Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Show HN: AI that watches and chats in YouTube/Twitch live streams

github.com

Post reply on HN
Print view View original

1–10 of 10 posts

Show HN: AI that watches and chats in YouTube/Twitch live streams

#1

Post by bOOOOc »

Hi HN,

I built an open-source tool called Live Stream Chat AI Agent. It lets an AI watch live streams (video, audio, and chat) and intelligently participate in the live chat using LLMs.

How it works: Userscript frontend (adds a control panel on live stream pages) Python backend Currently supports YouTube Live, Twitch, and Bilibili

GitHub: https://github.com/bOOOOcG/Live_Stream_Chat_AI_Agent I'm looking for feedback, ideas, or collaboration — any thoughts are welcome!

Note: Since this is a solo project, some comments and commit messages are in Chinese.

Show HN: AI that watches and chats in YouTube/Twitch live streams
github.com

Re: Show HN: AI that watches and chats in YouTube/Twitch live streams

#6

Post by barbazoo »

> This project enables an AI to act as a viewer in live streams. It captures stream data, sends it to a backend for processing by an LLM (like GPT models), and uses the AI's response to interact with the chat.

> Disclaimer

> Use Responsibly: Automating chat requires ethical considerations. Respect platform ToS and streamer rules. Avoid spamming.

Isn't this by definition spam? And how is one supposed to use it ethically if one doesn't control what is being emitted.

But then also, why at all? I'd say this is exactly what we don't need, AI messing with human discourse.

Re: Show HN: AI that watches and chats in YouTube/Twitch live streams

#7

Post by bOOOOc »

barbazoo wrote: post #6

> This project enables an AI to act as a viewer in live streams. It captures stream data, sends it to a backend for processing by an LLM (like GPT models), and uses the AI's response to interact with the chat. > Disclaimer > Use Responsibly: Automating chat requires ethical considerations. Respect platform ToS and streamer rules. Avoid spamming. Isn't this by definition spam? And how is one supposed to use it ethical…

Thanks for the feedback. Good points.

System prompts can guide the AI's behavior (e.g., its persona and rules).

My initial interest was mainly exploring AI in a live chat setting. Beyond that, a positive use could be to help moderate chat by flagging or filtering spam. Agree that responsible use is key.

Re: Show HN: AI that watches and chats in YouTube/Twitch live streams

#8

Post by bOOOOc »

KomoD wrote: post #5

Some of the use cases you list on your site are not cool. > Brand Marketing: subtly guide topics and recommend products like a human > Information Dissemination: Subtly guide conversations and share knowledge

Hi @KomoD, good call – those "subtly guide" use cases for marketing/info dissemination were indeed not cool. I've removed them from the project's website, Thanks for pointing it out.

Re: Show HN: AI that watches and chats in YouTube/Twitch live streams

#9

Post by barbazoo »

bOOOOc wrote: post #7
barbazoo wrote: post #6

> This project enables an AI to act as a viewer in live streams. It captures stream data, sends it to a backend for processing by an LLM (like GPT models), and uses the AI's response to interact with the chat. > Disclaimer > Use Responsibly: Automating chat requires ethical considerations. Respect platform ToS and streamer rules. Avoid spamming. Isn't this by definition spam? And how is one supposed to use it ethical…

Thanks for the feedback. Good points. System prompts can guide the AI's behavior (e.g., its persona and rules). My initial interest was mainly exploring AI in a live chat setting. Beyond that, a positive use could be to help moderate chat by flagging or filtering spam. Agree that responsible use is key.

Good points. Most of all you built something end to end and showed it off, that's a great achievement. Who am I to criticize :+1:

Re: Show HN: AI that watches and chats in YouTube/Twitch live streams

#10

Post by tomhow »

davydm wrote: post #3

[flagged]

Hey there, we've noticed that some of your comments seem to be breaking the guidelines, particularly these ones:

Be kind. Don't be snarky.

Please don't fulminate. Please don't sneer...

Eschew flamebait. Avoid generic tangents. Omit internet tropes.

Please don't use Hacker News for political or ideological battle. It tramples curiosity.

Please keep in mind that the purpose of HN is to satisfy intellectual curiosity, and we're trying to optimize for a different kind of discussion here. If you could please take a moment to read the guidelines and make an effort to observe them in future, we'd appreciate it.

https://news.ycombinator.com/newsguidelines.html

Post reply on HN