Microsoft's Copilot AI to pervade the whole 365 suite Microsoft's 365 productivity suite is the latest to get an injection of AI, with Redmond using what felt like an hours-long webcast to say Copilot AI will soon be available across the entire suite. If you're aware of Microsoft's addition of Copilot to its Dynamics 365 CRM and ERP software earlier this month, then you already …
The guys who run the vast corporation I work for are going to love this.
Pretty soon they'll be able to get rid of all the expensive messy humans they are forced to employ and the business can turn into a bunch of AI chatbots asking each other where the raw materials are and trying to figure out why deliveries are late.
Presumably our customers' chatbots will be mollified by a nice lunch out or a round of golf or whatever it is these guys do on their days off.
This is a capitalists dream.
The chatbots will be having endless meetings with each other. To make them feel more real, some will turn up late.
They'll start spouting "I'll circle back on that. We need to touch base and meet this head on. We can meet for coffee to discuss". Another will reply "We can't, we're just bots. We can't meet for coffee". The other bots in the meeting will be playing buzz word bingo. One chatbot in the corner will be struggling having been told they're all just bots as it didn't realise.
The meeting will end, it will be seen by the real humans as "A success. Fuck all got done like all our normal human meetings but the beauty is, we didn't have to pay anyone".
So Jane, what's your opinion on the new business plan?
"By leveraging our competitive advantages, and increasing our sales effort, large gains should be possible in the next quarter"
Alan, your thoughts please...
"By leveraging our competitive advantages, and increasing our sales effort, large gains should be possible in the next quarter"
Steve?
By leveraging our competitive advantages..."
Stop right there, Hello? IT? Can you remove everybody's access to this Copilot thing please?
How many managers can be replaced by these features? Most managers are simply pencil pushers and that bullshit can now be generated using automated means.
Seems only fair that the managers who decide to enable these features will be eligible for replacement by the decided features. Must also be a significant reduction in costs.
Be careful what you wish for. Even if it were simply a useful tool meaning that "knowledge workers" are 3x more productive, it would still put 2 out of 3 "knowledge workers" out of work..
However it's worse than that. It replaces them with "believable bullshit" of infinite volume. It threatens to drown out human discourse with a flood of crap that can only be productively used for disinformation and fraud.
But what I really worry is: As this thing improves, people will start to trust it. They may start to trust it so much that they throw their lives into it just like they did with Facebook and Google. They may fall in love with it, and even worship it. At that point could become a kind of religion. A "Spiritual Opium" in Marx's original sense. At that point, Microsoft has complete control over your lives..
I know I have said this a lot recently, but All of this (and, er, a few other things..) was predicted 23 years ago, by Deus Ex, a PC game based on a mash-up of conspiracy theories from Usenet etc in the 90s.
https://www.youtube.com/watch?v=WQZV5lwQZbY&t=11237s
How could a humble PC game make such accurate predictions about the future? One explanation is that some of the conspiracy theories it was based on turned out to be actual conspiracies.. :|
Be careful what you wish for. Even if it were simply a useful tool meaning that "knowledge workers" are 3x more productive, it would still put 2 out of 3 "knowledge workers" out of work..
It's not a zero-sum game though, not in the long term anyway. Just because mechanised looms put weavers out of work, doesn't mean that there's more unemployment now than there was before then.
If it actually works as advertised (and that's a big "if"), workers will end up doing more and/or different things with their time, and probably under more pressure.
(Thinks: for example, handling the court cases from people who had their cases messed up by AI support agents and were unable to get through to a human. It's when the courts get AI-driven that we're really screwed)
How could a humble PC game make such accurate predictions about the future?
By paying a little attention to literature and psychology?
There is nothing surprising about what's happening with LLMs right now. The technological advances are in line with what some people were predicting when DL showed more staying power than other flavor-of-the-week approaches in natural language processing, and very predictable once Vaswani et al. published "Attention is All You Need" in 2017. The economic and social consequences follow straightforwardly from those, because people Love New Shiny.
Obviously I added the last line :P and I did skip a couple of lines. But I did link to the original dialogue.
I'm not a fan of DX:Revision though - it doesn't change the dialogue but it does change the maps and character models, in a way that I think detracts from the game sometimes.. However, that seems to be the best YouTube vid of just the dialogue from the game.
The AI "morpheus" was not originally in such a grand room - it was a toy prototype of the corporate-state AI "Echelon V" that had been kept in Morgan Everett's bunker - it explains a lot of the backstory yet it was a part of the game that you wouldn't necessarily find on the first playthrough. The replayability was one of the best aspects of the original game, IMO.
I'd recommend "Deus Ex Transcended" for the best way to play the game on modern hardware while not changing the original game beyond simple graphical improvements.
Can see that being popular at HMRC.
Open Excel, look at data, look at agenda, ask Copilot to make numbers more believable and aligned with the agenda. Create a report, add a caveat in an annex that nobody reads that the data is "modelled" and the rating of accuracy is highly uncertain, so that when anybody questions it you are covered.
Make a shiny presentation, how the given policy is working as intended.
Receive pats on the back.
It’s nice to see MS jumping full-in on the latest bandwagon. And so quickly too! It was only a couple of months ago that nobody had even heard of Chat-Wotnot
They did the same when the Internet became a “thing” - suddenly everything in Windows wanted to talk to the Internet regardless of how un-helpful and inappropriate it was. Want to check that birthday list you kept in a text file? Hey - instead, why not search the new fangled Internet thing for “text files” or some such nonsense instead?
I really hope that Microsoft are not going to roll this out wihtout a mechanism ALREADY in place so admins can kill it stone dead.
Something tells me after the "fiaso" of "Reactions" that they will roll it out without a way of killing it.
What data does it collect and what does it do with it? There could be major privacy issues using this.
Let's speculate!
It will be on by default for all tenants, obviously - almost goes without saying.
The process to turn it off will be announced in the depths of one of their massive weekly bulletins (it probably won't get a 'major change notification' - those are reserved for new emojis and other such groundbreaking advances).
The bulletin mention (which may have a vague, unclear heading) will tell you that when the facility to turn it off becomes available in a few weeks (by which time you'll hopefully have forgotten about it) you can turn if off simply by running this many-line Powershell script against your tenant. There will of course be no simple toggle-switch to turn it off, and the script will have to be run individually against each 365 service to turn it off (and if new 365 services are added in future, it'll need doing for them too as they will have it turned on by default). In order to check which services have it on, you'll need to use Powershell - there will be no indication in the web GUI.
When you come to run the script you'll find that it doesn't work because there are some prerequisite steps which they haven't mentioned in the instructions, so you'll have to work them out first.
Then when the ability to turn if off actually appears, there will be no announcement / notification at the time, as hopefully you'll have forgotten by then so will leave it turned on. There will also be no indication in the web GUI.
If enough people get wise and turn it off quickly and take-up remains low, a few months later an option to allow individual users to opt-in to a 'trial' of the exciting new service will appear. For the process to block this, see above.
And so on...
After you've done all that is when they will tell you the only way to disable it is to "request" your tenant is added to an "exclusion list".
Having raised said request it will then take them ages to get around to doing it.... And then keep pestering you to review their "very helpful" support....
> What data does it collect and what does it do with it? There could be major privacy issues using this.
Too late! They have already collected the data to from all of your O365 documents to date, all your teams call transcripts and chat logs, all your GitHub contributions and comments, etc.
I suppose it might have it's uses, but I have no desire to find out what they are. I survived clippy by deactivating the damn thing as soon as it appeared, and doing the same for an awful lot of domestic and small business users. If this fresh hell doesn't have an off switch that is both easy to find and, crucially, stays off, even after updates, I foresee anyone who can jumping ship for libre office.
As I've said in other comments, these things are tools, to be used with an understanding that the material they produce might be utter cobblers. Used with that in mind, and enough knowledge to ensure you can spot errors, they can be handy time savers. But wedging them into every nook and cranny of the most widely used productivity tools on the planet isn't just asking for trouble, it's begging, pleading and demanding trouble.
The common or garden office drone cannot be expected to contend with that - many of them have enough trouble just using the tools they have, without foisting this kind of uncertainty upon them. Expecting them to double check what the fancy new thing tells them is both unfair and unreasonable - they won't do it, won't realise that they need to, and likely wouldn't have time to do it anyway. They wouldn't be asking the thing to prep them for a meeting if they had time to do it themselves!
Of course, I gather that this impending chaos is based on GPT4, and we haven't really seen what it's capable of yet. It might be much better than the version that proclaimed the demise of the unfortunate vulture mentioned in the article, and we might be being terribly unfair in prejudging the likely outcome. I doubt it, but we might.
But that doesn't really matter. The sheer scale of the potential for cockups is enormous, Microsoft don't exactly have a reputation for reliability, and I feel that a healthy dose of scepticism is absolutely essential here. Be afraid, be very afraid.
I can easily imagine companies starting to put into contracts wording to the effect of "you won't be using MS 365 or any other AI-enabled tools or cloud services to work on any part of this project" simply for fear of corporate information leakage.
This could well be the footgun that kills off the Office franchise if MS aren't very very careful.
We have Teams at work and the "Auto suggest" is fucking annoying as I've mentioned before. I'm tempted to turn it off for everyone but guess I should actually ask the business if it wants it on or not. I've turned it off for myself.
The other day I said to my work mate "I don't do breakfast. I need to go on a diet". She had left her desk when she got back and unlocked the auto suggests had come up and she's accidently clicked one (I hadn't seen this, she told me after as she'd deleted it before I read it). She said the auto suggest she clicked was "Yes, you do" :) Nice of Microsoft and Teams.
I'd be happy if they applied the same sort of effort to the help systems in 365 itself. Every time I use it, there is a guaranteed 10 minutes fighting to work out where some option has gone. I can see how this will go:
"How do I remove the green shading on the bullet points?"
"I'd advise rewriting your whole document with a more punchy stakeholder oriented style. Can I do that for you?"
Would it be possible to ask the AI to write an article that when analyzing the generated article again by AI it will summarize the key points you wanted to be interpreted in the first place?
Seems to me to be somewhat redundant or highly useful to influence stupid lazy people, and according to the theory of stupidity there are many more stupid people than you think.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
