this post was submitted on 14 Apr 2025
21 points (100.0% liked)
Ask
445 readers
186 users here now
Rules
- Be nice
- Posts must be legitimate questions (no rage bait or sea lioning)
- No spam
- NSFW allowed if tagged
- No politics
- For support questions, please go to [email protected]
founded 1 month ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
AI. For ... the fourth time or fifth time? (I've lost track of the generations of AI.)
Definitely not going to die, but the market/valuation will go way down as people realize it can kinda do a lot of things but mostly sucks at them compared to humans. It's a tool, not a product.
Every previous wave of AI died in the same hype/disappointment cycle. Yes each previous wave still has niche uses, but their economic activity is basically a rounding error.
They're dead.
The bodies are just still twitching a little from the chemical reactions of deccomposition.
The current wave will do the same thing.
It depends on what you consider dead. I was using generative AI in 2020 before this hype wave really began, but it was a healthy and growing niche.
The current state of the art models aren't worth the unsubsidized cost/investment (future models might be but I'm very skeptical). The problem is huge companies are investing expecting ROI that I can't imagine they'll ever get. It's going to crash.
I'm not sure whether the code completion models will ever die. They aren't great but they are useful. Which is basically my assessment of AI as a whole. The vastly more expensive models that are marginally better are probably not sustainable, but local models and less expensive ones will probably stay with us.
But if you're talking mostly from an economic standpoint, yeah I'm pretty sure it's going to crash barring some significant breakthrough on cost reduction or output quality.
I'm talking from the global take on the economy, yes. This wave of AI will go the way of every previous wave: some niche products will use it effectively and the rest of the world will look back with keen embarrassment at this phase of history when people took LLMs seriously.
I mean there's still practical uses for '50s-era "AI" out there. ("Symbolic AI" was it called?) But it is so tiny a segment it is basically nonexistent.