They're getting it by paying contractors at Scale AI to write content for them.

> They're getting it by paying contractors at Scale AI to write content for them.

Just a wild guess, but at best that content is probably pretty mediocre quality. It's probably Mikkelsen Twins ebook-level garbage.

Not anymore. There's arguably more than enough data to form a base for strong LLMs; extra data is nice, but doesn't have to come in such quantity.

(In fact, there's value in trying to filter excess crap out of existing training sets.)

We're talking about LLM-driven search engines here, the assumption is that they will always need up-to-date information. A "strong LLM" can't give you to latest on the presidential election if its knowledge cut-off is in 2023, so these companies "solution" is to scrape today's New York Times and get the LLM to write a summary.

You still need fresh data for many use cases.

Uh, sorry what?

What happens when you need to search something new? Just hallucinations all the way down?

LLMs arent embodied. They can not break news as they have no ability to gather fresh news

For all the hate it gets, Brave solved this half a decade ago already.

- Publishers no longer show you ads, they just get paid out of BAT.

- Brave shows you ads, but Brave does not depend on that to survive. Because of that there is no weird conflict of interest like with Google/Facebook, where the party that surfaces your content is also the party providing you with ads.

- Users can just browse the web without ads as a threat vector, but as long as you have BAT (either via opt-in Brave ads or by purchasing it directly) you are not a freeloader either.

The crypto part of brave is the worst part of brave..

How is it the worst part?

You opt-in to the ads, you get them in your notifications, and every time you tap on one of them you get a few BAT. You browse, the BAT get paid out to whichever sites you visit (or linger on, depending on your configuration). You can opt out of the ads at any time. Brave didn't pre-mine their own coins. And you can buy BAT if you want to support sites without watching ads.

> Brave solved this ... Brave shows you ads

Showing people ads is part of the problem to be solved.

It is solved. You can pay Brave to not be shown ads.

And the website gets paid either way.

How else are you going to pay the people writing articles, creating content and doing research?

Not every person/site can run on Patreon or sponsorship deals. And paywalling a lot of the web would exclude vast swathes of people.

1. With ads, I don't pay people writing articles, creating content or doing research.

2. "How else would you achieve X than by manipulating people visiting your website into paying for things they probably don't need, and be misinformed and tracked by powerful commercial and political entities?" - I can but shrug at this question.

3. The vast majority of written content is never rewarded or compensated monetarily, ads or no ads.

> With ads, I don't pay people writing articles, creating content or doing research.

You do. The ad broker sells access to your eyeballs to a company, and then gives part of that money to whichever parties have a monetization agreement in the content.

> "How else would you achieve X than by manipulating people visiting your website into paying for things they probably don't need, and be misinformed and tracked by powerful commercial and political entities?" - I can but shrug at this question.

Always fun to see people with strong opinions be critically misinformed.

Brave’s ads don’t have tracking, by design.

> The vast majority of written content is never rewarded or compensated monetarily, ads or no ads.

By that logic we should stop paying for art?

People who have something meaningful to say usually do it anyway.

The others can shut up.

Research is usually not paid by ads.

Some of the best scientists did it for fun. Einstein wrote his 1905 papers while working at the swiss patent office.

Probably no one here is trying to profit by their additions to the discussion, but they’re still regularly sharing very useful information.

I don't think I'd publish on a blog if I knew it would be consumed entirely from an LLM. Even if I don't want money, I kind of want people to know that I wrote it, they're my ideas, and I don't want them taken out of context.

From the look of the examples they use at least sites like TripAdvisor, which I'd assume would like to have people go to their site.

For services like rail companies, restaurants and so on, I can see them not being bothered, because they make their money offline. Blogs, and pure content sites might not be to happy about their content being used to prop up OpenAIs business, while getting nothing in return. If anything it seems like OpenAI is actively trying to get sued.

The whole interface and functionality seems really nice and a clear improvement for the users, for certain types of queries at least. It assumes that OpenAI has made their LLM stop lying, and that they can get the required data legally, but it doesn't seem like anyone cares about those details.

Something has to die, so that something new can live

But if the data provides die, so that LLMs may live, where will the LLMs get new data?

From other LLMs of course.

Which is the main issue I see with them, if no one publishes anything new, all you will get is whatever was there before, which may be incorrect or obsolete. If you want something new, an LLM can't discover it if no data source exists. So over time the LLM becomes useless for current information.