Microsoft starts boiling the Copilot frog: It's not a soup you want to drink at any price In 1968's Star Trek episode, "The Ultimate Computer," Captain Kirk had his ship used to test M5, a new computer. A copilot, if you will, for the Starship Enterprise. Designed to more efficiently perform the jobs of the human crew, the M5 indeed did those jobs very well yet with such a terrifying lack of understanding it had to …
... I was a visiting researcher at a US university some years ago and heard the tale of the Most Amazingly Engaging Powerpoint Presentation Ever!
One of the senior grad students presented a deck of slides, each with a different color scheme, fonts, and alignment/bullets combination. In one of the slides, he presented the research question he was working on, a very long paragraph in a smallish font (and he presented it talking like those car ads, in muted tones and very quickly). The next slide was a single "NO!" in a huge font.
The people who witnessed this miracle remember the craziness but not the details of the contents. Mission accomplished, I say.
A few years ago we had a guy from a storage vendor in, he spoke to a load of developers about the wonders of their object stores by on drawing on a flipchart and had pretty much sold their system by the time he was done.
I'd love to be able to speak publicly like that.
Having someone sketch it out in front of you is way more engaging than sitting through a bland-o-gram of powerpoint whether human or AI derived.
Not very good at drawing? Join the club, I'm rubbish. It doesn't matter. In fact it will seem more authentic.
Not very confident without notes? Ditto. But that's fine, have a list in front of you. Don't be afraid to keep checking you've covered everything. If asked, share them with your audience afterwards to remind them of the key points.
I’m so glad I (semi) retired from the workplace a little while ago.
Over the last few years a steady proliferation of meetings, reports, corporate communications, etc similar to what’s referenced in the last paragraph of the article had steadily sucked much of the joy out of a career I’d previously enjoyed. The thought of that being amplified through the application of the current generation of AI tools is just too depressing for words, and I’m just too old to kick against the bullshit the way I would have done 20 years ago.
Anon because I’m still doing a couple of days a week of consultancy for my former employer and I’d rather not jeopardise the income stream…
A fair question. At the time I first registered an account with El Reg (way back in the halcyon days of Verity Stob, The Moderatrix, and Flame of the Week) and started interacting through these forums it was a kinder, gentler world where the likelihood of an employer (or colleague) stalking you in what we rather quaintly called cyberspace was negligible.
By the time it started to seem like it might matter there was already a substantial body of (apparently not too) incriminating material out there under my real name. Still don’t want to put that sweet, sweet consultancy cash at risk though… :-)
A fair question. At the time I first registered an account with El Reg (way back in the halcyon days of Verity Stob, The Moderatrix, and Flame of the Week) and started interacting through these forums it was a kinder, gentler world
You have a very different memory of the Moderatrix to the rest of us!
Anon too because I spent my years working to fix data collections and computing calculation errors for my former employer, keeping all the users happy with the products. They moved on and left me behind and still working on the concept "say that everything is accurate" ... still "showing that everything is accurate" but never helping the users VERIFY that it's accurate.
That seems to be the universal software world now.
I agree completely. Now is a good time to scream "Screw this happy horseshit!" and retire. Which is what I did when I lost my major contract at 60 years old.
Trying to find work at my age was bad enough before the market got flooded by layoffs. Now we're supposed to compete with a machine because some ad says it can do useful work these so-called "AI" tools can't actually do?
I hear you. Slightly younger than 60 here (although not by that much…) and I lost my job 18 months ago. After 12 months of coasting on the severance package and catching up with all those deferred DIY jobs, I began seriously job-hunting 6 months ago. It’s depressing out there in our industry; every week, day almost, seems to bring news of yet more layoffs at Mozilla/Intel/Dell/Microsoft/HP/whoever.
I was offered - and accepted - a job a few days ago; pay’s at the lower end of my range, and the role, while it’s something I can do, wouldn’t be my first choice, but beggars can’t be choosers, et cetera.
Anon cos.
I noted the recent glorification of a UK former plunger fighting in Ukraine also.
https://amp.theguardian.com/uk-news/2024/nov/10/british-man-killed-fighting-for-ukraines-foreign-volunteer-platoon
Meanwhile - Shamina Begum - groomed as a child - boo, hiss, citizenship unlawfully revoked.
At the risk of channeling Beth Mole I would be thinking cerebrum rather than cerebellum.
Although MS present a fair semblance of brain death which might imply their brain worms have been partying in MS' brainstem.
Slightly paraphrasing, presumably an Encyclopædia Galactica entry under Microsoft Sirius Cybernetics Corporation:
"Genuine People Personality products were imbued with artificial intelligence and emotion, however, they are known for being pointless or irritating, often serving their desired purpose but in an unconventional or unhelpful way." (And people aren't?)
Quoting Marvin: "Ghastly, it all is. Absolutely ghastly. Just don't even talk about it."
...except to today's Senior Management, who are too young and hip to have ever heard the expression.
Computers are wonderfully useful tools, but, like all tools, the user needs to wield them with the knowledge that, if misused, they are capable of inflicting injury.
"It's not skepticism, it's enlightened self-interest."
-- a former boss.
PLease, why do you pretend that management are fools, they only exist because their entire position is about bullshit.
Anyting powerpoint or management or motivation is all about pointless bullshit, the only difference is the media, some speak, some clipart etc..
"If you're the human at the end of this AI-smart question and want to look smart enough to answer it, who are you gonna call? Copilot."
This one's easy to deal with. You throw it back at the "questioner". "You got Copilot to write that question? OK, get it to answer it. Next question, please."
I think maybe you missed the point here..
The problem the author is referring to is that all work conversations will eventually degenerate to "multiple chatbot instances chatting to eachother", with the humans just serving as a noisy communication medium between them..
In a bizarre inversion of who/what serves what/who in technology, we become slaves to the machine as the last "real human voices" get drowned out by those who are simply saying whatever the chatbot told them to say.. (ironically, in order to appear more knowledgeable)
I can't wait for the time when I can leave my AI to entertain itself having a Teams meeting with your AI. Both can create summary documents and us humans can safely ignore them. Why? Roll forwards a number of months and imagine the resulting conversation with a regulator/officer:
"And tell me, Mr. Smith, who signed-off on this change of process that directly infringed trading law? We need to know who we should be holding to account"
"It was the AI, Sir. No human staff members were present in the meeting".
"Umm..."
Here there be dragons.
> Even the seemingly innocuous business of Copiloting the Dall-E-powered Designer graphics tool into documents is vile poison. The pictures generated can have no insights beyond what the generative prompts contain, diluted by whatever fantasies exist in the heart of the machine. But those pictures in a report or a presentation trigger expectations of extra meaning, of explication. The reason clip art feels so empty, so tiring, is that it speaks up our cognition and delivers nothing.
You mean like the "every article must have a completely meaningless shitterstock image at the top" policy that the Reg adopted some time ago?
Mercifully, you culled them from the article body page. But they still appear in the "preview", either on the homepage or when a link is "shared" on some app
We have the possibility of Copilot creating PowerPoint presentations to be read by Copilot, which will generate more of same to show Copilot - have managers realised their job is being taken by AI?
I'm reminded of the "The Electric Monk" in Douglas Adams' "Dirk Gently's Holistic Detective Agency".
The Electric Monk was a labour-saving device, like a dishwasher or a video recorder. Dishwashers washed tedious dishes for you, thus saving you the bother of washing them yourself, video recorders watched tedious television for you, thus saving you the bother of looking at it yourself; Electric Monks believed things for you, thus saving you what was becoming an increasingly onerous task, that of believing all the things the world expected you to believe.
If that was written today would the Electric Monk actually be a Mechanical Turk, with a few buildings in a Low Cost country filled with thousands of their finest 'data scientists' 'leveraging data insights' to 'unearth new paths to value' like the recent Amazon Whole Foods AI thing....
Pop up its sidebar in Loop or Teams, and it can auto-summarize what has been said.
Presumably by sending the discussion to some central cloud-based server for analysis, something which almost certainly violates the security rules of every big company (ceryainly those I've worked for). How long before we see Copilot being banned in such environments?
"Presumably by sending the discussion to some central cloud-based server for analysis"
If it's Teams, then the discussion is already there in the cloud. Maybe the copilot bot sends it to another different cloud. But that would be idiotic: having the same corporation's different bits of software duplicating the information and sending it half way round the world in the process. So that's probably exactly what happens, you are right.
"How long before we see Copilot being banned in such environments?"
Where I work, we are able to use cloud based co pilot, but the restrictions on what you can send are very strict. We are moving to a cloud based M365 setup and because of the dangers of co pilot learning from input, I suspect that the highest level of corporately sensitive info we can send to co pilot will still be lower than the documents we will routinely be storing on the M365 cloud.
We also have our own garden walled LLMs that don't touch the internet, so we can use them for more corporately sensitive stuff.
A friend down the street's mother, when I was a medium sized kid, wanted to know exactly why that band was called, "The Grateful Dead". At the time, I hadn't read Jerry Garcia's answer to the question. So I replied to her, that sometimes life was so excruciating, dying would be a mercy. This is not the correct answer. But ever since that day in the 1960's, I have suspected that my supposition might still be applicable in certain circumstances. This may be one of them.
> Satya Nadella, Microsoft's own Captain Kirk, said at the end of October that AI was due to deliver $10 billion revenue in the next quarter.
In what universe is their Copilot shite possibly delivering $10B/quarter in revenue? This has to be some accounting fraud that says 'all our BIng ad revenue counts AI revenue now since Bing has some worse than useless AI results' or 'any Windows install that ever has Copilot installed (before it's ripped out) counts as AI revenue.' Or maybe they're just counting all OpenAI revenue since they're the major investor (and ignoring how much money OpenAI is losing on hardware and training costs)?
You summed it up with the accounting fraud section.
Of course it's not fraud, it you, quite rightly, throw it in with other figures such as revenue from Bing Advertising.
Revenue growth is everything to Investors. Sustainable profits are irrelevant. Once the bubble bursty, the Tech bro's will just invent some other bullshit hype machine.
Now if you don't mind, I need to look at my NFT "art" in the Metaverse.
Maybe corporate subscription costs for Microsoft products generally will nudge up and the increase is being called AI revenue. Maybe Microsoft have shares in (or even own) the data centre suppliers or the power companies that power them and in some strange self charging accounting cycle, they are making money that way. If I understood it, I'd be doing it too.
So worry less. But do try to avoid AI Clippy. And Teams for that matter.
Whilst it should be easy to spot a member of staff/school pupil parroting AI, perhaps a little red dot indicating that a user has AI turned on would be helpful.
In the meantime, we should all look forward to watching the epic fail videos that AI will inevitably spawn, whilst having an offline machine to fall back on, for when this all goes TU, as all Microsoft gimmicks inevitably do.
Whats the point of having everyone use Co-pilot and generating the same Powerpoints and clipart etc ?
I guess this validates my point that PP is a complete waste of time and a vehicle of people's who job is completely unncessary and a tax on time that could be used for better purposes.
My advice, fire anyone who makes PP presentations and similar as their primary task, and fire the people who requested it.
Imagine all those user-generated bots, er, AI constructs, replacing our actual faces with their perfect ones in every Teams meeting. While the people whose time would have been wasted "participating" in such a meeting can, you know, do productive work. Or hang out around the water cooler and discuss.... stuff.
And the fossil fuel extraction industry will love it: re-open more coal and hydrocarbon-fired electric generating plants do Micro$oft and others can open more GW-gobbling AI GPU-farms. Feed that loop. Kill that planet.
"I'd love if I could just search for an email in Outlook and get some context in my results like a Google search. It still uses straight keyword search from 1992."
I'd love it if that straight keyword search actually worked properly. I use outlook at work and the search generally just fails. I can manually find emails that it won't find that contain the words I am searching for. And vice versa it returns emails that despite opening and searching inside, I can't find why it returned them. Also what grinds my outlook gears is the utterly accurate but equally useless "there might be more results on the server" message, which omits from the end "but I just can't be bothered to look right now".
...you have to micro-analyse every word, graphic, chart, and table it produces, double-check facts, and just generally micro-manage it as if it was a 1st-month intern. In other words, you have to take more time over its output than you would have done producing it yourself.
The difference between an AI and an intern is that you have to keep doing that for as long as you keep using it - AI is never going to outgrow the 1st month stage, at least not in terms of trustworthiness. We've all seen the reports of ChatGPT and other bots Making Shit Up and arguing the toss over patently untrue statements - MAGA politcians, in other words - so there's really no point in ever going down that road, as far as I can see.
Thank god I'm old and will never have to work in a coporate environment where Copilot drives the dystopian nightmare. It's bad enough living in the dystopian political nightmare driven by China, Russia, and idiots like, well, the entire UK Conservative Party.
Ah no, but that's where you're wrong. It will outgrow the 1-month stage, just not in the way the human does. It won't get better at doing its job, but will get better at looking like it's doing its job. The errors will get harder and harder to spot, so the improvements in AI will lead to us simply having to do more work to find the errors that no human could ever have made.
As a human, looking for human errors is easy as we know how humans think, what with being one ourselves. We cannot understand AI, so we cannot anticipate its errors by reflecting on its thought processes.
AI is a great tool, it's just being oversold and in many cases used inappropriately.
Those too greedy and quick to use it to replace humans to cut costs and deliver less will get burned by the market. As will those churning out rubbish systems. It will. just like the Internet or the TV be used to promote selfish narrative as truth. The same old adage applies; rubbish in, rubbish out - it's all about the data and training. The majority of people still haven't learned to be sceptical of what the square screens tell them.
Conversley, working as an assistant on a pure task such as coding, they're great even if it doesn't always work, it presents ideas from many that would take 10 times longer to research. It can find information from reams of documents quickly and often assemble disparate pieces of info.
I had a really instructive copilot session yesterday when I asked it how to do something in typescript (iterate over the keys of a type). Over the course of our consersation it generated about 10 variations of a function, always taking into account my feedback and trying to fix it. None of them worked at all. I looked on stack overflow and discovered it was not possible to do what I wanted. I didn't have the heart to tell it.
In addition to this, Copilot will turn into the ultimate lock in for Microsoft. Users will become so dependent on the prompts and the exact context that Azure will provide that they will be unable to produce documents, presentations, meetings, notes etc without that exact environment. They will not be able to function in another company, nor will they be able to transition to a different cloud provider. The company will be locked into Microsoft Azure and the context built there until it eventually goes bankrupt.
Trekkie here...
Clichéd and kitsch? I beg to differ. Compared to many other productions that were made literally as long ago as I was, Star Trek has held up fairly well, I'd say. Yes, the acting is in the style of the era (i.e. dialogue-heavy and by modern standards somewhat overacted) and the special effects are mostly limited to stop motion, double exposure and optical transparencies which looks pathetic to those who believe that LotR, Avatar or The Expanse grade visuals are essential to enjoying the cinematographic telling of a story. Yes, it's dated. But that is, AFAIC, more than offset by the 'cuts to the bone' factor that applies to most of the series, and its continued relevance to modern-world problems and moral dilemmas.
My main issue with this particular episode is the plot hole: a well placed phaser blast to M5's innards would have solved the problem quickly and simply...
This sounds very much like the shoe event horizon postulated in The Hitchhiker's Guide to the Galaxy. In this case, it's recursive, redigested information being thrown at consumers in an endless loop burying them in utter nonsense. Eventually you can't filter past the garbage fast enough to prevent the total collapse of the system.
There are a number of way of committing commercial suicide, one is forcing your audience to have product they do not want ask about Internet Explorer, U2 and Google+ how well that worked . Then you can associate your product with a perceived poison and AI is to large audience about as poisonous as tech comes. Each update and back it comes like herpes its removable (at the moment) but continually pushing features customers do not want inclines them to look elsewhere .
There will likely be a huge difference between businesses that use AI well and those infested with pointy-haired types who use AI without the vaguest hint of self-awareness. Perhaps we will see some sued to oblivion when they use AI-generated output that happens to breach A.N. Other's copyright.
My boss was sent a marketing power point slide show for the product my team produces. It had reasonably accurate information in the bullet points, but had been populated with AI produced images instead of screenshots.
His initial verbal response to us was "I wish there was some way I could unsee this".
The next half hour was spent with the team taking a look and having a good laugh about it.
I don't know how his feedback was received, but knowing him, it would have been done very professionally.
Anon in case the creator sees this.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
