Microsoft reports are exposing AI's real cost problem: Using the tech is more expensive than paying human employees
Microsoft reports are exposing AI’s real cost problem: Using the tech is more expensive than paying human employees
Companies are racing to incentivize employees to use AI. But as some companies are finding, the more employees that use the technology, the heavier the bill.Jake Angelo (Fortune)
toiletobserver
in reply to ☆ Yσɠƚԋσʂ ☆ • • •pimat
in reply to toiletobserver • • •RobotToaster
in reply to toiletobserver • • •VoodooAardvark
in reply to ☆ Yσɠƚԋσʂ ☆ • • •FoundFootFootage78
in reply to ☆ Yσɠƚԋσʂ ☆ • • •There are ways to make it cheaper. Starting with maybe not encouraging token-maxing.
Generally, unless you're either a FOSS project or generating images/video, you have to be doing something very wrong to spend more on AI than on salaries.
tyler
in reply to FoundFootFootage78 • • •GamingChairModel
in reply to tyler • • •Yeah, the smarter way to use LLM-based agents is carefully defined tasks. Mozilla describes their vulnerability assessment processes in this blog post.
Mozilla describes the process they've used: building a harness that instructs a model to find a specific category of vulnerability on a specific interface, and then write up its findings. It's a narrow enough context that the model gets specific instructions, and a simple definition of success, and it sets up many such tasks that can be fed into the existing process for verifying and triaging bugs. Note that the output for this LLM pipeline basically feeds into the same interface for accepting bug reports from the public, or from their human contributors within the project.
There's a couple of takeaways here, too:
- This pipeline is model agnostic. Mozilla set it up before Mythos was released, and its description of other models (Opus 4.7, Codex) confirms that Mythos is better but not a
... Show more...Yeah, the smarter way to use LLM-based agents is carefully defined tasks. Mozilla describes their vulnerability assessment processes in this blog post.
Mozilla describes the process they've used: building a harness that instructs a model to find a specific category of vulnerability on a specific interface, and then write up its findings. It's a narrow enough context that the model gets specific instructions, and a simple definition of success, and it sets up many such tasks that can be fed into the existing process for verifying and triaging bugs. Note that the output for this LLM pipeline basically feeds into the same interface for accepting bug reports from the public, or from their human contributors within the project.
There's a couple of takeaways here, too:
There are ways to use these tools, but none of it really seems like a truly revolutionary/disruptive change to how large projects are managed.
Behind the Scenes Hardening Firefox with Claude Mythos Preview - Mozilla Hacks - the Web developer blog
Brian Grinstead (Mozilla Hacks - the Web developer blog)NoiseColor
in reply to FoundFootFootage78 • • •GamingChairModel
in reply to NoiseColor • • •Che Banana
in reply to ☆ Yσɠƚԋσʂ ☆ • • •Hahahahahhhahahahahaaha
g a s p
HhahahhahahahbahahhHahahahahahah
akakevbot
in reply to ☆ Yσɠƚԋσʂ ☆ • • •I don't feel like the companies are concerned about it costing more right now as much as they are betting that it will be cheaper in the long run. The cost of labor isn't unlikely to decrease drastically while the technology is likely to become cheaper.
While I would love to believe Microsoft is being burned by spending on AI, I think they don't mind spending more now so long as they can trade the cost labor for the costs of technology and maintain similar productivity.
Feels like they hope this will be to white collar jobs what Uber was for taxi drivers. Current profitability isn't really the goal as much as being able to reproduce similar outputs.
RobotToaster
in reply to akakevbot • • •This is my take too.
People were probably making the same jokes in the early mainframe era of computers.
Ismay
in reply to RobotToaster • • •They did not. Computers were able to do things that were almost impossible to do by hands.
LLMs don't do that. They regurgitate what we've been doing for decades.
They're great to brute force some problem tho. You give them shitloads of data and they're incredible to go through it.
Problem is, we took a REALLY unfinished tech and tried to package it as the saviour .
4am
in reply to Ismay • • •The “ AI as a savior” was only ever a marketing ploy to encourage enterprise to fall entirely and foolishly into the arms of SaaS, which then forces the consumer market to follow suit.
The fascists have built their panopticon.
Zephorah
in reply to akakevbot • • •arin
in reply to akakevbot • • •lemonwood
in reply to akakevbot • • •Even if AI someday does become profitable short term, total profits will still go down long-term, because all profit comes from human labor (or exploiting nature). All any technology ever does in capitalism is to replace human labor, thereby putting more pressure on the empirically proven tendency of profits to fall. Profit gains from technology can only ever be short term and relative to competition who hasn't yet adapted the technology. Once everyone has, prices drop, adjusting to lower socially necessary labor time.
The only way for the billionaire class to keep profits flowing a bit longer at this point is to do what we already see them doing now: get rid of the free market by enforcing monopolies with captive markets, bonded labor and merging big capital with an increasingly violent and warring state apparatus: capitalism inevitably leads to fascism-imperialism every time.
AI lends itself to this because of the ce
... Show more...Even if AI someday does become profitable short term, total profits will still go down long-term, because all profit comes from human labor (or exploiting nature). All any technology ever does in capitalism is to replace human labor, thereby putting more pressure on the empirically proven tendency of profits to fall. Profit gains from technology can only ever be short term and relative to competition who hasn't yet adapted the technology. Once everyone has, prices drop, adjusting to lower socially necessary labor time.
The only way for the billionaire class to keep profits flowing a bit longer at this point is to do what we already see them doing now: get rid of the free market by enforcing monopolies with captive markets, bonded labor and merging big capital with an increasingly violent and warring state apparatus: capitalism inevitably leads to fascism-imperialism every time.
AI lends itself to this because of the centralized nature of data centers, the already monopoly based business model of tech companies and the political power and influence those monopolies hold. On the other hand, there is some potential, if not revolutionary at least disruptive potential, in small scale, specialized, open source models that can be trained with fewer resources.
The only alternative road to fascism, of course, leads to communism.
A world rate of profit: important new evidence
Michael Roberts Blogmakeshift0546
in reply to lemonwood • • •Even if you take worst case costs Anthropic's "Profitability" Swindle share.google/UV5HNgJyMzfcknekF it's already approaching profitable.
If you slow down the model update cycle it's looking like at least anthropic can be profitable 🤷♂️. That argument is loosing it's weight quickly.
Anthropic's "Profitability" Swindle
Ed Zitron (Ed Zitron's Where's Your Ed At)4am
in reply to akakevbot • • •It’s a bad bet. AI is unprofitable now because they are building the datacenters.‘do you think one day those are going to be done and just never touched again? No, they’re going to constantly upgraded. They’re an ongoing and forever cost for upkeep, a fraction of- but a significant one - of their initial cost.
The AI companies, however, are never going to lower prices. They’re going to raise them until some of the market gets uncomfortable paying them. Then, they’re going to hold them there.
Will it be cheaper in the long run? No. Not really.
But it was never about replacing workers anyways. It was about fooling enough C-levels and middle managers and even “normies” that Super Amazing Smart Bot can do your job while YOU live in comfort! Incredible! Hey kids, tried of thinking? Let the agents do it FOR you!
And once enough were on board, and the “demand” was so great; oopsie they bought up all the compute on planet earth. All information flows through them now. Every company, every home, ev
... Show more...It’s a bad bet. AI is unprofitable now because they are building the datacenters.‘do you think one day those are going to be done and just never touched again? No, they’re going to constantly upgraded. They’re an ongoing and forever cost for upkeep, a fraction of- but a significant one - of their initial cost.
The AI companies, however, are never going to lower prices. They’re going to raise them until some of the market gets uncomfortable paying them. Then, they’re going to hold them there.
Will it be cheaper in the long run? No. Not really.
But it was never about replacing workers anyways. It was about fooling enough C-levels and middle managers and even “normies” that Super Amazing Smart Bot can do your job while YOU live in comfort! Incredible! Hey kids, tried of thinking? Let the agents do it FOR you!
And once enough were on board, and the “demand” was so great; oopsie they bought up all the compute on planet earth. All information flows through them now. Every company, every home, every document, every email, every PowerPoint presentation, every lewd text. Indexed, summarized, graphed, reported in real-time.
We just lost personal computing, and it’s literally capitalism’s design goal.
GamingChairModel
in reply to 4am • • •I think AI will be profitable for the next generation of AI business models that emerge from the abandonment of the current business model of developing the frontier. But the prerequisite is that the companies give up on developing the frontier and decide that the models they have are good enough, then get hardware optimized for inference on those models, stagnating into long term commodity infrastructure, like providing phone service or electricity for profit.
So yeah, I think many of these technologies are here to stay, but the growth will stagnate this year as data center construction swallows up companies that overextended.
Zephorah
in reply to ☆ Yσɠƚԋσʂ ☆ • • •FreedomAdvocate
in reply to ☆ Yσɠƚԋσʂ ☆ • • •The article is based off a false premise right from the get go, and their first line even points this out.
This is just common sense. Why pay for Claude licenses when Claude’s models are in GitHub copilot, which they own and is integrated directly into their dev programs? It’s not scaling back AI.
And while the cost of ai compute can get very costly very quick, having talented developers using the AI tools enables them to get through mountains more work than they could without it, so it’s really paying for higher “productivity”/throughput.
makeshift0546
in reply to ☆ Yσɠƚԋσʂ ☆ • • •Meh, right now, and only if you're trying to replace the work force. At is current state, on a $30 a month Copilot plan you'll already see a huge gain in efficiency with supervised coding and agents doing minor chores and maintenance.
The average coder isn't better then opus 4.6. No, is not ready to run production code bases unsupervised. Yes, it's absolutely ready to do many many simple tasks autonomous and more complex coding with supervision.
If you take even a week to try out this shit with an eye for what's possible currently and have an ounce of common sense, I fail to see how folks don't realize this will absolutely change how software is delivered. Yes humans will be involved but there will be much much less direct coding and a lot more supervision over multiple concurrent tasks that have had the time to delivery cut significantly.
☆ Yσɠƚԋσʂ ☆
in reply to makeshift0546 • • •I use these tools extensively, and they absolutely do not replace the need for a coder. The reality is that they're fundamentally incapable of telling whether something is correct or not in the business sense. And Simply churning out a ton of wrong code really fast doesn't actually help anybody.
They certainly can be a help for a developer. For example, I can fluently write code in any language now even if I'm not familiar with the stack or syntax. A skill that would've taken months of effort to build previously. But in terms of actual workflow, it's not all that much faster because I still have to review what the tool is doing, and human comprehension is still the bottleneck in the whole process.
Etterra
in reply to ☆ Yσɠƚԋσʂ ☆ • • •unbuckled_easily933
in reply to Etterra • • •GitHub - dtnewman/burn-baby-burn: Be the most productive engineer on your team 🚀🚀🚀
GitHubTheoriginalthon
in reply to unbuckled_easily933 • • •License
MIT (but it's all vibe-coded so who really knows who owns it?). Burn responsibly. 🔥
That's hilarious if true
P03 Locke
in reply to ☆ Yσɠƚԋσʂ ☆ • • •Sure, if they are using Copilot. Copilot is fucking garbage.
If they made a model that was half as good as Claude or Kimi or Qwen, they wouldn't be in this mess.