Hacker News Reader

Comment by slopinthebag | Hacker News Reader

Meanwhile the world keeps spinning and most people don't even know what Anthropic is, much less anything about Fable.

If AI lived up to a tenth of the promises the American labs produce, the world would be drastically different today. It's not. I'm doubtful of future impact based on that.

I'm happy we can utilise current OSS models to the extent we can now. They'll improve. The world will continue as usual. And hopefully we can put this bubble behind us.

View on HN

Ask a recent college grad if the world is drastically different today then when they started college.

If you mean employment, the world is different because of rising debt, declining economies, and a crazy leader currently in charge of the most powerful country on the planet. If you asked me when I graduated if the world was drastically different from when I first entered university I would also say yes, and I graduated well before GPT2.

I can’t understand that you think designers, swe, technical writers, help desk etc etc at a jr level aren’t being affected by genai.

Did you think 5 years into the invention of electricity the world already was vastly different? The internet? Would you have written them off because random people didn't know much about them at that point - which isnt even true as chatgpt has been ~ the 5th most popular site in the world for a couple of years now?

> which isnt even true as chatgpt has been ~ the 5th most popular site in the world for a couple of years now?

That part is kind of their point - it doesn't have the distribution issues your other examples have.

I mean, besides the fact that electricity and the internet are orders of magnitude more transformative than a statistical next-token prediction machine, none of the predictions behind LLMs were made of either in the first 5 years.

Gangnam Style is the most popular video ever, surely it means something right?!

If we're cooked, it's only because of a mass hysteria behind this thing. It's an extremely useful technology, we're just losing our collective mind because of it.

100% correct, but the last thing a person suffering from delusions wants to hear is that they’re suffering from delusions. LLMs inadvertently hack some very primal parts of the human brain, and specifically the part that anthropomorphizes things with seemingly human-like behavior. Combine that with herd behavior caused by social media and you have a perfect mass delusion machine. It’s nearly impossible for people inside the delusion to see their way out.

This is going to be studied for many, many generations to come.

You think machine learning has only been around for 5 years? It has been around for 76 years, since 1950. The first pattern matching chatbot was Eliza, created in 1966. Shannon-style next-word statistical model chatbots were created in the 1980s. The first neural models were created 26 years ago. Large language models have been around for 10 years.

This technology is not new and has been around since well before the internet.

Reply on news.ycombinator.com