One real reason AI isn't delivering: Meatbags in manglement

Every company today is doing AI. From boardrooms to marketing campaigns, companies proudly showcase new generative AI pilots and chatbot integrations. Enterprise investments in GenAI are growing to about $30-40 billion, yet research indicates 95 percent of organizations report zero measurable returns on these efforts. In fact …

[Doctor Syntax]

"They think they have a smart system, but what they really have is a stateless algorithm that never improves."

And what were they sold as by the sales-snakes? A smart system that could learn.

[herman]

It is good for writing bed-time stories and prototypes. Not so hot for full size novels and production code.

[VicMortimer]

What an idiotic article

The best way to include AI in your company is to not use it.

You're setting yourself up for failure if you do. When this bubble pops, AI is going to get a LOT more expensive, and you'll have skyrocketing costs that may just bankrupt you if you can't rip out AI and replace it with people. Oh, you fired the people? Well, bye.

[RoloH]

Yes, it truly is....

"The models are more powerful than ever."

Duh, the "models" and the entire generative paradigm ARE the problem.....

[Anonymous Coward]

Re: Yes, it truly is....

The real measure of an AI model is how large a room it would heat.

[elsergiovolador]

Marketing

The need for marketing department is an indicator that your product is shite.

[nematoad]

Re: Marketing

I disagree. It's no use making something and then not being able to sell it.

I know because I found myself in that position.

I started a tiny company which designed and made etched-brass items for model soldier enthusiasts. A fair bit of work on my part and a small outlay to get the parts made. Then I hit a brick wall, the people I was hoping would sell the parts backed out and I had nothing to replace them with. So the company folded and I was left with inventory that just sat there. Eventually I went to Ebay. It's not perfect but it at least helps me market what I have and get back some of the money invested, but not my time.

You may not like the tactics used by marketing departments but they are a distasteful necessity. Just like lawyers.

[DecyrptedGeek]

Re: Marketing

True but I used more shitty products because of a fancy sales brochure when all the tech savy staff pointed out it was shit this other company made a superior product. Seems to me its the higher ups and sales staff that fuck up projects.

[elsergiovolador]

Re: Marketing

You’re describing a distribution failure, not a product quality problem.

If a product needs a marketing department to convince people it is worth buying, that is usually compensating for weak pull. Strong products sell through word of mouth, reputation, and users doing the selling for you. Marketing can accelerate that, but it is not the same thing as needing it to exist.

Your example wasn’t “the market didn’t want the product”. It was “the planned sales channels fell through”. That is logistics and access, not persuasion. Ebay worked precisely because it reduced marketing to discovery, not manipulation.

Marketing departments exist mostly to manufacture demand where organic demand is insufficient. Sometimes that is necessary. It is still a smell

Good products spread. Mediocre ones need explaining.

[MachDiamond]

Re: Marketing

"The need for marketing department is an indicator that your product is shite."

If nobody knows you are making something, you don't get any sales. Marketing is how you get people to know you exist.

[M.V. Lipvig]

Re: Marketing

There's advertising, then there's marketing. Marketing may be advertising, but advertising is not marketing.

[MachDiamond]

Re: Marketing

The marketing people need to sort out what to put in the ads, where those ads will be seen by the target audience and hand that off to the advertising department to come up with some implementations. The two things work hand in hand, but handle different aspects of it in companies large enough for there to be a big distinction. For a small company, management working with whoever is coming up with the ads is "marketing".

[Mimsey Borogove]

Re: Marketing

If nobody knows you are making something, you don't get any sales. Marketing is how you get people to know you exist.

As a young and idealistic librarian in the 90s, I turned up my nose at the idea of marketing. I thought, it's a library! People know libraries exist, and that they provide useful services to everyone from CEOs to poor children. But around 1998 or so, when we hired a new head librarian with good ideas about marketing, I realized that I couldn't assume that. Even if people knew, vaguely, that the library existed, it might not occur to them that even if they didn't need a best-seller to read at the beach, it could still be helpful to them. Our new boss started a marketing program that told people what we did that would help them in their jobs (being a government library), and usage went up. I adopted her ideas, helped with the program, and finally knew that, no matter how good the product may be, people still need to be told about it.

This applies to everything out there, products, services, shops, software, and all.

[mickaroo]

Interesting Etymology

You can’t spell “fail” without “AI”.

[StewartWhite]

Since when are LLMs intelligent!?

"In simple terms, the AI was intelligent..."

No, it wasn't intelligent - go to the back of the class and repeat 1,000 times "LLMs are stochastic parrots" and no cheating using ChatGPT because it will eventually degrade and return "Limbs are pork-fascistic carrots".

[Decay]

The sad reality is most users, business departments and management want a no thought required solution to a problem and AI is sold as that solution. I want AI version x, y or z to do this work for me faster and cheaper without putting any effort into thinking about how it will do it, what we are doing that could be improved. Just set it up and we will push the button and it all just works. This mindset isn't limited to just AI. We have all been victims of sales drones convincing the powers that be that product X is plug and play, just install it and your accounting system, CRM etc. it will be fantastic, just sign this contract and expect invoices every month.

The difference with AI is that the hype, and the decision makers experience with AIs writing grade A management waffle emails and memos, has convinced many that if it can write my emails so well, then just imagine what it could do building reports, performing complex analysis etc. I have somewhat solved this problem by having same people perform some basic analysis, stats or reports using a fairly simple dataset that they understand. Once the AI has mangled it a few times, made obvious errors or just plain made stuff up they usually respond with shock or occasionally with it must be that particular AI, so we try a few other flavors and they soon realize how bad it can be. Doesn't stop the sales drones trotting out the "our product is trained specifically for this scenario" bull but at least it gets the users or management asking questions and doing some critical thinking which seems to be a lost art.

[martinusher]

Its the outsourcing mindset, only this time using a machine instead of anonymous foreign contractors.

[ecofeco]

This.

[Doctor Syntax]

"it must be that particular AI, so we try a few other flavors and they soon realize how bad it can be"

The risk here is that they decide you're the common factor so it must be your fault. TThen they decide to get rid of you.

[wub]

Try understanding the problem BEFORE solving it

Great article - very insightful. I'm a bit disappointed in myself for needing to have this explained to me. I do have a small bit of relevant experience in a similar area.

I was tasked with obtaining a LIMS (laboratory information management system) for the lab I worked at in the dim and distant past. As part of my education, I attended a dog and pony show for one of the leading providers. In my mind, this was still early days for LIMS products, but the first thing the speaker did was to ask the audience to show hands whether this was the first LIMS their companies had bought. It turned out that there were already folks looking for a fourth LIMS system.

I am far from a management expert, but even I realized immediately that if you had already failed three times at trying to make one of these products benefit your company, there might well be a systematic failure in that organization that needed to be addressed first.

When it was time for actual implementation of a product for us, I was struck by the fact that no one wanted to do the work of figuring our IN DETAIL how our internal processes actually worked, and in the cases where I did get some cooperation, no one seemed to understand that the only things the system could do for us were the ones we told it about - if we only did something once in a while, we could never do it at all, unless we also explained how and when to do that do the new system.

I'm surprised any of these things work, at all. You're absolutely right - many if not most of the failures come down to management failing to understand the problem first.

[Tim99]

Re: Try understanding the problem BEFORE solving it

I'm long retired, but I installed several small LIMS systems in the 1980s - Installed a large proprietary UNIX system, then wrote and sold my own (smaller custom systems). Here is some (unwanted?) free advice...

If you buy a commercial system, change your procedures to match it; try to avoid Oracle (I bear many scars). Alternatively, write and support your own, making sure that you employ and train sufficient staff to keep it running, and expand it if necessary - preferably using standard (FOSS?) tools, or at least (if you really have to) "industry standards".

Under no circumstances, buy a commercial system, and then change/add to it, even if the changes are done by the supplier, in a few years time you will have an unsupportable mess. The supplier will have upgraded their software, and left you stranded.

[David Hicklin]

Re: Try understanding the problem BEFORE solving it

> If you buy a commercial system, change your procedures to match it

Have an upvote for just that, I have been part of one ERP migration which was a success as the management were committed to changing processes, it probably helped that we had just been sold to new owners who replaced all the older, long term upper management with younger more enthusiastic ones.

[frankyunderwood123]

Stopping treating it like a replacement for human Labour would be a start...

I do hope the bubble bursts soon - just so long as it doesn't hammer my investments, right?

That's the thing, those rubbing their hands with glee over an AI bubble burst may be somewhat pissed when it rocks the global economy at the worst possible time for them.

However, it does need to pop and reality does need to shine a harsh bright light on foolish behaviour.

In the meantime, the "agentic" hype marches on and development carries on at such a frenetic pace it's impossible for companies to keep up.

The checks and balances are all but non-existent.

Companies will be unleashing poorly thought out agentic systems which WILL fail at massive scale.

Even if the systems themselves don't hallucinate themselves into infinite loops, the ongoing AI security arms race is going to make so many of them huge targets for hackers, for bad actors.

Taking too many humans out of the equation is just so massively dumb at this stage, it beggars belief that it's already happening.

Unleashing these agents and entrusting them with complex multifaceted tasks? Madness. Absolute madness.

When this one bursts, it's going to be an almighty mess - but I don't think we're even close to maximum balloon inflation yet.

Sadly, I reckon we're going to have to suffer a good few years longer yet and there's going to be some MASSIVE clusterfucks that are going to cause a whole lot of hurt.

Some of the smaller ones are going to be hilariously sad, such as entrusting agents to run and maintain compute clusters which go wrong and end up costing companies millions in cloud compute fees!

Oh yeah, that's one to watch for sure - get the popcorn ready.

[Doctor Syntax]

Re: Stopping treating it like a replacement for human Labour would be a start...

"That's the thing, those rubbing their hands with glee over an AI bubble burst may be somewhat pissed when it rocks the global economy at the worst possible time for them."

The longer it goes on, the more investment it eats up. The more investment it's eaten up the worse it will rock the economy. There's no point in hoping that there'll be a safe time somewhere in the future.

There's no time like the present.

[Excused Boots]

Re: Stopping treating it like a replacement for human Labour would be a start...

Indeed, but firstly let’s assume that the current AI craze is indeed a ‘bubble’, no different to tulips a few hundred years ago. Yes if it bursts now, it will be bad, it’ll wipe out a number of companies, pension funds etc. Basically it’s not going to be good. Well, except for the likes of Musk and Altman, who, on paper will be a lot poorer, but they’ll survive unlike Mrs Jones in downtown Manhattan whose pension fund has been wiped out and likely getting nothing from now on.

But if it burst tomorrow or next week or next year, then the fallout will be even worse. Maybe even the likes of Musk and Altman feel real pain.

No sorry, who am I kidding, of course they won’t!

[MachDiamond]

Re: Stopping treating it like a replacement for human Labour would be a start...

"Well, except for the likes of Musk and Altman, who, on paper will be a lot poorer, but they’ll survive"

I haven't yet put anything in Elon's "skills" column that would let him claw his way back up out of the hole he's dug. If the gubbmint isn't in a position to hand out big piles of free money as they have been, all of Elon's enterprises are sunk. Just look at what Tesla made in net profits in 2022 and each year since. For 2025, there has to be some guessing about Q2 and Q3 as people rushed in the US to buy a Tesla vehicle before the tax credit expired. Q1 was propped up by Carbon Credits and the numbers from different markets are showing sales way down while the EV sector as a whole is rising. Without Tesla stock being worth ludicrous multiples of the company's value, Elon might be far out on a cracked yard arm. Starlink is rather expensive for those that don't really need it (rural with no other options). xAI is tossing $1bn/month into the wood chipper with the output not even suitable for compost. Poof, gone.

[MachDiamond]

Re: Stopping treating it like a replacement for human Labour would be a start...

"That's the thing, those rubbing their hands with glee over an AI bubble burst may be somewhat pissed when it rocks the global economy at the worst possible time for them."

Learn lots of different skills so you can jump in many directions when the revolution comes and try to not take on too much debt. A good store of food and stout locks are a good idea too.

[Mike 137]

Déja vu

"Simply bolting AI onto old processes doesn't work"

I seem to remember exactly the same problem when "e-government" became the rage.

[CorwinX]

I may have mentioned this before

El Reg should start a lottery of some sort on when (not if) the AI crash happens.

Winner gets an official Reg t-shirt.

----

Sorry, can't help myself, runner up gets two ;-)

[Bebu sa Ware]

Re: I may have mentioned this before

"El Reg should start a lottery of some sort on when (not if) the AI crash happens."

A misspelt Chinese knock off of the MSDOG sweater might be a reasonable prize for this "guess the weight wait" competition.

My prediction — first or second week of November 2027… specifically 686 day from today (2025·12·25)

[Random as if !]

TL>DR

It's a search engine

[coredump]

Re: TL>DR

Pretty much.

The only diff I can see from here is the "AI" peddlers have programmed their search engine to return results in (more or less) complete sentences, rather than URL's and such. So it fools more people into thinking there's magic behind the curtain.

Thing is, regardless of the presentation, both kinds of results could very well be complete bollocks. But it's easier for the inexperienced to disbelieve a page of URL's and search results snippets than a computer which "sounds intelligent" or whatever.

[David Hicklin]

Re: TL>DR

> "AI" peddlers have programmed their search engine to return results in (more or less) complete sentences

Which inevitably are complete garbage, I just ignore them and scroll down past them as fast as I can

[takno]

Re: TL>DR

Let's not get ahead of ourselves. It's not that good.

[Mimsey Borogove]

Re: TL>DR

It's a search engine

It's not a search engine. If what it tells you is accurate, and if the citations listed at the end are accurate, it might have found something decent. But you have to check that anyway, because if it doesn't know the answer to something, it makes it up. It might get wrong information from the right place, or right information from a non-existent place, or it might just make up the "information" and the citation both. Since you have to check it anyway, you might as well just do the search yourself.

Besides, when is a search important enough to throw away the amount of power used by a LLM (yes, they're large language models, not any kind of "intelligence")? Hundreds of dollars'-worth of power and water wasted for what? To settle a bar bet? To tell you when the latest blockbuster will be on at your local theater? You don't need "AI" for that, you just need a little of the old-fashioned kind yourself.

TL;DR: It's not a search engine.

[LateAgain]

"the AI tools didn't learn."

Strangely in the terminator story where the terminator DID learn, it was because Connor got to the AI chip and flipped off the write protect.

(That bit was in the book but not the film)

[s151669]

Re: "the AI tools didn't learn."

It was in the directors cut.

[dlc.usa]

Management Has Moved On

from airline magazines to AI magazines.

[sarusa]

Ahaha, no

The problem is that 'AI' is only more competent than useless at very specific things, like OCring text, upscaling pictures, doing translations. And you have to know what those are. It is a tool that is good at very specific things. Like a knife, a craftsman knows not to hammer things with it.

For things like coding, doing budgets, doing schedules, summarizing meetings, LLMs follow the usual AI rule - It is absolutely better at producing something than completely incompetent twats. You get something instead of the nothing they could have done themselves so they think it's pure magic. For anyone competent, AI just slows things down because it does worse things (and radically wrong things) than we could have done on our own in the same time and we have to spend more time fixing and checking it.

But then... management is usually completely f@#$ing incompetent, so it is kind of a big plus for them? So I guess the big takeaway here is just replace all of management and all of the executives with LLMs and everything will be much better. I'm sorry, I guess you were right, it really is a management problem. Management needs to be replaced entirely by 'AI'. But by no means attempt to use it for real work, it mostly sucks at at that.

[Bebu sa Ware]

"AI is absolutely better at producing something than incompetent twats."

Could be the OpenAI corporate motto or mission statement.

Even if it were, the incompetent twats would still queue to buy OpenAI Kool·aid.

[cd]

Re: Ahaha, no

"The problem is that 'AI' is only more competent than useless at very specific things, like OCring text, upscaling pictures, doing translations"

Put another way, plagiarism.

[MachDiamond]

Re: Ahaha, no

"Put another way, plagiarism."

Plagiarism is copying off one source.

Research is copying from many.

AI is research with sources invented to fit the summary (done first).

[Ben D]

Re: Ahaha, no

Agreed - OCR is a great example of a boring area that is being transformed by LLM. Better results, less time training, and the people at the coal face want to use it.

[iron]

Re: Ahaha, no

No need to use an LLM for OCR, ML has been doing it for years and is cheaper in terms of cost, energy, time and every other metric.

[Boolian]

Man Down

What the article is screaming, is that AI is worse than a man short. It requires several additional skilled employees with deep insights into the company business, to hang over a congenital idiot and hold its hand.

We have all been there. No value is derived from, or extracted from them.

When you hear the phrase "He's a good wee worker if you keep him right" you've got a chancer and a waster on site - one is doing half a shift while the other is doing worse than half a shift for him.

[dmesg]

Re: Man Down

"It requires several additional skilled employees with deep insights into the company business, to hang over a congenital idiot and hold its hand."

So, shall we initiate a new Reg -standard abbreviation for AI? "TBN": The Bosses Nephew. We just need to figure out where the apostrophe goes.

[Armus Squelprom]

"AI won't transform business until the enterprise is willing to transform itself."

AKA

"AI won't transform business until the enterprise is willing to subordinate humans".

[Anonymous Coward]

The phenomenon outlined in the article is *not* specific to AI implementations.

What is being described is a misalignment between what the Business "wants" and what the Business actually "needs" to improve productivity/profitability.

This is perfectly illustrates why it is advisable to engage an external pair of eyes before you start any Business Transformation project, otherwise you are just pumping money into a bottomless pit.

This is the raison d'etre of the Business Analyst.

[An_Old_Dog]

Office Politics and Business Processes

it is advisable to engage an external pair of eyes

Office procedures are formed and warped by office politics. The less-successful businesses are unwilling to make the necessary rationalisation and streamlining changes to become more-effective.

Make the changes, then see if the technology (computers, AI, whatever) is worth using.

A place I worked at had a technical review of our server setup (we had low-hundreds of Novell servers) each year by Novell staff, as part of our support contract. For ten years, the various reviewers made the same set of recommendations and remediations, yet the changes were never made.

[David Hicklin]

Re: Office Politics and Business Processes

> the various reviewers made the same set of recommendations and remediation’s, yet the changes were never made

My experience of that is when you look at the list they made, you quickly realise that they don't have a clue how your business works and the changes would just makes things worse.

[Denarius]

Ah yes, the advisers/ consulting firm that borrowed your watch to tell you the time theme. How about asking those who may know, such as coal face staff and instantly dismiss any manglement who insist on being included in this first stage of information gathering

[AdamG]

"The phenomenon outlined in the article is *not* specific to AI implementations."

Completely agree.

But on the rest, I completely disagree. What is being described is incompetence, and for solving that, there is no substitute for competence.

With regards to what you said specifically: the business doesn't "want" anything, it's not sentient. You're talking about what idiotic people *within* the business want, from a personal bias and completely independent of what the business actually needs to improve. (But I suppose maybe that's what you meant with "wants" in quotes.)

Stupid managers who think AI/LLMs (or whatever else) are a magic bullet that they can use to avoid doing their job (or to avoid paying other people to do other jobs), are also going to be stupid when it comes to choosing/listening to BAs and consultants.

You might be lucky and end up with a BA who actually helps, but you're just as likely to end up with some idiot that isn't qualified to analyse anything, and who just tells the manager what they want to hear / sucks up to them to keep getting paid, kicking off all kinds of wasteful projects and distractions, and that burns money just as badly.

No, if that's the situation you're in, there are only two options: the management are corrected, or they are removed and replaced with people competent at the job.

[RojCowles]

"The models are more powerful than ever."

[Citation Needed]

[Uncle Slacky]

I think this relates to the SNAFU principle, as in “It promoteth growth, and it is very powerful.”

(true meaning: “It is a crock of shit, and smells as of a sewer.”)

https://www.catb.org/~esr/jargon/html/S/SNAFU-principle.html

[jpennycook]

agile again?

> Simply bolting AI onto old processes doesn't work. Yet that's what most companies do. They treat AI as a plug-in to existing workflows that were never designed for predictive or adaptive tools.

So management didn't learn their lesson when they bolted-on agile to waterfall? AI-agile-waterfall sounds terrible.

Maybe we should add some blockchain for suppliers, then issue NFTs for customer engagement?

[Roland6]

“ The evidence is clear that this isn't a tech failure. It's a management failure.‘

Clearly need to clean the glasses (aka spectacles - for non-English readers)

The evidence is clear - is the normal disconnect between hype and reality.

I would like, just for starters, the article’s author to explain how Mark Peace’s AI Troubles [https://www.theregister.com/2025/12/10/vibe_coding_is_good_enough/ ] can be ascribed to “management”…

Been here before with previous IT “silver bullets”…

Which would suggest author has no previous experience of IT hype cycles…

[Roland6]

Re: “ The evidence is clear that this isn't a tech failure. It's a management failure.‘

Going further, it is telling that the “AI” evangelists only talk about hand sweeping “AI”.

Stepping back into the sort of detail people here are more accustomed to, “AI” is just a label for “statistical programming”. Ie. “AI” is just a field descriptor, like “procedural programming” - something ElReg readers will be very familiar with. Hence in many “AI” articles it is possible to replace “AI” with “Procedural Programming” and the article still to make sense…

Looking at things this way, we see many are already using AI, just that they aren’t calling it AI, instead they are calling it by the type of AI being used: image analysis/processing, offer selection, etc.

So for example if your call centre/website is creating a list of offers for a specific customer/caller, they are using AI…

So it is telling that an AI evangelist isn’t ware of this…

[[e-mail]]

Investment capitalism can be weird

Quite often in business, being the one guy who figures out that a business can generate profits is the best thing. This is the Warren Buffet thing. Investment capitalism is different. The profitability of the business is besides the point as long as you can convince other people that they should buy your stock for more than you paid. You want everybody to believe that the investment is great and thus bid against each other. That's what's going on with AI. There is little evidence that the development at Open AI, Google, Microsoft, and others is AI, much less profitable. Those early investors are not going to let the high probability of failure stop them from trying to get people to pay more for the stock than they, the early investors did.

[M.V. Lipvig]

I'm certainly doing my part

... to help it fail. At work, I add the information management decrees with ai, then proceed to ignore it. I don't care if it's accurate or not, I don't even look at it. At home, I refuse to use it, turn it off where possible, and find non-ai products to replace ai products. I have zero interest in it.

[tyrfing]

This sounds a lot like the problems with automating any other process.

The current process relies on people moving stuff along and handling exceptions without really having to think about them.

Automating it, for instance using workflows, requires mapping things out in detail. Including what happens when "unexpected" stuff happens.

This part is far more difficult than actually programming the automation part.

Especially if you have to persuade people to use the new version and not fall back on the old way. Which means you have to completely map things, or people *will* move back, and good luck getting them to trust it again.

AI is being sold as "well it will just handle that stuff". Which is completely false.

[nsimic]

No. Just no.

I keep hearing this all over the place."AI works! You are just using it wrong"

Everyone compares AI with some of the previous enabling technologies like the internet, mobile or even software itself.

The gains in every one of those technological revolutions were immediate and obvious. The only reason there are no AI related productivity/revenue gains right now (besides for the companies slinging AI hardware or software) is because there aren't any to be had.

[David Hicklin]

So to summarise the summary

AI is:

1. Still a solution looking for a problem that has not been found yet

and

2. It is just like a new starter at your workplace who on day 1 has no idea about processes, where to coffee comes from etc . But unlike the new starter it never learns and on day 365 it is *still* a new starter.

[TheRealJoMo]

Not a Paradox - just collective hysteria

“This is the paradox of the current AI boom. Adoption is high, hype is higher, but meaningful business impact remains elusive. AI is everywhere except on the bottom line.”

Hilarious.. a “Paradox” ? No this is not a Paradox its called collective insanity. The entire economy of the USA is being focused on something that in 95% of cases provided no measurable value. A Paradox? No its just feckin nuts.