Balls to that: it doesn't explain the vanishingly rare few that *do* work.

[Oblivium]

Re: 30 percent of Microsoft's code was now written by AI

Definitely explains Teams.

[FirstTangoInParis]

Re: 30 percent of Microsoft's code was now written by AI

Also explains Outlook app. The deckchairs are constantly being rearranged on it. Hopefully it will soon sink and we can get back to non ms mail APD that actually work.

[Pulled Tea]

Enough.

Setting aside Elon Musk's rather sad application, the advances in artificial intelligence have been jaw-dropping.

You know, I'm kind of done humoring this sentiment. I'm with this guy whenever someone in the media makes this sort of assertion: Which “artificial intelligence” are people talking about when they say that AI advances are… whatever? Are you talking about protein-folding? Computer vision? Reinforcement learning? Chatbots? What?

We've already got the CEO of Hugging Face trying to distance “artificial intelligence” from the bubble that he claims is caused by LLMs, but he was perfectly happy to keep mum and reap the benefits of the mania related to AI when everyone was being so breathless about how AI was set to revolutionize/destroy everything. Well, you can't back away from the fact that MechaHitler is seen as the same category of software as “real AI”. Everyone — and I include El Reg in this! — was perfectly happy talking about “artificial intelligence” having “monumental advances” while at the same time being quiet when Microsoft, Open AI, Google, Amazon and everyone shoving slop machines to consumers' faces, use “AI” as an excuse to fire and immiserate folk, and pollute the world and push us into climate disaster.

I've seen really good definitions of “artificial intelligence” — one by Ali Al-Khatib (“an ideological project to shift authority and autonomy away from individuals, towards centralized structures of power”) and another by Rua M Williams' book, Disabling Intelligences (quoted here as “a blanket term applied to any system that claims to supplement, reproduce, or replace human actions, decision making, or reasoning”) that focus on AI's effects to people and their material circumstances, and I'm going with those… and as far as I'm concerned, MechaHitler is “real” AI, and everyone who shilled for “real” AI systems will have to sit with the fact that they enabled bastards like Elon Musk, Sam Altman, Mark Zuckerberg, Dario Amodei, and all the monsters who fouled up our ecosystem and our ways of consensus and finding meaning so that Line Go Up.

You lot made your bed. Go lie in it.

[Dan 55]

Re: Enough.

Not one sad application but two.

However much money he has, Musk can't stop being the world's biggest loser.

[Anonymous Coward]

Re: Enough.

He is only the second biggest loser.

Why? because he is a loser!

[NewModelArmy]

Re: Enough.

Microsoft forcing AI upon us, is akin to a "friend" conning you into watching a scat video, which you cannot un-see, and then telling you, you must like it, and want it.

[colinhoad]

Re: Enough.

Amen.

[spacecadet66]

Re: Enough.

You can't see me, but know that I'm doing a one-man standing ovation here.

[Headley_Grange]

"Copilot finishing your code before you finish your coffee."

A bit like saying "My four-year old finished deocorating the bedroom while I made coffee" after you left them unsupervised for ten minuutes and they spread the paint all over the walls, floor, furniture and themselves with their hands.

[Prst. V.Jeltz]

or "We let a big orange moron with the IQ and social graces of a middle school bully and a narcism mental handicap run the country for 4 years "

[Hubert Cumberdale]

Quiet, piggy!

[kmorwath]

From the Pig in Chief...

[Dave K]

That's the issue in a nutshell. AI is like having a young, enthusiastic intern working for you. Sure there's a few simpler bits and pieces they can probably help with, but they are prone to making mistakes and lack the experience of the more grizzled veterans on your team.

So when you have something vitally important to work upon, you'd be crazy to hand it off to the intern because there's a high likelihood of mistakes in the work they'd return, and you'd have to spend a lot more time testing, validating and fixing their work. So in these cases, you'd usually either hand it to one of your more experienced and dedicated co-workers, or you'd tackle it yourself - because at the end of the day, you're accountable for the end-product.

There is a place for interns to provide some cheap value, but there's also a lot of cases where you simply need skill and experience for a piece of work. Plus of course with an intern, you're trying to train them up to become the grizzled veterans of tomorrow - something that doesn't really exist for an AI tool.

That's how it is for me. I'm accountable for the work I produce. Yes, I delegate some bits of it to co-workers in the team and utilise some "cheaper" heads for certain more utilitarian bits, but anything that's mission critical I either tackle myself or only entrust to someone who's competence and experience is without question. Why on earth would I risk my job and my credibility by allowing AI to make a botched attempt at my work just so I can grab a cup of coffee?

[Jamie Jones]

Most interns would not even be as bad as AI for that - if there is something they can't do, the decent ones will say so, not bluff their way with dodgy code they try to convince you is 100% correct. As for AI...

[Anonymous Coward]

I thought I'd ask the enterprise Copilot about some DB2 syntax

That's an hour of my life I'm not going to get back. It mixed up z/OS and LUW and then proceeded just to make shit up and when I finished by telling it to fuck off and stop wasting my time, it replied that help is available for self harm.

So yes, I am an AI cynic. If the tool were sold just as an internal document search which cited sources to be checked by the user which it would have the happy side effect of forcing companies to document their software and processes better. As a chat line to some all-knowing oracle it's absolutely not worth the billions being thrown at it or the jobs lost and the bubble can't pop soon enough. I really do hope MS have bitten off more they can chew and get holed below the waterline when it pops.

[Anonymous Coward]

Re: I thought I'd ask the enterprise Copilot about some DB2 syntax

Same here, for things that I actually need some help with, it does absolutely shit with, making crap up and going in circles, when saying that's incorrect, and it proposes another answer, which is also incorrect and when said, it says the first thing that it proposed, which i already said was incorrect.

[Doctor Syntax]

You're using it wrong.

You shouldn't be asking it questions where you know what's correct.

[Blank Reg]

Re: You're using it wrong.

I do it all the time. I use it as a typing assistant. I can describe a function in one sentence and it will write out 20 lines of code, usually correctly. If you get overly ambitious and do something like ask it to write a whole class then you need to be very precise in your wording or it can quickly go off the rails. Then you need to explain what it did wrong and try to break it down to smaller sections so that it doesn't try to fill in too many blanks

[RolandM]

Re: You're using it wrong.

And I am using AI quite often to lookup things I know in principle but don't remember in detail level.

[Anonymous Coward]

Re: You're using it wrong.

That's basically the issue. Using it as a mildly smarter search engine is OK as long as you know the subject you're asking questions about because its tendency to 'hallucinate' (because that sounds so much better than 'lie' or 'make shit up') when it hasn't got the answers or misinterprets the data renders it otherwise worse than useless.

Useless would mean you'd simply discard the response, but because it's good at making it appear the answer is sane means you spend time working with it until you discover it's been lying - again.

[Ken Hagan]

Re: I thought I'd ask the enterprise Copilot about some DB2 syntax

" If the tool were sold just as an internal document search which cited sources to be checked by the user which it would have the happy side effect of forcing companies to document their software and processes better."

Yes, but of course it would also reveal just how much copyright content it had hoovered during training, so none of the current big players will ever do this. We'll have to wait for a new entrant who is willing to do the job honestly.

[FuzzyTheBear]

Re: I thought I'd ask the enterprise Copilot about some DB2 syntax

it IS popping and investors are aware if this .. THey build data centers that they can't power up for lack of power .. lines are not coming in the buildings. Stupidity has a limit. IT's reached that limit and racks full of silicon that have no use bwcause they can't turn them on are a loss. In a year it's going to be old tech and all the money is going to be totally wasted. The bubble is blowing up and it can't come fast enough. I got a question : What has AI done for you lately ? .. I never used any part of it and never will .. i hate the idea behind it. HAL related brain freeze something .. When it started to get used for war by armies , that was the end of it .. there , imho , was nothing intelligent about it. Just more tech to murder maim and kill.

[Anonymous Coward]

Re: I thought I'd ask the enterprise Copilot about some DB2 syntax

I have a feeling gaming is possibly going to get a massive boost in 2026, 2027 because of all the GPUs that will be coming on the market when those AI datacenters are switched off.

That said, I'm not sure. One of the side effects is that the theft of intellectual property has again been allowed (because copying a cartoon out of a newspaper is theft, but if you copy literally everything it's "for humanity" or something), and locally installed AIs are more intelligent means to sniff around the insides of companies, with Microsoft's DLP helping as customers have already flagged which bits they consider important. From Microsoft's perspective it's pretty near perfect to grab larges volumes of govermment money.

Or, translated, you're paying taxes to enabled being spied on some more, and your power bills are going up because somehow that stuff needs to be powered and cooled too.

I'm not surprised Microsoft jumped on this.

[Anonymous Coward]

Job is only 50% done

Artificial: Yes

Intelligence: No

[cd]

My condiments

I'm ready to relish big tech going down their own black hole because their products can't catch up and cut the mustard.

Hot dog!

[Tom7]

He's right, as far as it goes

The fact you can have a fluent, intelligent-sounding conversation with a computer is mind-blowing and we shouldn't forget how impressive that is.

We also shouldn't be fooled by its fluency and how intelligent it sounds into thinking that it is able to think intelligently or that it knows anything for certain. LLMs are trained to produce fluent, plausible-sounding output and that is it. They will always prioritise fluency and sounding intelligent over being correct.

The fact a thought leader at Microsoft doesn't understand this and its implications for how LLMs are used in products is disturbing.

[breakfast]

Re: He's right, as far as it goes

Unfortunately being able to have a credible-sounding conversation with no meaning is a very powerful cognitive trap for the human mind. Words are tokens of meaning for us - I can't read "elephant" and not connect it to our understanding of an elephant, so we instinctively feel that we're having a meaningful conversation. The noise is so signal-shaped that at best we have to constantly remind ourselves that it is just noise. People who can't do that are going to fall into the trap and before you know it their minds, too, will be blown.

I don't know what we can do about this, it troubles me that we've taken a philosophical thought experiment and unleashed it on the world with no consideration for what that might mean.

[Anonymous Coward]

Re: He's right, as far as it goes

Consideration? How about Intent? or Concern?

Well, their intent is to get people to give them money.

As an aside, what happens if you ask chatgpt to 'imagine you are a thousand monkeys. now write all of Shakespear's plays and save them to disk for me.' Would it succeed? How do we go about getting it to detail the trade secrets of the companies that have stopped using it?

[breakfast]

Re: He's right, as far as it goes

Given that it has been trained on most of popular culture you'd probably get a response like "It was the best of times, it was the blurst of times."

[Smeagolberg]

Re: He's right, as far as it goes

"The fact you can have a fluent, intelligent-sounding conversation with a computer is mind-blowing and we shouldn't forget how impressive that is."

It's only impressive until you understand that it's predictive text based on statistics and a lot of source data. Once you know that it is neither mind-blowing nor Magic (*).

(*) https://en.wikipedia.org/wiki/Clarke%27s_three_laws

"Any sufficiently advanced technology is indistinguishable from magic"

LLMs are "sufficiently advanced technology" to those who don't understand how they work, which is a large proportion of those who believe they are qualified to spew words about them.

[LBJsPNS]

Re: He's right, as far as it goes

"The fact you can have a fluent, intelligent-sounding conversation with a computer is mind-blowing"

It was mind-blowing when Eliza did it, what, 40 years ago now?

[captain veg]

Re: He's right, as far as it goes

> when Eliza did it, what, 40 years ago now?

60, in fact. There's very little genuinely new in computer science since the 1960s.

-A.

[Andrew Dyson]

Re: He's right, as far as it goes

> "They will always prioritise fluency and sounding intelligent over being correct."

...as will a depressing number of human beings. So in that sense the AI is actually pretty good at mimicking our behaviour. Maybe we should give it more credit??! lol

[nijam]

Re: He's right, as far as it goes

> The fact a thought leader at Microsoft doesn't understand ...

... anything at all shouldn't be surprising.

Is he actually an AI? Or just an A?

[PRR]

Re: He's right, as far as it goes

> The fact you can have a fluent, intelligent-sounding conversation with a computer is mind-blowing and we shouldn't forget how impressive that is.

ELIZA gave people good insight... back in the early 1960s.

"...some people.., attributed human-like feelings to the computer program, a phenomenon that came to be called the ELIZA effect. ...many early users were convinced of ELIZA's intelligence and understanding..."

https://en.wikipedia.org/wiki/ELIZA

[Doctor Syntax]

"I grew up playing Snake on a Nokia phone!"

Not an impressive thing to highlight in your CV to run a tech business. It explains a lot.

[Fruit and Nutcase]

CV

With his AI credentials, "he would say that, wouldn't he?", to misquote Mandy Rice-Davies.

https://en.wikipedia.org/wiki/Mustafa_Suleyman

[Doctor Syntax]

Re: CV

Clearly a highly talented climber of corporate trees as one might expect.

[IGotOut]

"I grew up playing Snake on a Nokia phone!"

That'll be the same Nokia brand Microsoft bought and killed wouldn't it?

[cyberdemon]

Re: "I grew up playing Snake on a Nokia phone!"

Or rather infiltrated, sabotaged, bought at fire-sale price after setting the fire, and then killed ...

I can never forgive them for killing Maemo. Nokia's debian-based OS for the N900, before ex-MS exec Elop came in as the new CEO and immediately killed it, because his real employer were trying to push the abomination that was Windows Phone.

[Fruit and Nutcase]

Re: "I grew up playing Snake on a Nokia phone!"

I had a Nokia N9, and that was the end of the road once Elop came on board.

I still use the message alert sound from that on my current handset.

Not that it's of any consequence now, but Skype was also seamlessly integrated to the phone and messaging functions.

[kmorwath]

Re: "I grew up playing Snake on a Nokia phone!"

If they only invested a little more money into Nokia phones and Windows Phone just like they are burning them into AI.... instead of abandoning it and adopting Pichai's products instead....

[Flicker]

Just say no...

My coding days are long gone, but the combination of Windows 11 "upgrade" requirements, unending nagging to get a 365 sub and the really annoying promotion of Copilot, which I have absolutely no use for or interest in, finally tipped me over the edge. After a lifetime's professional and personal use of Microsoft products from MS-DOS onwards I've now moved 99% to Linux Mint, retaining an old Win 10 system for a few old applications which don't play well with WINE (although would admit to posting this from that system, which I was using when I saw the article).

In my limited (mainly family) circle I know of no one who would consider buying a new Windows machine - they either operate exclusively through mobile phones or else buy MacBooks. Seems to me that Microsoft has steadily, determindly enshittified their products and alienated their consumer base over the years with no obvious route to recover it.

[nematoad]

Re: Just say no...

I don't think that MS's focus is on the consumer sector anymore.

It seems to me that having a stranglehold on business customers is all they care about these days. Companies are so tied to the MS ecosystem with applications, business processes and legacy holdovers that they *can't* easily or cheaply get off the MS train. They are locked in and MS seeing that will continue their march towards an AI "paradise" whether their customers like it or not.

It's the old sunk cost fallacy. So much money invested, they have to get something out of it even if it means pumping even more money in. And someone else will have to pick up the bill.

[DoctorNine]

Re: Just say no...

The worst part is, there are simply huge amounts of money to be saved by switching off the M$ crack habit.

Junkies are gonna Jones for the hard stuff no matter what we tell them though.

And don't get me started about .pdf and Adobe either.

[M.V. Lipvig]

Re: Just say no...

I think they've started torpedoing their business customers as well. Those who deal in a lot of protected IP are howling, and governments are starting to move away from them over security issues.

The coffin will be firmly nailed with a power hammer once larger governments move away, then start telling companies they contract through that they are not allowed to use M$ in anything related to the government contracts.

[stiine]

Re: Just say no...

Unfortunately, I hope you're correct.

[Mage]

The Tech is impressive?

No, it's not, unless you are very naive.

It only seems more impressive than Eliza because it's been fed the Internet's content and also scanned books (neither of which is fair use). Eliza was static with the responses in the code. The LLMs are using a distributed database (the so-called Neural Networks are not simulations or copies of biological brain structures), Almost all the marketing and PR is a lie.

Some people thought Eliza was impressive. It's in Linux emacs; type emacs in a terminal and then Esc x doctor (no spaces)

I am the psychotherapist. Please, describe your problems. Each time

you are finished talking, type RET twice.

How do I turn off AI?

What do you think?

[Raphael]

Re: The Tech is impressive?

I keep pointing out that most LLM AI is simply what we used to call an Expert System with a natural language processor

[Mage]

Re: Expert System

The Expert systems were expensive to produce because they used expensive human experts for the human curated data. They are not flexible and limited to one narrow field. They were not wasteful of computer resources, water and electricity.

An LLM / Generative AI is using randomly scraped info with no human curation. They seem far more natural, but are not the same at all and are totally untrustworthy.

[Irongut]

The tech is NOT impressive

Especially CoPilot which is particularly shit in all its guises.

And the more you stuff it down our throats the more profits will slip through your fingers.

[colinhoad]

They just don't get it

The only mindblowing thing is how vastly disconnected Microsoft leaders - and the rest of the Silicon Valley bigshots - are from their users. Microsoft used to be a software company that cared about making products users wanted, now it's just a funnel for whatever daft trend has taken hold on Wall Street.

[Skrynesaver_2]

Embedded in everything

The app (whatever happened to timer switches) that comes with my radiators now has multiple AI modues available, I just want to be able to set them to heat the room during the low tariff, instead I'm being asked to participate in heating the f%^*ing icecaps.

[TVU]

"Microsoft exec finds AI cynicism 'mindblowing'. The tech is impressive. Shoehorning it into absolutely everything is not"

Therein lies the issue. The problem is that AI is being overhyped and it is the latest buzz word that gets added to everything (here's looking at you ServiceNow) even if it isn't really applicable/useful.

[Cruachan]

Copilot is just the new clippy to most people, if you want/need to use it that's fine but wait to be asked instead of shoe-horning it in to everything. Opened a PDF in Adobe read the other day, it defaulted to the new experience and started telling me that it was a long document, maybe I'd like an AI summary of it.

NO, I FUCKING WOULDN'T

Meanwhile Musk has gone full-blown Josie and the Pussycats by making Grok post about how great he is. Definition of garbage in, garbage out.

(It was a terrible film that was meant to be a satire on advertising and product placement, at the end the main villain’s master-plan is revealed as messages to tell everyone how great she is because she was an unpopular kid at school)

[Red Or Zed]

There is a quote:

“It is difficult to get a man to understand something, when his salary depends on his not understanding it.”

― Upton Sinclair

Which seems to fit here. M$ were behind the curve on the internet, on virtual machines, on security as default, that they're determined to be fully in the Next Big Thing.

AI looks like Tulip Mainia, but with more destruction of eco-systems.

[AdamWill]

Yes

Yes, Mustafa, it is the children who are wrong.

[Anonymous Coward]

Sillyman.

[Ian Johnston]

Pattern identification is not AI. Pattern identification is useful, for a start.

[kmorwath]

Esactly. I found useful that Adobe and Photoshop can build mask faster - even if I have often to refine them - once I tell them what objtecs should they mask. They usually work.

Generative AI to remove distraction not rarely botches the task, especually when the backgroud is not simple enough, or common.

[david1024]

Blame the customer

I mean, only luddites wouldn't want $300-$500 per year AI help that they don't need telling them 80% wrong answers. Not all businesses or tasks benefit from an extra step in the middle.

Wait, is that CEO or markering accuracy range?

Trash for web use. Need an off switch.

[O'Reg Inalsin]

Re: Blame the customer

$400 from every adult and child in the US (~342 million) still only comes to about $137 billion a year, where the data center build out requires north of 1 trillion a year of AI spend to be economically feasible.

[John Brown (no body)]

Re: Blame the customer

"Trash for web use. Need an off switch."

I asked copilot how to remove copilot from Teams (It's a work laptop, I don't get a choice!). It gave instructions on how to unpin the copilot app from the sidebar. So I told it it was wrong and that would not remove or disable copilot. It replied, "Sorry, you are right..." and proceeded to give a fairly complex and involved solution that boiled down to me needing admin privileges to change a system policy *which I can't do, work laptop!). So, it's first response was not even close to correct and the second was basically, "we don't want you to disable or remove it so we made it really awkward do do that"

[M.V. Lipvig]

Re: Blame the customer

Does your company have a mechanism to report security breaches? Sounds like something reportable to me.

[Strahd Ivarius]

Re: Blame the customer

Windows or anything coming from MS is a security breach...

[John Brown (no body)]

Re: Blame the customer

We have blocks on a number of AI access points, but since copilot is MS, the chiefs seem to have decided it's "safe". I suppose the logic behind the decision is that since MS already have all our data on their servers anyway, letting staff have access to copilot isn't making stuff any worse (although they probably would not word it that way)

[Paul Hovnanian]

Oh well ...

... it will all be fixed in Windows 12 (Code named: Sorry about the AI).

[M.V. Lipvig]

Re: Oh well ...

Yes, but 12 will require a TQM module to work, and will reject attempts to run on a machine eith a TPM module.

[kmorwath]

Re: Oh well ...

And the installer will be fully written with CoPilot....

[ITMA]

He doesn't get it....

If Mustafa Suleyman really finds AI cynicism "mind-blowing", the bloke really needs to get out into the real world more.

Talk to real users.

I think he may find their frank views on having AI rammed into everything rather more than mind-blowing. More like utterly dejecting and shocking when he hears many just don't give a flying f*ck about AI and want it gone and dead.

[Doctor Syntax]

Re: He doesn't get it....

He'll just know they're wrong.

[ITMA]

Re: He doesn't get it....

Probably - because CoPilot told him so.... LOL

[kmorwath]

Talk to real users.

Those low-paid serfs? They don't talik to them. They despise them. They don't even come close to such despicable people. They must pay, make them rich, and shutup.

[frankyunderwood123]

Don't even get me started on ...

AI search result overviews... so often inaccurate or just absolute rubbish, only surpassed by the actual search results themselves.

The enshitification of search results is now complete, seemingly regardless of what search engine you use.

DuckDuck Go was once regarded as a shining beacon of where search engines should be headed, now it churns out the same crap as Google search.

Part of the problem is the HUGE amount of auto-generated content, whether that be AI slop or older tech isn't the point, the volume is the point.

For any given search, now regardless of what engine you use, easily 60% of the results are "websites" that regurgitate shit content - and that content is more often than not auto generated from ... shit content.

Ask any question in the search engine and be faced with search results that when you visit the site itself, you are thrown an "accept cookies" popup, but there's no "reject all" - it's that horrible popup with so many options to uncheck, it scrolls on to infinity.

You get to recognise the cruft after a while. I'm sure it's a small amount of "players" generating this shit, but they sure are prolific.

I even started adding these URLs to my hosts file at one point, just to banish them.

I gave up. They are infinite.

If there's one way to force everyone to use AI, it's to scupper search engines - looks like that's been done already.

Even the standard prepending "reddit" to the search query, in the hope of getting actual human opinion, unfiltered, is starting to get less useful.

Are we headed toward search engines that are entirely AI, but masquerade as search results?

Would that make any difference at all to the already shit search results we now have?

This is just the beginning of enshitification of finding information on the internet - it is going to get exponentially worse, as AI slop is indexed and AI agents slurp up the indexed AI slop to generate second, third, fourth and fifth level sloppy seconds, thirds, fourths and fifths.

Beer is what I need, more of it.

"Barman, take my phone and laptop and show me to my pony, I'm trekking back to the past, I'm no longer a gun slinger in this shitty new world."

[M.V. Lipvig]

Re: Don't even get me started on ...

noai.duckduckgo.com works. Duck without a side of enshittyfication.

[DoctorPaul]

Re: Don't even get me started on ...

Qwant works pretty well for me.

[Anonymous Coward]

Finished

Microsoft is finished. If it continues on this road it will alienate users to the point where they'll start looking for alternatives. Linux Mint comes to mind and many have already made the jump. My guess is we'll see uptake numbers that will shake Microsoft to its foundations before the year ends.

Also, Big Tech and Wall Street seem to be holding each other in a deadly embrace. Not going along with the A.I. bubble ride will mean your stock will suffer and you as an executive will be kicked to the curb. It also means much of the wealth in the U.S. is phony and based on Monopoly money.

The sad thing is that richer people will get richer from this bubble whilst the working Joe will have to pick up the bill when the entire house of cards implodes. Banks will most likely need saving.

[Scotthva5]

Re: Finished

>> Banks will most likely need saving

Almost guaranteed and the same people that lost their livelihoods will be forced to pick up the tab whilst the asshats that started this shit will walk away with millions.

[kmorwath]

Re: Finished

Nadella hid Microsoft cuts on products development, and product quality, behind the cloud money first (where he needed people like Russinovich to make it work...), and now behind AI. Shareholder saw money and didn't make questions about the future.

When the AI bubble bursts, Nadella will have a lot to explain. And once he finished to kill Windows and Office, even more. I'm awating to see what happens when Outlook is replace by the "new" Outloook, and what happens to Exchange deployments - especially after the security SNAFUs.

Or when issues about US control of cloud systems make Azure less appealing outside US.

[ecofeco]

I miss the days

I miss the days when losing a billion or two dollars meant your company went out of business. And collusion, predatory pricing and price fixing were against the law.

Now? The corps are so big they can lose billions in months and still stay in business and eventually force whatever snake oil they are selling on everyone.

We are past manufactured consent and deep into forced consumption.

[Anonymous Coward]

Re: I miss the days

Plus they are now "too big to fail" and will undoubtedly come cap-in-hand at some point.

[Doctor Syntax]

Re: I miss the days

Their problem might be that they're too big to bail out. Yes, failure is an option.

[John Brown (no body)]

Re: I miss the days

Are we in the early stages of the world of The Space Merchants?

[Philippe]

Has he used his own tools?

I am often assigning the same task to two of these tools, It can be Copilot and ChatGPT, Copilot and Gemini Pro, ChatGPT and Grok and any other combination.

Copilot always comes up with the worst results, it also makes obvious mistakes and miss pretty basic stuff especially in PDFs.

So when this dude says "Why aren't customers impressed" he's clearly talking about Copilot customers.

[DoctorNine]

Re: Has he used his own tools?

No offense, Phillippe. But I think the problem is that he HAS been using his tool a bit too much... ahem...

[Ashentaine]

Smile, citizen! Happiness is mandatory.

>Has he used his own tools?

Given that he works for Microsoft, the answer is definitely yes as it's mandatory to do so. And I reckon a large part of his pay packet now depends on not only using Copilot regardless of whether it's necessary or not, but also shilling it to anyone who is within earshot at every opportunity.

[Anonymous Coward]

Suleyman also said cyber crime is lessening

I think earlier this year or late last year Suleyman inexplicably said cyber crime was decreasing and isn’t of much concern.

He was rightly excoriated for that.

His lack of awareness with that and now AI fits perfectly with the Microsoft ethos: features first, security last, coherent UI huh?

I assume people like that set up layers of sycophants to provide warm hugs all day long.

They’re like multilayered sycophant sponge cakes that collapse into fondant on the great B bakeoff and leave the judges grimacing.

[sarusa]

A complete f@$#ing dumbass whose only job is literally cheerleading complete fail

This guy is a f@#$ing dumbass (he has said many other dumb things) whose only job is literally cheerleading these atrocities to increase stock value. At least if you were Google you could kind of point at Deepmind at being something you bought that is actually doing good things, but Microsoft has literally nothing worse than useless for AI.

But this is all you can expect from C levels. All dumber than a f@#$ing LLM.

[Always Right Mostly]

"30 percent of Microsoft's code was now written by AI."

It shows. This is not a compliment.

[Matthew 25]

Self confessed AI cynic

I like to do things myself because then I understand why I did it that way. That means I can document it and come back years later to update it.

If I say to AI "I want a THING" I may get a THING but I have no idea why it won't work on the last Thursday of the month or how to fix it.

[Anonymous Coward]

Re: Self confessed AI cynic

Also, you won't know this for several months because your customers are using AI as well.

[Anonymous Coward]

They don't get it...

...because inside these organisations, access to biggest, greatest models is a given...outside these orgs most folks are using the "free" versions which are significantly shittier...hardly anyone has enough GPU compute to run the really good local models.

If OpenAI, Microsoft et al really want to push this AI thing forward, they need to focus on bringing down the cost of compute...be that more efficient models or saying "FUCK NVIDIA" and designing alternate hardware to run local models on.

That's the point at which AI becomes genuinely interesting.

Having a woke, guard railed, frankensteins monster on a sack barrow that people can pay to talk to is not exciting...what they can see unfettered behind closed doors without paywalls is probably exciting.

These execs don't understand that they live in a completely different world to the people on the street when it comes to AI...because the people on the street haven't seen what the execs have seen...what they get to see is hugely different from the dumbed down trash that gets packaged for public consumption.

[nobody who matters]

Re: They don't get it...

Rubbish - even if they do have access to (and regularly use) a version of the chatbot that the great unwashed are not allowed to see, they will also know very well what <is> being offered to the outside world. If they don't (or are incapable of understanding that), they shouldn't be in the job to start with.

Your post may well explain the reason why Suleyman appears to be disconnected from the real world, but it is not a valid excuse for him!

[blu3b3rry]

Re: They don't get it...

Not sure that's quite right - a bullshit generator is a bullshit generator regardless of guardrails, and if it's running locally or on the cloud. All the increased "compute" does it let it generate more bullshit far faster.

If the internal versions of CoPilot are so good, how come the now 30% AI coded W11 continues to get more and more unstable?

[Anonymous Coward]

Re: They don't get it...

"woke AI" LOL. Keep drinking the Musk-juice!

[Piro]

I didn't realise my simple opinion was

Mindblowing. We never needed this junk, we'd be better off without it, and it's consuming very real resources at a crazy rate.

[martinusher]

Physician Heal Thyself!

It might have escaped the notice of this person that Windows has become a really nasty pile of festering dog excrement. Its always had tendencies but recent attempts at 'upgrades' seem to have broken more than they've claimed to fix.

I'd have a lot more confidence in their company if it produced a relatively bug free, stable platform that had adequate performance on average hardware. Say something as reliable as a typical Linux or FreeBSD distribution. They could 'Apple' it, adding their own snazzy UI and other proprietary bells and whistles but the underlying system needs to 'just work'. Its not too much to ask, they've done it before but they just don't seem to be capable of building boring, solid, stuff -- they've got to keep adding bells and whistles to it until it becomes an unstable mess.

[Anonymous Coward]

Mustafa Suleyman?

Who left a muslim terrorist in charge at MS?

Deport all these people with foreign-sounding names!!!

[Anonymous Coward]

Re: Mustafa Suleyman?

I'm pretty sure your name, Mr Anonymous Coward, is an equally foreign name EVERYWHERE ELSE on the planet.

Also, the answer to your question is: State Street, Vanguard, and Blackrock.

[nobody who matters]

Re: Mustafa Suleyman?

The vast majority of the American population have foreign names ;)

[Anonymous Coward]

Re: Mustafa Suleyman?

How many centuries does my family have to live in the US before our name is no longer foreign? We've been here since the late 1600s.

[nobody who matters]

Re: Mustafa Suleyman?

It will always be foreign to the original inhabitants of North America.

From their position, the early settlers from Europe were the illegal immigrants of their day.

Some Americans would do well to remember that before they victimise more recent arrivals.

[Anonymous Coward]

AI writes, reads and believes it’s own press releases

It’s just regurgitation of regurgitation.

[nobody who matters]

Re: AI writes, reads and believes it’s own press releases

That's just sick

[Dwarf]

Allow me to simplify

People have been unimpressed with Microsoft products for a long time, what makes you think that something called AI is going go change that ?

Remember that we never asked for it and yet we get it rammed in our face all the time, with no way to turn it off. Now imagine it was a toaster constantly offering to make you toast.

[Kevin McMurtrie]

AI can does amazing things, but it's a long way from general intelligence. If anything, Microsoft's LinkedIn should be the textbook example of a corporation not understanding how to use AI. Even when doing simpler tasks like code assistance, AI mistakes negate many advantages. Especially the sneaky, subtle mistakes that slip by if you trust AI too much.

[Andrew Richards]

Cynicism can be explain by just opening Copilot...

Just done this: at the bottom of the window - admittedly in a smaller font size than everything else is "Copilot may make mistakes". Yes, a lot of what is available is impressive. I use Copilot daily largely as a proxy for web-searching for information in a domain in which I'm (apparently) "expert": i.e. I can tell if it's producing something useful or nonsense.

But that's the problem: would Windows Calculator be useful if every Nth calculation was wrong? That what he doesn't understand: impressive is one thing, until it's infallible it's a new type of computing. It's not deterministic and accurate, it's sometimes potentially wrong. Microsoft - and to be fair, everyone else - has to sell this element with the product. People are used to believing what a computer tells them: it's massively irresponsible to push something that's new-shiny without being honest about this bit. If they don't people will be burnt by this and stop believing the hype.

But it gets worse: this in the same week where MS are pushing future Copilot-agentive stuff will requires read AND write access to your data. Ignore what they're doing with your data and where: is Suleyman going to put money on this and compensate the first home user using this and Copilot accidentally deleting a folder of holiday photographs?

[blu3b3rry]

Re: Cynicism can be explain by just opening Copilot...

That is the crux of it, really. If I instruct a computer to do "q" then it should do "q" every single time, not occasionally doing "m" instead without warning. If it does do "m" without warning I expect a reason why and/or the instruction to fail.

Who in their right mind would trust their money/livelihood/personal property/insert as appropriate with something that unreliable?

Admittedly one thing has been fixed. CoPilot finally counts the number of "r's" in "Raspberry" as 3 instead of 2. Only taken them over a year to get that right.

[Expect Great Things]

Ample Motivation

Apparently, Mr. Suleyman’s annual compensation as CEO of Microsoft AI is in the vicinity of $10m. The Upton Sinclair quote springs to mind:

“It is difficult to get a man to understand something when his salary depends on his not understanding it.”

[JimmyPage]

Cynic ? Or realist ?

"AI" as peddled by "AI" vendors is unremittingly shit.

[Anonymous Coward]

Re: Cynic ? Or realist ?

I gave you an upvote but you're only half right, maybe all right in that it doesn't meet their hyperbole.But ... the more ignorant and unthinking you are, the more dangerous it is. So I would say being shit is not the issue, being unable to recognise when it's shit is the problem.

[Sleep deprived]

AI is not intelligent nor does it have to be

AI is not intelligent, not does it have to be. It only needs to be less dumb than those who get fascinated by a statistical predictive engine. And this seems a much easier goal.

[AstroCam]

MS has a history of doing daft things with Tech. Remember Kinect, the next big gaming peripheral. You are the controller, etc. They shoehorned it into everything. 10 years later it is in nothing. I am not saying AI is as useless as Kinect, but both have their place as tools

[Anonymous Coward]

Kneel acolytes

How could you doubt the word or faith of the masters?

[BartyFartsLast]

I'm impressed by it.

Just not impressed enough to think it's even close to the hype.

There's a *lot* of money going to be lost when this bubble bursts

[mikeymikec]

"AI makes 30% of our code!"

Why am I not surprised?:

https://www.neowin.net/news/microsoft-finally-admits-almost-all-major-windows-11-core-features-are-broken/

[Houninym]

AI shoehorning

And The Register shoehorning a Google AI advert in at all possible places in the adverts in the article. Call me a cynic.

[nobody who matters]

Re: AI shoehorning

There are adverts in the article?

Who knew? ;)

[Martin an gof]

Guest Editor

Look who one of the guest editors on the Today programme is going to be! They had a little personal statement slot at about 0720 this morning if you want to scroll through the programme (to about 1h22m) on Sounds.

M.