5 comments

[ 2.8 ms ] story [ 22.5 ms ] thread
I think there is an opportunity for GenAI chatbot platforms to offer something similar to Quora, Stackoverflow, Reddit, etc and then train on it. answers.tld.com, for example. Human mods curate and perform QA (assisted with LLMs), and the models are built off of the corpus. Reddit is trying to turn into ChatGPT using Reddit Answers, when OpenAI and other platforms already have the traffic, MAUs, and investment.
If I use a chatbot it’s precisely to avoid going to your website. Have you tried using it from a mobile device without an account and ad blocker? You will quickly understand why people don’t want to go to your website.
I want to thank reddit for imposing limits on the exposure of their content to LLM scrapers.

> What’s more, Reddit is experimenting with a new and simpler onboarding flow to increase early user engagement.

Calling it now - depreciation of old.reddit.com with simultaneous rollout of a login wall like X.

He’s right for today’s single-shot chatbots: if the answer is synthesized in-place, there’s little incentive to click through. The shift will come from agentic LLMs and multi-agent systems that plan, browse, cite, and hand off—e.g., parallel agentic AI that retrieves, verifies, and selects authoritative links, then routes users to the source. When models are rewarded for attribution and task completion (not just word prediction), distributed agentic AI can become a net traffic driver to forums like Reddit.