Hacker News Reader

Comment by codingwagie | Hacker News Reader

People are just click farming with these posts. The technology is ~4 years old. We are in the infancy of this, with hundreds of billions of capital behind making these systems work. Its one of the biggest innovations of the last 100 years.

I propose an internet ban for anyone calling the generative ai top, and a public tar and feathering

View on HN

If you had said the last 20 years then maybe but last 100 years? For a really good autocomplete?

Just think about the innovations over the last 100 years, how the world looked in 1925

It's incredibly good autocomplete, though!

I remember the 90s, and the time when translation programs were truly awful. They had pretty much no memory, not knowing what "it", or "she" might refer to, from the previous sentence. They tended to pick a translation based on the selected dictionary, and choked on typos and trademarks. A popular past time in my neck of the woods was feeding a translator the manual for a computer mouse, selecting the medical dictionary and giggling at all the incoherent talk about mouse testicles and their removal and cleaning procedures.

In light of that, ChatGPT is scifi magic. It translates very well, it deals with informal speech, it can deal with typos, and you can even feed it a random meme JPG and it'll tell you what's it about. It's really like living in the future.

Well human beings are just auto complete if thats all LLMs are

Yes, the classic retort.

It’s absolutely amazing what these things can produce with simple prompts. I do think it deserves recognition as a top 5 innovation of the last 100 years. We can call it really good autocomplete but it really is incredible.

That doesn’t mean I agree all jobs are going away, AGI is here, or all these other guesses people are making about the future.

If all you've used it for is autocompleting code, you're doing yourself a disservice. In 1925, there were already cars and airplanes and telephones, even if they were still very new. There have been great leaps and advances in medicine and genetic engineering like CRISPR and mRNA vaccines. Computers and the Internet are the two really huge things from the past 100. The industrial revolution predates 1925. Where a man could make one widget before, with a machine, he could make a thousand of them. So there's a multiplicative effect from that machine. Where you needed 10 humans you could now use one. AI is still developing, but promise of it is even greater because we can, in the hoped for future, copy paste the human who does the work. It's a ludicrous idea, and the societal impact would be immense, and we're not there yet, but given that there were already telephones in 1925, an optimists view of what we're on the cusp of really is revolutionary.

That, of course, is the optimist's view. If you cynically see LLMs as a dead end that won't ever get that far even with Moore's law, then any day now we're going to come to our senses and give up on the whole thing, but looking at how we've come to our senses about crypto and Bitcoin is now worth $0, all I can say is that I'm along for the ride.

I am with OP. Being able to have full meaningful conversations with a machine about any topic in existence is up there with invention of flight.

People have been saying it's still early in crypto for over a decade.

This much capital being poured into something and having very little to show for it is actually a bad sign, not a positive.

Putting it on the same shelf as the transistor, the jet engine, and the nuclear bomb is pretty funny. It's a probabilistic token generator. Relax.

this is nothing like crypto

Besides the reliance on GPU compute, the insistence of its proponents that it is inevitable, and the plowing of resources into the space by venture capital, sure, it's nothing like crypto.

>We are in the infancy of this, with hundreds of billions of capital behind making these systems work

Just like IoT, just like web3, just like blockchain, just like...

Agreed. To me, this is reminiscent of the "dot-com bubble" 25 years ago. The internet changed the world permanently, even if the stock market got ahead of itself for a few years. The same is true of generative AI.

https://en.wikipedia.org/wiki/Dot-com_bubble

Damn I feel old that you had to link the dotcom bubble

The funny thing is, when the first dotcom bubble had been going on, people would have been making references back to the first AI winter in 1974-1980, where funding for AI related projects dropped precipitously as interest waned.

Reply on news.ycombinator.com