Ask HN: How much [more] value to AI do declining social media platforms have?

[AbstractH24]

Clearly both X and Meta are trying to leverage the massive historical dataset they have of real human interactions as an advantage as they pivot into becoming AI companies. Neither of their social media platforms are on the assent anymore.

I’m starting to wonder though if these datasets really are as valuable as they think and if they will remain valuable as this space matures.

Thus far, I don’t know of anyone using Grok or LLAMA. Are these pivots gonna be too little too late? Are these companies on the verge of rapid implosion?

[ttyprintk]

How do you think they derive worth from old posts and tweets?

Meta will make more money from Instagram than Facebook. Inside X/Twitter, I assume they’re desperately trying to negotiate for corporate-palatable influencers.

[AbstractH24 → ttyprintk]

Training data for what real human interactions sound like.

I agree, the value is limited. 18 months ago or so it seemed more important. Thus, the question.

[ttyprintk → AbstractH24]

I feel like the most-valuable silos for that are LinkedIn, ClearanceJobs, and GovConNet. Those datasets are more-often between humans and have money and/or trust on the line.

[AbstractH24 → ttyprintk]

I understand the first one, but why the second two?

[ttyprintk → AbstractH24]

Among the sites which facilitate connections between strangers, those have the strongest incentives to ensure both are human. In terms of value for training, you’d want to avoid any conversations which are part of the dead Internet bot-to-bot problem.