In a few years, I don't think we'll be navigating to Facebook.com. We'll be asking our LLM: "what are my friends up to?" And it will provide a nice UI to browse interesting updates, collated from any open source, including Facebook. No ads, no influencers, no dark patterns, just what the user asks. And it won't require APIs. It will just browse the site itself as an agent and pick out the parts the user cares about.
I really think the new internet is going to be a radically different experience soon.
> No ads, no influencers, no dark patterns, just what the user asks.
For a while now, I've seen blogposts from marketers desperate to do to AI what SEO did to search engines.
I hope they fail to enshitify the LLM experience the way they enshitified search, but I wouldn't count on it in the absence of AI messing with production so hard that nobody even wants to market stuff in the first place.
The risk vector shifts from Facebook to the LLM provider. This is why I'm a big fan of the open weights models. They're not immune to attack, but it's much easier to identify censorship, propaganda, and advertising. They're getting really good, really fast.
In a few years, I don't think we'll be navigating to Facebook.com. We'll be asking our LLM: "what are my friends up to?" And it will provide a nice UI to browse interesting updates, collated from any open source, including Facebook. No ads, no influencers, no dark patterns, just what the user asks. And it won't require APIs. It will just browse the site itself as an agent and pick out the parts the user cares about.
I really think the new internet is going to be a radically different experience soon.
> No ads, no influencers, no dark patterns, just what the user asks.
For a while now, I've seen blogposts from marketers desperate to do to AI what SEO did to search engines.
I hope they fail to enshitify the LLM experience the way they enshitified search, but I wouldn't count on it in the absence of AI messing with production so hard that nobody even wants to market stuff in the first place.
At least there is no chance of manipulation and subliminal advertising that way. /s
The risk vector shifts from Facebook to the LLM provider. This is why I'm a big fan of the open weights models. They're not immune to attack, but it's much easier to identify censorship, propaganda, and advertising. They're getting really good, really fast.