Microsoft overhired due to the spike in global usage during the pandemic.
Some jobs are being automated but it’s more a question of "How many." And Microsoft was one of the first to start dumping its cash reserves into data centers. There is a good chance these will remain profitable even after any sort of bubble pop because they still remain a hyperscaler that can provide services beyond just AI compute.
overall, given where the economy is at and how much of their war chest is gone, im not surprised to see this happening.
If you look at Microsoft's total headcount over the past decade, it's really not that outlandish, especially if you combine "over hired during the pandemic" with "end of the ZIRP era". In 2019 employee count was 144k, now it's about 223k.
That said, for the past 3 decades at least, tech was a hugely growing sector that provided ever growing employment opportunities. It seems like that time is now over. Even if you take the "Jevon's paradox view" of AI, that is that the cheaper cost of writing software will feed a demand explosion, it's just hard to imagine that many more people will be needed to fulfill that extra demand.
This is why I think it's fair to think "this time it really is different". In past technological revolutions, it really wasn't that hard to see where the new jobs would come from. Now even the leaders of the AI revolution think it will kill tons of jobs. I get a bit worried about the number of young people and friends' kids who decide to major in "business" in college now. It just feels very difficult to imagine a white collar career that will be stable over the next 10 years or so, let alone an individual's 40 year career.
You can't just look at the number from 7 years ago and compare it to today and say "see pandemic over hiring" look at it with more resolution. Microsoft indeed increased hiring during the pandemic, but they laid people off and then rehired more than they laid off. Since the 2022 layoffs tech companies have consistently increased staffing levels.
This isn't pandemic over hiring, its layoffs to create fear and give them an excuse. I guarante you, and I'll put money on this. No matter how many people they let go with this, in six months they'll be employing even more than they do today.
Google has only about 2% more employees than 2022. Management decided that's the biggest the company is going to get.
It impacts absolutely everything. The dynamism from a growing org is gone. Managers have very slim opportunity for promotion. If you need people, you have to find them from another team. I'm relatively new on my team with 5 years there.
> In past technological revolutions, it really wasn't that hard to see where the new jobs would come from.
I think this is an extremely pollyannaish view of how technological revolutions have happened. In fact, it is always the exact opposite, people always fail to predict where new jobs would come from.
I'll just speak for myself then. The "PC revolution" launched in the 80s. A primary goal of the PC (from a business perspective) was to allow individual workers to do much more by themselves, e.g. in the 60s and 70s you had legions of secretaries and assistants for basically every mid level or higher worker.
It was completely obvious in the 80s, and I say this because I lived it, that you would need tons of people building software and in tech generally as PCs and related tech killed off a slew of those admin assistant jobs.
I think it was obvious that more programmers would be necessary, but if you look at BLS forecasts from 1985, they expected +~250k programmers/computer analysts each over 10 years, which is basically what happened. Secretarial employment was 3m at the time and predicted to stay basically flat, which is basically what happened.
Typists and data entry work cratered, but overall admin support actually went up.
There are hundreds of millions of workers in the US, and about 5 million computing professionals right now. There are 3.5m workers in admin assistant roles.
The growth in computing employment is totally overshadowed by parts of the workforce. BLS' "Management, business and finance" grouping grew by 19m since the early 80s.
Nobody knows what the equivalent to this is going to be until the economy starts reorganizing itself.
I'm careful to not make any assumptions here - as it might very well "be different this time" - but by the same logic, you could say it's totally obvious that we will need tons of people to build and diffuse AI systems and workflows into large orgs and SMEs. I have so many friends (in traditional industries) who could easily 5x their efficiency with AI but they have no idea and they need help. I don't see this changing nearly as quickly as AI becoming more capable.
I mean, in the very short term I could buy that scenario - there are a host of "AI consultants" selling their wares.
But even then, I would only buy that for the very short term, because AI is designed to (eventually) implement that step, too. As an analogy, I always wonder who is paying for these "how to use agentic AI" courses I see pop up all the time now. Literally just ask it! I asked Claude to build a course for me to get comfortable understanding agents and harnesses and it did a fantastic job.
Ive been motivated recently to start making my appliances mine by using Ai to reverse engineer and replace their firmware.
So long as the appliance has been slow to update its security channels, the recent eMudrha certificate leak was the perfect opportunity to place yourself in the middle of the relationship between device and vendor. Its allowed me to make some trivial updates to my television's firmware and start to have AI build me firmware that better serves my purposes. Hoping to lobotomize it and replace everything with a simple linux system that executes a startup script to launch a Jellyfin client.
I bought it and its mine, this is the freedom we were originally robbed of.
The idea was always that Facebook or Google were listening to us and whenever we would talk about something we would shortly thereafter see an ad related to it.
Now we can basically say that its not Facebook and Google doing this directly, its our televisions and appliances that are listening and just selling that data back to the ad-networks.
This is similar to Flock and Palantir who collect info the government is not allowed to and then sell it to the government.
In this case Meta, Google etc can honestly say they aren't surrepticiously listinging to you or collecting data not in their privacy statement. They just buy it instead.
So I need to write a chaffing agent that runs when I am not home and randomly spouts synthetic conversations through non-smart speakers about nappies (kids a re long gone), nuclear weapons development, plumbing, union membership, crochet, video games, crypto coins, and other things we don't normally do.
Facebook was always doing this. It just wasn't proven they were.
VoIP has been around for 25 years. Compressed audio is a tiny amount of packets. It doesn't need HD audio, it just needs enough to recognize words. Enterprise VoIP only needed a 128kb connection 25 years ago to work. On a smartphone, 128kb is nothing and will blend in with background connections. Facebook could record and monitor your audio 24/7 and it won't even be a blip in the network traffic, especially if it does a quick process of the audio before sending it to Facebook.
Has this not been proven by now? That smart devices are used to do this, including smartphones? Their OSes just ship with apps that that record that audio information right out of the box.
> Now we can basically say that its not Facebook and Google doing this directly
Finding out that TVs are listening does not mean Google and Facebook aren't. I've had plenty of times in an old car, or on a walk talking about something only to have related ads pop up later.
They can already do that with most of the information they already have, they don't need to listen for that type of thing (though they still might be as you see here, data is data after all).
Most people are not that unique. They know where you are and who you are with and any searching or sites you have been to and what is in your text messages or whatsapp as well as that same information for all the people 6 degrees away from you. They know people like you, who are your age and like the things you like and are where you are with friends like yours etc etc it goes on and on.
People rarely talk about things unprompted and you will only notice when you think you talked about something and then see an ad for it so there is a confirmation bias as well.
That's a pretty good point. We kept hearing these rumors for a while now and all the usual suspects keep saying "we don't listen to your microphone on your phone" but yeah nobody expects the TV and fridge to do it, so maybe that's how it happens and people are not crazy.
That's just weasel language. For "hey, Google" to work, the phone needs to listen. Just like it can react to "hey, Google", it can also react to any amount of other trigger words without ever actually recording your conversation.
DOGE was impressively criminal. The team tried to remove the CVE reporting system entirely and only failed to do so because of intevention from the military and several agencies.
1) the government is currently doing the work and the work is critical. Yoinking the system without any replacement causes catastrophic consequences that reduce profits (and maybe get a few folks killed, but really, think of the bottom line.) Anyways, if you want to transition a critical component you do so with a plan.
2) It is critically necessary for the CVE system to be trusted by many entities worldwide. The government has historically not played favorites and had that trust. This is still mostly true. No for-profit commercial entity has that same level of trust; the for-profit model removes the ability for that trust to truly be universal.
That is straight up not what happened, and what’s impressive is that this comment is already being cited in AI summaries about it.
The contract with MITRE was set to expire under budget cuts from DOGE, along with basically everything else. That is not them “removing the CVE reporting system” like it was done specifically. It was not saved by ‘several’ agencies or the military (???), CISA restored funding last minute.
What's yours? I'm not defending it, that's just what occurred. The parent comment makes it seem like DOGE cuts were aimed at the CVE program specifically when they were not.
Ai labs look to be desperate to try and stave off the cooling hype that looks to be snowballing in the industry right now. These next few months are going to be crucial to their potential survival and future.
Its not surprising to me that this press release looks half-baked.
After AI more or less “solved” large parts of mathematics overnight and literally broke the internet by uncovering decades-old PHP vulnerabilities, you still think this is a hype that’s cooling off?
To me, this mostly shows how much our baseline for what counts as “noteworthy” has shifted since ChatGPT came out. I’ve seen more disruptive events in the last couple of years than in the 30 before that.
What would it take to get you excited again? AI discovering that aliens exist?
AI didn't solve large parts of mathematics. Which is not to say the the results aren't interesting or remarkable. Proving that Navier-Stokes blows up with certain external forces is important. But it's just a step, like seeing a pattern in data with no insight into what the data represents. Understanding why and where the fluid dynamics model is wrong is more interesting, if this enabled that then it's great.
All the proofs they've produced are interesting but they seem to be based on the work of mathematicians who nearly got there. As yet, no big leaps of insight. Again, I'm not saying it's a bad thing, it's also not a demonstration of intelligence beyond reckoning. I also read it cost something on the order of $15m of inference per problem, maybe a whole army of mathematicians could have got there too if that was available.
Isn’t that the point though - they’ve generated such strong results in mathematics, so why now are they delivering something so weak here?
Hype rather than realistic attempts to solve things is actually a valid explanation for that. They want to give the impression it’s a solve-everything machine even if it’s not well suited to areas outside of CS and math (or really anything where the logic of what is being done is not 1:1 with syntax, which is where LLMs are clearly excelling)
> So why now are they delivering something so weak here?
Was anyone under the impression that AI would, in its very first attempt at biological research, suddenly be able to get answers without having to need physical follow-through on its leads? Even if AI had a hypothetical cure for cancer, it would need a human with cancer and a biolab and governmental approval to actually do anything with it.
The nature of biology makes the feedback loops much larger and more expensive, which means the AI has to follow the "normal" collaborative scientific process instead of being able to hunker down in its proverbial study and work all night.
> Was anyone under the impression that AI would, in its very first attempt at biological research, suddenly be able to get answers without having to need physical follow-through on its leads?
Why can’t it build a simulator for this stuff? It has textual and formulaic explanations of everything it’s working with. Seems like it should be less than a days work if it can solve all of math in a few hours?
Any computational bottleneck on running such a simulator should also be solvable right, Anthropic have already told us this is the singularity already
AI boosters need to make up their mind: If you are claiming this is the Singularity and you have obsoleted human thought, expect to be judged by the standard you have set.
If you are saying you have made a really cool piece of enterprise security automation software, then okay I am impressed this is a very powerful fuzzer.
Has it really? I wonder because all these "breakthroughs" are published in AI magazines and not mathematics magazines. Seems to me a bunch of advertisement but I am not qualified to make a claim
The staggering losses anthropic is facing are mostly due to their valuation increasing to just shy of 1T at the beginning of this year. That valuation increase caused the value of what they owe on future stock to triple which is shown as a massive loss on paper but doesn’t represent actual cash leaving their hands. The rules are being rewritten by frontier companies whose revenue growth and proposed spending are outpacing any thing we have seen in the past. Anthropic has committed to spending over 500B over the next 10 years on cloud infrastructure which might be what you are quoting but many project their revenue growth to outpace the 50B a year in spending so that may not even equate to any loss but will certainly be less than 50B/yr. The whole thing is bonkers but and it’s easy to get overwhelmed by the breathtaking numbers on their balance sheet.
Paper loss. It's a sort of debt they have on bonds that is tied to their valuation. Because their valuation has grown significantly, so has their debt. It's standard accounting process to put that debt as a "loss", even though they're not paying anything.
Think of it this way. It's as if a public company gives you a certain amount in RSUs that won't vest for a while. A year later, the company is worth 10x more. So they list the increased value of the RSUs they owe you as a loss.
In reality, the fact that this "debt" has ballooned so much is a positive sign - the valuation is high (it's a separate issue whether that valuation is realistic).
If they would be profitable without further spending on R&D they would stop spending, but as you say they are in an R&D race so they can't stop and thus the hypothetical is irrelevant.
I think Anthropic actually has enough revenue to function. I think they cover their compute obligations with revenue, I think they can probably cover reasonable salaries too.
I made some calculations earlier, but with 16.8B for Q3, what is the problem? 2.5B per annum should cover the necessary salaries. After that they have 64.7B. Compute obligations over the coming ten years are only 518B. 8x their their current revenue.
I think this is actually a stable business. Probably not compatible with a 2T valuation though, especially not at present interest rates.
I don't see why its such a difficult idea to square up that AI companies attract high tier mathematicians and many people just genuinely want to do research in math and natural sciences with the bonus that it also shows off the model's capabilities?
That must be hard to say with a straight face. Over the last 4-6 months I think very few corporate developers have handwritten any code at all and the “intelligence” growth from one model to the next is very tangible.
Odd. I barely perceive the growth, and I use them a ton at work. Not saying the advancements aren’t there, I just wouldn’t describe every model release as having tangible growth.
Also the handwritten code doesn’t remove the work. If anything, it makes the thinking more difficult (for me) because instead of thinking during the coding, I have to before and after (in iterations).
For big projects I actually go back to tradcoding some to make sure I understand the system, since I’m the one responsible.
I actually dont let llms code anything non trivial, I will have it help with the planning a bit usually, but I write most of the code, and wow turns out my brain doesnt rot in the process and I actually understand whats going on.
Im convinced the people letting llms do their jobs are going to get replaced, not by agents but people who use them more reasonably.
Everyone wants to be the one with the power to remake the world according to their own designs.
Everyone also doesn't want to be the one being remade according to the designs of others.
The AMD hardware is x86_64, which means lower battery life on average for normal users, and as an architecture, it is very well supported across the Windows ecosystem.
The RTX Spark is ARM64, which means more efficient power management at the cost of a small hit to sustained performance loads. But comes at the cost of being less supported when it comes to available software packages.
Aside from this, I’m led to believe that the AMD stack when it comes to local models has more overhead related to its need to translate more NVIDIA native instructions and therefore can be less performant than Nvidia.
Not sure how true this is anymore, but ive yet to hear otherwise.
> The AMD hardware is x86_64, which means lower battery life on average for normal users [...]
A Ryzen AI Max+ 395 powered tablet with 70Wh [0] gets the same battery life given the same task as a Nvidia N1X powered laptop with 99.9 Wh [1].
Power Consumption in their "Cyberpunk / Stress Test" is also lower at Ø109.3 (108.8-110.8) Watt for the x86_64 vs Ø134.5 (132.1-137.5) Watt for the arm64 system using the same settings, resolution and display brightness. They also get the same performance when normalising for resolutions and settings in Cyberpunk, so this is not an instance of one providing significantly more performance.
To add, battery life on this chip hasn’t been that impressive [1]. It’s basically matching recent MacBook Pros, while sporting a significantly larger battery. Both AMD and Intel have made significant strides in improving power consumption on their latest core designs, to the point where there isn’t a lot of difference anymore.
> Aside from this, I’m led to believe that the AMD stack when it comes to local models has more overhead related to its need to translate more NVIDIA native instructions and therefore can be less performant than Nvidia. Not sure how true this is anymore, but ive yet to hear otherwise.
Uh, well you'd just download lemonade server and use models designed explicitly for the hardware from hugging face.
Windows Pro > Work or School Account > Join a Domain instead > Set password & recovery questions
The one component here that is not present on most machines straight from the manufacturer is the presence of the Pro version.
The home version of windows requires the account with increasingly limited alternatives.
As the author mentions in TFA, she's on the Arch staff as a package maintainer. I don't think there was a technical obstacle around the windows setup process.
I wonder. I sort of accidentally discovered that Rufus could edit my arm64 windows 11 install media without problem - adding a local account and bypassing the need for domain login (just as it does for amd64).
This isn't a new trick - just the regular approach to editing windows install media.
But that process still results in having Windows 11 installed on the computer. I don't think it's a stretch to assume someone on the Arch Linux staff has reasons for avoiding installing Windows which do not involve an inability to accomplish it.
In my experience you're just replicating the problem.
Slop code doesn't stop being slop code when you add in slop tests to account for the sloppiness.
If you generate code at a certain velocity, you eventually get to a point where you generate tests the same way you generate code. The model does its best effort in creating tests, but you'll inevitably get to a point in your work where the tests themselves are faulty and unreliable, the same sloppiness you were trying to control for.
There have been multiple times where my software dev team has gone back to prior implemented tests they had the model generate and found that after some time eventually the model got frustrated with the test no longer passing eventually modifying it to fit current needs around a feature push.
You can probably anticipate how we came to discover this, as it ended up introducing a regression around multiple edge cases which the test was meant to control for.
The only answer is to inject actual engineering into the process as much as possible without giving it all up to the models to generate more slop in response to slop.
tests are a great answer, but only if you treat them with serious intentionality.
Your engineers weren't doing what I'm describing I promise.
There are multiple places in my post where I explicitly describe the process by which you intimately:
- make the agent do verified work (the e2e tests and test driven thing)
- watch the agent work (the second monitor thing)
- read your own code (the tests/prod mirror thing
- play with the work and perfect it (the tech debt and UI perfect thing)
Plus a ton of other places in my post I very clearly describe being intimately involved in building and selling the work.
That to me is basic god damn competency. What your engineers did is a fad, and it won't last, and I'm sorry they abdicated their professional duty.
If you wish to screen for that sort of shit in future/current hires, show them my post and see if they turn up their noses or if they get excited and creative. You want only the latter.