Kevin Beaumontvibe working
vibe working
I am gone
lol
lmao even
I have access to the new vibe working feature (lol) in Microsoft 365 and just asked it to "make a spreadsheet about how our EDR tool blocks 100% of ransomware".
It did it, completely made up the data for me, off to send this to CIO now and go play Fortnite.
I also asked it make a presentation to compare CrowdStrike with Microsoft Defender.
Gonna log off for the day and go fight a goose.
Right, no need for Linux security controls, gonna pump this one out quick.
Amusingly it uses my own research about BPFDoor to justify this.. but BPFDoor doesn't use any vulnerabilities, it just makes that up.
I've tried the Excel, PowerPoint and Word vibe working tools now - you can give it any incredibly dumb idea and it'll write papers, presentations and make up data justifying it.
For bonus points, if you don't know it's a dumb idea... you won't know still.
The future is now. Vibe going out of business, soon.
“Microsoft says its (vibe working mode) in Excel has an accuracy rate of 57.2 percent in SpreadsheetBench, a benchmark for evaluating an AI model’s ability to edit real world spreadsheets”
Imagine being a hiring manager and hiring somebody to use Excel who says they do their job wrong about half the time. Welcome to AI.
Today I got Microsoft Excel’s vibe working feature to generate me metrics about how well my cybersecurity programme at Lumon Industries is going in a one sentence prompt.
Good news everybody - we’re doing everything brilliantly! All of the data is made up, but I logged off anyway and smoked a kipper for breakfast.
Kaspar (allegedly not haunted)@GossiTheDog Excel verified it! Can't be wrong then.
Panderito 
@GossiTheDog shareholders rejoice!
Henryk Plötz@GossiTheDog It spent 10 minutes on this?!
Now, while in principle I have nothing against sub-second, one-shot inferences, I must say that blasting a couple hundred watts for 10 minutes does not appear entirely appropriate if all you get is this kind of output.
Ketumbra@GossiTheDog as useful as any other KPI document I've seen to be fair.
Brandon@GossiTheDog [insert are you winning son? Meme]
Sophie Schmieg
Led By Gilded Fools@GossiTheDog Lumon Industries would like you to complete this test to prove you did the work.
johnduggins@GossiTheDog What a guy!
Simonoid@GossiTheDog so in one corner we have trained intelligent people with efficient brains.
On the other we have (possibly) dumb people generating arguable quality sheets and the data source still needs to be plumed in. Oh and it's killing the environment with inefficient models.
Cool and normal 👍👍
On the other we have (possibly) dumb people generating arguable quality sheets and the data source still needs to be plumed in. Oh and it's killing the environment with inefficient models.
Cool and normal 👍👍
Martin Hamilton@GossiTheDog Strong possibility that this is only happening so that the VP of Vibes can strut out onto a stage while gyrating unconvincingly to the tune of a certain Bee Gees song...
Felix DF3LX@m @GossiTheDog You mean the Vibe President?
A feral Natalie@GossiTheDog It may be right only half the time, but when your performance review hinges on it ... maybe you get a better performance bonus for being wrong.
I'm going to love the malicious compliance posts coming up for this half's review cycle on Blind 😈
Logan Fick@GossiTheDog Thanks for sharing your impressions on all this. I am dumbfounded there are people out there who think this is a legitimately good idea.
Huntn00@logan @GossiTheDog The straw that broke the Capitalists’ back…I mean they were already going down.
Simon ZerafaAs long as the theme from Shaft is playing in the background, you can have all the good vibes 😎
Shaft (1971) - Opening Credit
Antony
lambtor@GossiTheDog has this society hit terminal velocity yet on the free-fall?
toto@GossiTheDog I mean if I want something that’s right about half the time I will just throw a coin. There is something’s where „ai“ is useful. This surely is not it.
Si Dawson@mrtoto I believe Excel 1.0 was shipped with a function which does exactly that (RAND()).
Could have saved themselves billions.
PetterOfCats@GossiTheDog and you know if they are saying it's 57% it is much much lower than that.
Internet Rando@GossiTheDog this whole excellent thread about the future of work, and all I got was this awesome GIF.
Dirk HohndelBut Kevin... I'm sure it will get better. We may just have to change the benchmark a bit...
@GossiTheDog The fact that they even released it....
Alexandra Jusino@systemadminihater @GossiTheDog I guess 57% is better than nothing. Which I gotta say, is a sad state of affairs.
TheTomas@GossiTheDog According to the Pareto principle, I should be satisfied with 80%, which in turn means an acceptable Downtime of 6 days per month for IT. No company could operate and surive. But People keep telling this Pareto BS to IT.
Arthur van der Harg@GossiTheDog 42.8% inaccurate. Remember the Pentium floating point bug? 1 in 9 billion calculations were inaccurate. It cost Intel $475 million dollars. 1994 dollars.
And Microsoft dares foist this crap on the world?
Garrett Latimer@ArtHarg @GossiTheDog Unfortunately, the managerial caste craves the smooth taste of productivity-shaped objects. This prion disease will not have material consequences for them even as it gums up and suffocates companies with reams of bogsity.
Rupert Reynolds
Max Fenton@GossiTheDog love to guess which 42.8% is wrong
KanaMauna“Slightly more accurate than a coin toss” is not the winning proposition they think it is.
Juan Per¢ent,🇲🇽 🍉,🇻🇪@GossiTheDog source https://www.microsoft.com/en-us/microsoft-365/blog/2025/09/29/vibe-working-introducing-agent-mode-and-office-agent-in-microsoft-365-copilot/ "These breakthroughs allow Agent Mode to not only generate outputs, but also evaluate results, fix issues, and repeat the process until the outcome is verified. It’s like you’re handing off work to an Excel expert—while you steer and guide. "
@GossiTheDog got to love all these vibe worked quarterly earnings reports…. Hey we made more money than is available in the world, must be true!
VessOnSecurity@GossiTheDog If you pay the new hire only 1/4th of the original salary, that's counted as a pure win in business management circles.
Erik Jonker@GossiTheDog i first thought this was a joke...
Jan Sandbrink@GossiTheDog Hey, they did a vibe check on whether the AI is worth it and that check succeed. So they continue...
jonathanhoey
ifrik@GossiTheDog Does somebody has a link for that for me? Need to update some slides for a talk now....
Simon Oak@GossiTheDog my work is only profitable when we make no mistakes. And the longer mistakes pollute our system, the more costly they get.
Maybe focus on making AI detect existing mistakes? If it detects 50% that would be great, on one condition: your data must be a big enough mess.
I'm afraid that's opportunities everywhere! 🤣🤣🤣
DianeMicrosoft "It's a mediocre worker, but it will always say your ideas are the best."
Management: take my money.
Bastian Senst@GossiTheDog How well would a human pass this test?
Adrian Sanabria@GossiTheDog OMG no it didn’t hahahahahahaaaa 🤣
Peter Goulborn@GossiTheDog It got confused with https://clownstrike.lol/ ?
@GossiTheDog Even the vibe coding tools are calling them Clownstrike! LOL!!!
Maxilo Overit@GossiTheDog Vibe as a service as valueless as vibe as a substitute for a clue. News flash?
Julian Wolf@GossiTheDog Nice, Defender for free. CIO will love it, everybody happy.
Ivor Hewitt
Brett Haines@GossiTheDog Amazing! I can get the quality of a teenager that's barely trying for the cost of a junior salesperson. The future is truly here.
Jonas Köritz@GossiTheDog it forgot to include the recovery costs when your crowdstrike agent breaks your whole fleet at once.
@GossiTheDog the c-suite value of this can't be understated...
Luigi 
@GossiTheDog
I was vibing with Gemini on Google spread sheets from our vulnerability tools saner from secpod, what a joke how was messing results
I was vibing with Gemini on Google spread sheets from our vulnerability tools saner from secpod, what a joke how was messing results
babble encat@GossiTheDog I love that the ring pie chart has a little gap in it. All that money and their graphics library can't handle an edge case?
the vessel of morganna@GossiTheDog gonna try this out when I log in today, smh