this post was submitted on 16 Mar 2025
456 points (97.9% liked)
Technology
66783 readers
5673 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related content.
- Be excellent to each other!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, this includes using AI responses and summaries. To ask if your bot can be added please contact a mod.
- Check for duplicates before posting, duplicates may be removed
- Accounts 7 days and younger will have their posts automatically removed.
Approved Bots
founded 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
That’s always been their MO though. Take a recent innovation, and implement it better. That always means it’s later than tech from other places, but they get it “right”. Yes, I know that’s subjective.
In the case of AI, they scrambled to announce the feature with barely any work done on it. Had they kept mum about Apple Intelligence features for a year or so and then revealed, that would be the Apple way.
Also by all indications the current direction in Machine Learning (stuff like LLM) is a dead end which will never yield a "reasoning artificial intelligence" (even whilst quite a lot of other areas in ML have already reached sufficient capabilities in their domain to actually be useful) so there really isn't any space to "implement (the main subdomain of ML that has been promoted as) AI better" IMHO.
It has its uses and who though we where anywhere near general intelligence. They are still making strides in LLM and it can still improve and is useful.
LLMs have already massively slowded down in terms of improvement from generation to generation and they're not at all improving when it comes to logical errors (because they're not structured for that at all - they're massive statistical engines for language, not reasoning devices), so it seems unlikely that this stage of LLM evolution is the beginning of something massive, rather it looks like it is has gone as far as it can.
Not saying they're useless, just not at the early days of a game changer technology.
When Apple got into personal computers, that tech was just about to go from a niche thing to mass adoption (from big machines in universities and very large companies to mass adoption by consumers and businesses) and 3 decades of advancing by leaps and bounds, and similarly Apple's entrance into portable networked computing (with iPhones and iPads) pretty much turned the niche of ultra portable computing devices (such as the Palm Pilot) into an omnipresent mass market product.
A lot of that was getting in early and then ridding the wave of incremental tech improvements in those areas and related areas.
What exactly wave of tech improvements is there going to be from now onwards on LLMs given that they've barelly improved in terms of output and the only significant improvement in the last couple of years was Deepseek's significant reduction in required computing power from "insane" to "massive"? Even some kind of amazing fall in required resources crossed to mass adoption of NPUs and TPUs would still not solve the reliabilty problems of the Technology and so far nobody has managed to crack that specific nut.
I was there when personal computing took off, when mobile networked personal computing took of, when the interned took of and so on, and what I'm seeing with LLMs now (not 2 years ago, but now) doesn't at all feel like being at the brink of a revolution like with at least the personal computer and the internet (the smartphone looked more like a cool gimmick back then, to be honest).
Frankly the AI "Revolution" at this stage feels a lot more like the Bitcoin "Revolution" after a few years and it having been taken over by greed and speculation, than the Personal Computer Revolution or the Internet Revolution.
It probably doesn't help that the tech in question, LLMs, are kinda shit, to put it plainly. You make the shiniest, most polished turd and it's still just a turd. They are interesting and can be neat to play with but, they lack practical applications where cost to run them actually makes sense and benefits humanity. The iPod shuffle was more impactful, when measuring positive impact on people's lives.
Ah, the mad rush to be Second Place.