all 60 comments

sorted by: hot top controversial new old
[–] 157 points 4 months ago (22 children)

And they achieved... what exactly?

More faces per book? Instant Instagram? Whatsapp is now appier? Finally copied Second Life in meta horizons?

Like... What does $221,000,000/month of AI even get you?

  • source
  • hideshow 22 child comments
  • [–] 23 points 4 months ago (1 child)

    And they achieved… what exactly?

    I sometimes wonder whether AI is just a way to launder money at the same time as burning the planet.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 3 points 4 months ago

    The big push happened right after the crypto crash. I have a suspicion both are pushed by energy companies. If power consumption goes down, we start switching to renewables and stop consuming dirty sources. They can't have that, so they pushed crypto to consume it. They let it crash because they started pushing "AI" instead. Politicians always say we can't decrease power consumption, and we just need to build up to replace dirty energy sometimes in the non-specified future. We will never reach that if they can instead keep creating excuses to expand demand.

  • source
  • parent
  • [–] 8 points 4 months ago (2 children)

    Right? I haven't noticed anything change, everything is as buggy, slow, and difficult to use as it ever was.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 9 points 4 months ago (1 child)

    The dark patterns on meta products are downright evil, and infuriating. How anyone can stay on their products baffles me. They punish you for using it.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 3 points 4 months ago

    Only decent way to see local events and updates on stuff happening in my area, also most people use marketplace so it's how I find stuff to buy locally.

    Facebook with ublock and a userscript to clean all the crap off it is at least generally usable.

  • source
  • parent
  • The app is falling apart. From marketplace discussions the link back to the listing is mostly broken. The app crashes frequently. Images and video will glitch back and forth by several pixels, randomly.

    I think they are doing a massive study on how automating code production kills products. It is a brave move and I look forward to the papers they publish when the company folds.

  • source
  • parent
  • [–] 5 points 4 months ago (3 children)

    May have proved senior engineers are more effective having a team of agents rather than a team of junior engineers

    Doubt that is the case, but that is likely what all these middle managers expect in these companies

  • source
  • parent
  • hideshow 3 child comments
  • [–] 7 points 4 months ago (2 children)

    An agent can produce code quickly, but at the accuracy level of a co-op unless a significant amount of time is spent to "prepare" the agent. Trouble is, you have to know how to do the job you're asking the agent to do to be able to effectively steer it to an acceptable solution.

    With a co-op (or junior engineer), I still need to do the equivalent of the above to train them on the tasks/concepts they need to perform their work, I still need to have a firm grasp of what it is that I'm asking them to do to be able to guide them to the correct solution.

    The difference is that the co-op/junior retains that knowledge and is able to extrapolate that to different situations as well as come up with innovative solutions based on that initial knowledge. One of the biggest risks to tech work from LLMs (aside from the general problems with LLMs) is it encourages managers to not hire juniors/co-ops, which means that there will be no new generation of senior developers coming up to replace us when we retire (lol, as if).

  • source
  • parent
  • hideshow 2 child comments
  • [–] 2 points 4 months ago* (1 child)

    I want to push this outside engineering. It sounds like they want to train computers, not employees. Which, I don't know about your fields, but folk aren't exactly looking to mentor their future competition in the financial industry. So you pretty much get to teach yourself or go back to school.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 3 points 4 months ago

    Historically (at least in orgs that aren't maximally toxic) software engineering has actually had a great culture of seniors helping Junior and midlevels grow. IME most of us enjoy mentoring, and further enjoy having someone else who can do the boring shit for us. It's why, despite incompetent management being the norm, the industry has succeeded as much as it has.

    Unfortunately I think this is changing now. The ownership class has finally pushed too far and I think a sizeable portion of us have realized that we devs have to stop maintaining the engineering cultures of our orgs to fight back.

  • source
  • parent
  • [–] 4 points 4 months ago

    Management has proven to the board that they're using the technology everybody in corporate likes. They'll surely get a bonus for that. And once it's obvious it fails they'll lay off thousands to compensate for the billions spent and get a bonus for the thorough cost cutting. Then hire a bunch offshore to get work done.

  • source
  • parent
  • [–] 2 points 4 months ago* (last edited 4 months ago)

    I would imagine the answer Meta execs would want, would be “effectively microtargeted advertising to those most likely to buy the product, creating sales for our advertising clients and allowing us to charge them more”

    Going through massive sets of user data to figure out which of them would most likely be interested in a product does sound like a plausible use case for AI

  • source
  • parent
  • [–] 51 points 4 months ago (2 children)

    Is that cost in electricity cost? Or they are not using their own AI models?

  • source
  • hideshow 2 child comments
  • [–] 11 points 4 months ago

    They have their own models, it depends on a teams manager and a few other constraints if they are required to use them or if they are pestered to use generative models at all.

    I'm betting that cost is just the cost for the token usage and nothing else.

  • source
  • parent
  • [–] 1 point 4 months ago

    Meta is not using their own models internally because their internal models are far behind the state of the art.

    This is why Zuck is in a frenzy too, trying desperately to catch up, failing consistently to actually do so

  • source
  • parent
  • [–] 48 points 4 months ago (7 children)

    im required to use ai at my job. they don't pay attention to what i do with it. so it now answers my emails 1 business day after i receive them, and responds (appropriately) that im working on their request and ill check back in with them with progress. its a constant loop.

    its funny how many tasks in corporate america get forgotten bc they weren't executed immediately :D

  • source
  • hideshow 7 child comments
  • [–] 7 points 4 months ago (4 children)

    corporate elsewhere too. it's all for The Metrics, to see usage and adaptation go up. i'm not required to use it at work, but our developers are.

    let's just say clients are increasingly choosing the (worse) fallback browser version of our service instead of the vibecoded and more and more broken app full of vulnerabilities. because at least the browser side has issues they expect and usually know by now.

    but hey, new graph go up, right? ignore the old graph we cared about before.

  • source
  • parent
  • hideshow 4 child comments
  • [–] 7 points 4 months ago (3 children)

    my fav is the fact that our dev team is use 80% of the companys ai tokens, but have to push to production at least four times bc of bug fixes. every week. we used to publish once, maybe twice.

    somehow this is better?

  • source
  • parent
  • hideshow 3 child comments
  • [–] 28 points 4 months ago (2 children)

    Surprisingly difficult to find a source on this that isn't a Twitter Screenshot of some right-wing grifter's account.

    https://www.theinformation.com/articles/tokenminimizing-meta-moves-curb-employee-ai-usage-ai-costs-reach-billions

    This seems to be it, but I don't think there's currently a working paywall bypass for the information (solid site, though).

    https://mlq.ai/news/meta-caps-internal-ai-token-spending-after-costs-approach-billions-in-2026/

    This might be that article, but not entirely sure. Ironically, that site seems to be another questionable AI project.

  • source
  • hideshow 2 child comments
  • [–] 1 point 4 months ago

    We've been seeing posts of people using AI for stupid stuff because their bosses wanted to show their bosses that they were all in on AI.

    Now that they've effectively churned the waters for effect, the companies have realized that - Surprise! - there was no result. Now they want AI used for real stuff, which is what they should have been doing in the first place. Companies got way out over their skis way too early on this one.

    And the AI companies made billions, and they're still broke, and years away from profitability, if ever. And viability, if ever.

  • source
  • parent
  • [+] 42 points 4 months ago (4 children)
  • [–] 16 points 4 months ago (1 child)

    I'd like to join the chorus of people saying "Zuckerberg will be bankrupt soon", but I said the same thing about Musk and he became a trillionaire

    People will continue to use his dumb products, not in the least cause he does his best to monopolize any market he's in

  • source
  • hideshow 1 child comment
  • [–] 7 points 4 months ago

    The system prevents people this rich from going down. He could literally toss a coin on every decision and at the end of the year he will be richer than before. These people basically cannot stop themselves from getting richer.

  • source
  • parent
  • [–] 12 points 4 months ago (5 children)

    Is this whole token thing really just the same as paid in-game currencies, just designed to get you to not think about the fact that you’re spending money? Or is there another use for it?

  • source
  • hideshow 5 child comments
  • [–] 17 points 4 months ago (2 children)

    Tokens are an AI context of units of text. Could be a character or a word. Why use the term tokens instead of words? Who knows. Maybe because some units aren’t words? The tech industry delights in creating stupid terms for things that already had terms.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 4 points 4 months ago

    There are to many words to have each be it's own input, but bundling frequently used snippets is more efficient than using characters.
    Then the investors came and decided to measure success by wasted energy and the rest is history.

  • source
  • parent
  • [–] 12 points 4 months ago

    and that is not even the actual cost, because the vendors are still not making a profit.

  • source
  • [–] 11 points 4 months ago (1 child)

    Observe the dead-eyed stare into the middle distance.

    He almost looks human in that photo.

  • source
  • hideshow 1 child comment
  • [–] 7 points 4 months ago

    And the marketplace search is still a broken pile of shit...wish everyone didn't jump to it and stayed on CL...

  • source
  • [–] 12 points 4 months ago (1 child)

    I hope Cuckerberg is bankrupt at the end of this AI clusterfuck.

  • source
  • hideshow 1 child comment
  • [–] 5 points 4 months ago

    I'm sure making Instagram clog my feed with suggested posts and other shit I don't follow has done wonders for their engagement metrics. I really love how meme posts and other shares are now clogging that shit up as well.

    It's definitely not a main contributor to what made me finally delete my Facebook. Kind of ironic that, given how much of my data they scrape, they still couldn't piece that together.

  • source
  • [–] 1 point 4 months ago

    Unfortunatly in the heart of empire they pick the winners and losers ..... news flash the losers become winners.

  • source
  • [–] 1 point 4 months ago

    Still less than what Uber spent

  • source
  • [–] 1 point 4 months ago

    There are more than 7.000 AI apps out there, most used are all from big corps (Gargle, M$, Zuckerbot, etc), meanwhile there are pretty good and private indie alternatives which are struggeling to survive. That is the real problem, not the AI as tool as such, but those which biased these for own greedy advantages, to profile and manipulate the user. Adding users which use it without criterio and knowledge to substitute with it own research and brain, instead of an tool which can help with research and work, using it for what it is.

  • source