AI fact check integration

dutchbb

Well-known member
Yes please.

With option to check only in selected forums, groups or posts containing certain keywords.

AI is not always correct, but it could help steer some discussions to something more productive.
 
Upvote 2
Seems more like something that would be better served with a more general approach that admins and devs could build on. Ability to send requests/prompts to any AI via its API but then developers could build fact checkers, content generators, or whatever using that. That also lets the forum admins/owners/devs decide which AI (ChatGPT, Gemini, etc.) to use.
 
Looks like the best option would be OpenAI API using gpt-3.5-turbo.

Cost Example:If a user submits a 500-word post (approximately 667 tokens) and the AI generates a 250-word response (approximately 333 tokens), the costs would be:
  • GPT-3.5 Turbo:
    • Input: 667 tokens × ($0.0015 / 1,000 tokens) ≈ $0.001
    • Output: 333 tokens × ($0.002 / 1,000 tokens) ≈ $0.0007
    • Total per fact-check: Approximately $0.0017
 
Looks like the best option would be OpenAI API using gpt-3.5-turbo.

Cost Example:If a user submits a 500-word post (approximately 667 tokens) and the AI generates a 250-word response (approximately 333 tokens), the costs would be:
  • GPT-3.5 Turbo:
    • Input: 667 tokens × ($0.0015 / 1,000 tokens) ≈ $0.001
    • Output: 333 tokens × ($0.002 / 1,000 tokens) ≈ $0.0007
    • Total per fact-check: Approximately $0.0017
If you go through OpenRouter you’ll have about 300 options and spend less.
 
Wouldn't that be a bit much to expect admins to go through though to set up a API? Also I wold think support would be a nightmare with so many options to use.
 
Wouldn't that be a bit much to expect admins to go through though to set up a API? Also I wold think support would be a nightmare with so many options to use.
Easy
 
Wouldn't that be a bit much to expect admins to go through though to set up a API? Also I wold think support would be a nightmare with so many options to use.
OpenRouter AI API has a standardized API to interface with all 300+ LLM models. So you only need to setup a single API to have access to all 300+ LLM models. OpenRouter has free LLM models https://openrouter.ai/models, and I have been pushing ~40 million tokens/month for practically free for my script https://github.com/centminmod/or-cli :) Extended examples at https://github.com/centminmod/or-cli/tree/master/examples :D

Xenforo thread summarization examples
For fact checking, you probably need to add web search support to LLM and OpenRouter natively supports Exa web search https://openrouter.ai/docs/features/web-search. I added Linkup and Brave API web search as they have a free tier for testing. Linkup has 1000 free standard searches/results per month and Brave API has 2000 free standard searches/month. So combined 3K free web searches/month. Then for paid, OpenRouter's Exa is actually cheaper at $4 per 1K searches vs $5 per 1K searches.
Give OpenRouter AI API a spin https://openrouter.ai/docs/quick-start ;)
 
Last edited:
Back
Top Bottom