Hacker News Reader

Comment by wilg | Hacker News Reader

Who cares, just check back in a year and see how its going.

Yep, the progress will be slow but inexorable on this front.

Sooner or later we'll arrive at what I see as the optimum point for "AI", which is when I can put an ATX case in my basement with a few GPUs in it and run my own private open source GPT-6 (or whatever), without needing to get into bed with the lesser of two ShitCos, (edit: and while deriving actual utility from the installation). That's the milestone that will really get my attention.

You already can run a local llama instance on a high-end graphics card (6+ GB VRAM).

Yes, I can, but (see my edit) there's very little utility because the quality of output is very low.

Frankly anything worse than the ChatGPT-3.5 that runs on the "open"AI free demo isn't much of a tool.

And it's hilariously bad (in comparison to regular chatgpt).

And slow. They never tell you that quantization of many LLMs slows down your inference, sometimes by orders of magnitude.

It depends on the quantization method, but yes some of the most commonly used ones are extremely slow.

Precisely my point I don’t think a lot of people will go back. Even somebody like me who’s willing to put several hours into trying to see how both work won’t do that for every blog post about an “improvement”.

Bard was rushed, and it shows. You only get one chance to make the first impression and they blew it.

I think there's a way in which ChatGPT is paying this, by having released GPT-3.5, rather than just waiting 6 months and releasing it with GPT-4 out of the gate. In this thread everyone is making a clear distinction, but in a lot of other contexts it ends up quite confused: people don't realize how much better GPT-4 is.

I don't think so for stuff like this, it kinda has to be built in public, and iteratively. If it gets good enough they'll surface it more in search and that'll be that.

Partially agree with that sentiment but I don’t think it negates my point that they released something inferior because they were caught flat footed.

I agree they did release it because they were caught out by OpenAI. But also I'm fine with them starting there and trying to improve!

Yeah, competition is good. Glad Nadella and Altman are making them “dance”.

What? After a year, they'll hear that Bard is really good at code assistance now and then they can try it again.

Yes, but switching costs increase over time, especially with API integration, and it’s not like OpenAI isn’t also improving at what seems to be a faster rate. My code results on ChatGPT seemed to have gotten a real bump a few weeks ago. Not sure if it was just me doing stuff it was better at, or it got better.

DuckDuckGo is closer to Google Search than Bard is to ChatGPT at this point, and that should be a concern for Google.

I hope it's less than a year when I hear that Bard remembers your last chat on refresh or either one (Bard or OpenAI) implements folders...

Competition is competition and I respect that.

I'll use whatever is best in the moment.

And if chatgpt start trying to network effect me into staying locked with them, I'll drop them like a bad date.

Been there, done that. Never again.

Ymmv

Reply on news.ycombinator.com