How's AI different than the dot-com era?

Question

I am old enough to have lived the dot-com era. In the mid-90's I had a "cushy" job but once I realized that the internet was going to be huge, I quit my job and rode the dot-com wave.And now here we are with AI.The way I see it AI is going to be evermore disruptive of old business model than the Internet has been; but contrary to the dot-com era, the barrier to entry is a lot higher. Is it?Aside from YC's recent batch, where/how does HN see Innovation leveraging AI and where are the opportunities?I asked ChatGTP and what I got back was average cliche' answers.Can we have a human-powered large scale brainstorming?TIA

atleastoptimal · Accepted Answer

Barrier to entry is very low in terms of leveraging existing tech. You can make a venture funded app just by building a wrapper around OpenAI/Anthropic, etc.
What's different is that AI is a far broader field than dot-com era businesses, so many people will be lost in the shiny hype-stuff and fail to see the true scope of what's changing.
AI is about creating intelligence, and intelligence is the source of all innovation, developing, technology, invention, art, etc. in the world. It's clear to see that with all the benchmarks AI is surpassing humans in, one by one, despite decades of people saying such feats are impossible, that AI overtaking human intelligence is inevitable.

incomingpain · Answer

I agree there's some smoke holding up the AI companies; but it's not industry scale smoke.
There's a significant difference between a industry scale bubble busting and AI which is breaking out right now which could theoretically become a bubble but it's far from it. Obviously its possible later next year we see the bubble burst but I expect something rather extreme would need to expand to get there today.
When I look at this index for example, https://finance.yahoo.com/quote/AIQ/
They had a steady decline for all of 2022. Then a big climb, but I estimate they probably have a 25-30% drop coming soon. Just looking at the holdings, they dont scream AI exactly.
Nvidia/cuda obviously being a huge part of AI and they have a pe ratio of 70? There's certainly an nvidia bubble there unless they show some serious profits soon to avoid crash. Though if they do, we'll be kicking ourselves for not buying now.
Moreover, what's important to note. Just because there's some serious smoke going on there, that bears no witness on the future. If these startups showing up can bring a game breaking product to market, then valuations today are imaginary. You want to get in on them now or asap.
I'm not a gambler investor so I never touch anything like this; but I also don't see any dotcom scale bubble that's going to pop.

bodantogat · Answer

In many ways, it feels the same to me. Specially alarming to me is that some of my friends, who have no idea how LLMS & AI work, are seriously considering investing in it. The barrier to entry is quite low, and getting lower. Almost all the models out there can be used cheaply via API - millions of tokens/words processed for less than a few hundred dollars (if that much). For those willing to spend the time, services like replicate, anyscale are even cheaper. Basically, anyone who can call an API & feed it data, can get into the "AI" business. And they are.

Jesus_code · Answer

I think AI in its current iteration (LLMs) has some value but personally speaking I feel it is over-valued.I think the fact that the cost:value prop is 20:1 to get a basic level output is really alarming.

sk11001 · Answer

It's different in most ways? How is it even similar? Some level of trendiness and hype, both related to technology... that's about it? That's like 1% of all the dimensions on which you can compare these things.Saying that x is the new y, or the same as y, just because they have one or two things in common is intellectually lazy.