3 ms·
> they'd have to get ChatGPT to stop spewing up made up bullshit and fake academic citations firs Why? The alternative is reading your way through SEO spam (an
by luckylion 4y ago
> they'd have to get ChatGPT to stop spewing up made up bullshit and fake academic citations firs
Why? The alternative is reading your way through SEO spam (and soon: reading yourself through ChatGPT-generated SEO spam), which is often just as wrong. Source: am working for SEOs.
- dharma1 4y agoYes - I’m more hopeful of LLMs (or whatever replaces them over time) getting more truthful and accurate sooner than we get rid of SEO spam. I would also imagine SEO spam gets worse soon with LLMs.
- luckylion 4y agoThe incentives are very different. With SEO spam, you have a direct monetary incentive. More spam = more clicks = more money. With LLMs extracting information and answering questions without actually sending the user to the site ... what's the incentive to create low-accuracy content to be ingested, beyond "I want to mess with the data set"? I'm sure there will be a few people going down that route, but that needs a lot of intrinsic motivation. It's a very different beast vs having a multi-billion dollar market draw in millions of interested parties, and you can concentrate on identifying the saboteurs and filtering them out, which feels easier, especially because it's hard to impossible to identify "shadow-banning".
- ipaddr 4y agoOn that note where are they extracting new information from? If sites are never visited people will stop creating them