Wikipedia:WikiProject AI Tools/Idea lab
Controlling hallucinations with LLM's
Plausible approach for controlling hallucinations suggested by Gavin Mendel-Gleason who is working with Peter Turchin.
Summarizing discussions
Having an easy way to summarize discussions - main points/arguments for/against.
Training data
- Are there any efforts to improve FA/GA/FL with the idea that it would provide a stronger dataset for various machine-learning or AI models?
- Are there any efforts to have editors specialize in an area of contributing to Wikipedia so that models could train on their edits when making improvements to sections like Further reading, the lead, infoboxes, etc.?
- Are there any lists of reliable sources that explicitly allow for training and use by AI models?