I can't wait to not integrate with their service because of how they will be deprecating Bard in 2024, launching Dylan and Hex, basically the same products but by two different teams, neither doing what people used Bard for as well as Bard did.
This makes me wonder if Google is about to have its Internet Explorer vs Netscape Navigator moment?
ChatGPT isn't there yet, but I can see it replacing Google search in the future. Just like Linux never won the desktop, but Android replaced Windows by numbers, it's quite possible that the old way we search the open web will be replaced by a different model. An AI that answers all your questions. The timing is good, because Google search is getting worse every year, mostly due to SEO spam. Having tools like ChatGPT feels like SEO spam will get even easier to generate and harder to detect.
ChatGPT plans to charge for usage, and that could be their biggest mistake. If I were Google, I would jump on this, make my own version and give it away for free until all competition runs out of money. Gradually add ads on top of it and keep the leading position. They can even use their dominant position with Android and Chrome to make their AI the default "AI engine".
If you feel like they stand no chance, remember that Internet Explorer won and used to be the default for years and years.
They probably have enough money at the moment. The question is how long can they offer the service for free and whether they will go with subscription model or ads model.
As a content creator, what's my incentive to not block any bot from my site? Personally I think business models need to evolve along with the paradigm shift. I'll happily provide an interface for any bot to come and extract content from my site. I'll even add a bunch of context to help the lil' guy out. But I want to be paid for it. But it's not realisitc for google to have relationships with every single website. I think we need an open protocol to help build a market. Especially since some data is worth more than others (my pokemon database is not worth as much as your stock prediction database).
Know some people will downvote me for saying this, but I think crypto might be the best way to automate this. Would suck if Google or Microsoft or some startup ended up centralizing that market.
> ChatGPT isn't there yet, but I can see it replacing Google search in the future.... The timing is good, because Google search is getting worse every year, mostly due to SEO spam. Having tools like ChatGPT feels like SEO spam will get even easier to generate and harder to detect.
Honestly, what's stopping SEO spam from taking over this "ChatGPT search," except that it's a new model that they haven't yet learned how to game? It's not an AGI, it's a algorithm that does some mindless but neat stuff with text. At some point it will run into the same type of SEO problems Google does, and then the question is if they're willing to try hard enough to stay ahead in the arms race.
Bard is actually not launched to the public at all, they just made a PR announcement and are picking up some extra testers so their PR can have a grain of truth to it. They seem to be giving access to an API for a small group of individuals or companies they have chosen. Which seems like nepotism and very unfair to me. Although OpenAI did it initially also.
But.. some people may not be aware but some of the key Google researchers started their own company awhile ago called https://character.ai which has an amazing large language model. Not as generally useful as the OpenAI models but still amazingly humanlike and fun.
I think basically the character.ai people looked at the tendency of these things to hallucinate and lack of citing sources and decided it was more appropriate for simulating fictional characters.
I can’t wait until it asks me between every sentence if I finally want to upgrade to YouTube premium. My decision might have changed in the last three breaths.
24 comments
[ 5.3 ms ] story [ 46.5 ms ] threadI was very tempted to say "announces an announcement" (https://hn.algolia.com/?dateRange=all&page=0&prefix=true&sor...) but self-calmed.
"Bard will be used to a group of testers before being rolled out to the public in the coming weeks, the firm said."
https://blog.google/technology/ai/bard-google-ai-search-upda...
ChatGPT isn't there yet, but I can see it replacing Google search in the future. Just like Linux never won the desktop, but Android replaced Windows by numbers, it's quite possible that the old way we search the open web will be replaced by a different model. An AI that answers all your questions. The timing is good, because Google search is getting worse every year, mostly due to SEO spam. Having tools like ChatGPT feels like SEO spam will get even easier to generate and harder to detect.
ChatGPT plans to charge for usage, and that could be their biggest mistake. If I were Google, I would jump on this, make my own version and give it away for free until all competition runs out of money. Gradually add ads on top of it and keep the leading position. They can even use their dominant position with Android and Chrome to make their AI the default "AI engine".
If you feel like they stand no chance, remember that Internet Explorer won and used to be the default for years and years.
Know some people will downvote me for saying this, but I think crypto might be the best way to automate this. Would suck if Google or Microsoft or some startup ended up centralizing that market.
I thought they pretty much do via AdSense.
Honestly, what's stopping SEO spam from taking over this "ChatGPT search," except that it's a new model that they haven't yet learned how to game? It's not an AGI, it's a algorithm that does some mindless but neat stuff with text. At some point it will run into the same type of SEO problems Google does, and then the question is if they're willing to try hard enough to stay ahead in the arms race.
But.. some people may not be aware but some of the key Google researchers started their own company awhile ago called https://character.ai which has an amazing large language model. Not as generally useful as the OpenAI models but still amazingly humanlike and fun.
I think basically the character.ai people looked at the tendency of these things to hallucinate and lack of citing sources and decided it was more appropriate for simulating fictional characters.