Comment by tracker1

Comment by tracker1 3 days ago

37 replies

View on Hacker News

I'm somewhere in the middle on this, with regards to the ROI... this isn't the kind of thing where you see immediate reflection on quarterly returns... it's the kind of thing where if you don't hedge some bets, you're likely to completely die out from a generational shift.

Facebook's product is eyeballs... they're being usurped on all sides between TikTok, X and BlueSky in terms of daily/regular users... They're competing with Google, X, MS, OpenAI and others in terms of AI interactions. While there's a lot of value in being the option for communication between friends and family, and the groups on FB don't have a great alternative, the entire market can shift greatly depending on AI research.

I look at some of the (I think it was OpenAI) in generated terrain/interaction and can't help but think that's a natural coupling to FB/Meta's investments in their VR headsets. They could potentially completely lose on a platform they largely pioneered. They could wind up like Blackberry if they aren't ready to adapt.

By contrast, Apple's lack of appropriate AI spending should be very concerning to any investors... Google's assistant is already quite a bit better than Siri and the gap is only getting wider. Apple is woefully under-invested, and the accountants running the ship don't even seem to realize it.

daheza 3 days ago

I think apple is fine. When AI works without 1 in 5 hallucinations then it can be added to their product. Showing up late with features that exists elsewhere but are polished in apple presentation is the way.

Reply View 14 replies
  • rafaelmn 3 days ago

    Have you used Siri recently ? It's actually amazing how it can be crap at tasks consistently considering underlying tech. 1 in 5 hallucinations would be a welcome improvement.

    Using ChatGPT voice mode and Siri makes Siri feel like a legacy product.

    Reply View 4 replies
    • rTX5CMRXIfFG 3 days ago

      I don’t think that’s the point. Yes, Siri is crap, but Apple is already working on integrating LLMs at the OS level and those are shipping soon. It’s a quick fix to catch up in the AI game, but considering their track record, they’re likely to eventually retire third party partnerships and vertically integrate with their own models in the future. The incentive is there—doing so will only boost their stock price.

      Reply View 3 replies
      • bbarnett 2 days ago

        Wait, Apple is going to introduce an AI Clippy?!?!

        Reply View 0 replies
      • mrheosuper 2 days ago

        Just because the LLM can access kernel does not miracle make it better. Do you think the problem with Apple AI right now is because they don't have access to OS components?

        Reply View 1 reply
        • Imustaskforhelp 2 days ago

          There was this project that recently dropped open source https://github.com/minitap-ai/mobile-use

          I do think that kernel level access is not needed as this is some good amount of automation though I assume what apple can make do however is actually not require another laptop connected to automate your mobile but rather the npu?/gpu? inside your phone.

          I am surprised by why they haven't done it already.

          Reply View 0 replies
  • makeitdouble 3 days ago

    In general I don't think Google or Apple need AI.

    In practice though their platform is closed to any other assistant than theirs, so they have to come up with a competent service (basically Ben Thomson's "strategy tax" playing in full)

    That question will be moot the day Apple allows other companies to ingest everything's happening on device and operate the whole device in reaction to user's requests, and some company actually does a decent job at it.

    Today Google is doing a decent job and Apple isn't.

    Reply View 3 replies
    • platevoltage 2 days ago

      You’re right, they don’t need AI. I finally stopped using Google search after they added the AI summary and didn’t add a way to turn it off. I’m just as bothered by Apple’s lack of AI as a am their lack of a touch screen on MacBooks. I use AI when I need AI.

      Reply View 2 replies
      • Imustaskforhelp 2 days ago

        Hm, so what'd you want from apple and google.

        One went too far in one direction and the other went too fair in the opposite direction. And it seems that you want to be somewhere in the middle?

        Reply View 1 reply
        • makeitdouble 2 days ago

          I think in general we want granularity and choice.

          So not just in how much AI there is, but what AI, where it's applied and where we can turn it off, and what context it has access to and where it's off bound.

          Reply View 0 replies
  • platevoltage 2 days ago

    True. I mean, how long did it take us to get a right click button.

    Reply View 2 replies
    • tracker1 2 days ago

      If you're talking apple... if you used a "standard" 2-3 button mouse, It worked in OSX from pretty much the start iirc. I always used a PC mouse for that reason.

      Reply View 1 reply
      • platevoltage 2 days ago

        oh absolutely. They have had support for aftermarket mice for a while. Their track pads have supported "right click" for a long time too.

        Then again, they've always been way better at making track pads than mice. They have probably the best track pad in the business, and also the Magic Mouse, which everyone hates.

        Reply View 0 replies
  • vivzkestrel 2 days ago

    or maybe Apple realizes the whole thing ll crash in 18 months and are waiting for the fallout

    Reply View 1 reply
    • Imustaskforhelp 2 days ago

      the AI technology might be nice imo but its nowhere near the amount of money being spent. Its dumpster fire amounts of money and the amount of weirdness just everything being AI wrapper slop is so.. offputting.

      Things can be good and they can still be a bubble just as how the internet was cool but the dot net bubble existed

      They become bubble when economically things stop making sense.

      AI ticks this checkbox.

      Reply View 0 replies
giancarlostoro 2 days ago

> they're being usurped on all sides

They did it to themselves. Facebook is not the same site I originally joined. People were allowed to people. Now I have to worry about the AI banning me.

Reply View 5 replies
  • platevoltage 2 days ago

    I deleted my Facebook account 10 years ago, and I’ve been off Instagram for half a year. I recently tried to create a new Facebook account so that could create a Meta Business account to use the WhatsApp API for my business. Insta-ban with no explanation. No recourse.

    Reply View 4 replies
    • Imustaskforhelp 2 days ago

      You don't have to use whatsapp api for business if all you want is simple automation.

      Like, there is beeper which can theoretically allow you the same but you might have to trust their cloud but they are giving options locally too

      in the meanwhile, you can use what beeper uses underneath its hood which is https://github.com/tulir/whatsmeow and use this for automation.

      I used this for some time and I didn't seem to get banned but do be careful. Maybe use it on a side sim. I am not sure but just trying to help

      Reply View 1 reply
stephen_g 2 days ago

> They're competing with Google, X, MS, OpenAI and others in terms of AI interactions

Am I the only one that find the attempt to jam AI interactions into Meta's products useless and that it only detracts from the product? Like there'll be posts with comedy things and then there are suggested 'Ask Meta AI' about things the comedy mentions with earnest questions - it's not only irrelevant but I guess it's kind of funny how random and stupid the questions are. The 'Comment summaries' are counter-productive because I want to have a chuckle reading what people posted, I literally don't care to have it summarised because I can just skim over a few in seconds - literally useless. It's the same thing with Gemini summaries in YouTube - I feel it actually detracts from the experience of watching the videos so I actively avoid them.

On what Apple is doing - I mean, literally nothing Apple Intelligence offers excites me, but at the same time nothing anybody else is doing with LLMs really does either... And I'm highly technical, general people are not actually that interested apart from students getting LLMs to write their homework for them...

It's all well and good to be excited about LLMs but plenty of these companies' customers just... aren't... If anything, Apple is playing the smart move here - let other spend (and lose) billions training the models and not making any real ROI, and they can license the best ones for whatever turns out to actually have commercial appeal when the dust settles and the models are totally commodified...

Reply View 2 replies
  • tracker1 2 days ago

    I was thinking about this.. if you look at (I think OpenAI) the scene generation and interaction demos, it's a pretty natural fit for their VR efforts. Not that I'm sold on VR social networks, but definitely room for VR/AR enhancements... and even then AI has a lot of opportunities, beyond just LLM integration into FB/Groups.

    Aside, groups is about the only halfway decent feature in FB, and they seem to be trying to make it worse. The old chat integration was great, then they remove it, and now you get these invasive messenger rooms instead.

    Reply View 0 replies
  • olyjohn 2 days ago

    God the AI answers that it gives in the Facebook Groups are so wrong that it's hilarious.

    Reply View 0 replies
malfist 3 days ago

How many years of not seeing returns this quarter does it take before its all hype?

Reply View 5 replies
  • tracker1 3 days ago

    How long did it take Space-X to catch a rocket with giant chopsticks?

    It's more than okay for a company with other sources of revenue to do research towards future advancement... it's not risking the downfall of the company.

    Reply View 0 replies
  • throw310822 3 days ago

    "technology too expensive to be offered at a profit (yet)" != hype

    Reply View 3 replies
    • ehnto 2 days ago

      The MIT report that has everyone talking was about 95% of companies not seeing return on investment in using AI, and that is with the VC subsidised pricing. If it gets more expensive that math only gets worse.

      I can't predict the future, but one possibility is that AI will not be a general purpose replacement for human effort like some hope, but rather a more expensive than expected tool for a subset of use cases. I think it will be an enduring technology, but how it actually plays out in the economy is not yet clear.

      Reply View 1 reply
      • ringeryless 2 days ago

        writing spam emails, that's what LLMs enduring market niche will end up being.

        without structural comprehension, babbling flows of verbiage are of little use in automation.

        CAD is basically the opposite of such approaches, as structural specifications extend through manufacturing phases out to QA.

        Reply View 0 replies
    • malfist 2 days ago

      Is it too expensive? Or not a valid solution to real problems?

      Reply View 0 replies
deepsun 2 days ago

Deja vu, Zuck has already scaled down their AI research team a few years as I remember, because they didn't deliver any tangible results. Meta culture likes improving metrics like retention/engagement, and promotes managers if they show some improvement in their metrics. No one cares about long shots generally, and a research team is always the long shot.

Reply View 0 replies
mvdtnz 3 days ago

> they're being usurped on all sides between TikTok, X and BlueSky

Good grief. Please leave your bubble once or twice in a month.

Tiktok yes. X and Bluesky, absolutely not.

Reply View 5 replies
  • tracker1 2 days ago

    Monthly active users:

    From DemandSage:

        Facebook - 12 billion!?
    TikTok - 1.59 billion
    X - 611 million
    Bsky - 38 million
    That's according to DemandSage ... I'm not sure I can trust the numbers, FB jumped up from around 3b last year, which again I don't trust. 12b is more than the global population, so it's got to be all bots. And even the 3b number is hard to believe (at close to half the global population), no idea how much of the population of earth has any internet access.

    From Grok:

        Facebook - 3.1 billion
    TikTok - 1.5-2 billion
    X - 650 million
    Bsky - 4.1 million
    Looks like I'm definitely in a bubble... I tend to interact 1:1 as much on X as Facebook, which is mostly friends/family and limited discussions in groups. A lot of what I see on feeds is copy/pasta from tiktok though.

    That said, I have a couple friends who are die hard on Telegram.

    Reply View 4 replies
    • Imustaskforhelp 2 days ago

      pardon me but I am just a little surprised as to how telegram came in the last paragraph? we were talking about social medias and telegram is a messaging app...

      Reply View 3 replies
      • 4ggr0 2 days ago

        Telegram Groups sometimes blur the line between message-only interactions and social media dynamics, depending on how many bots and features they got running.

        Reply View 2 replies