Maybe their AI should be trained to fact check stuff posted on Facebook?
So if you post something that claim outrageous things, it would be visible right from the start that the claim is false.
It will be an interesting arms race, but I'm hopeful for truth there.
I can imagine good enough AI being able to spot truth even better than what humans do - by veryfing sites and commenters with sources of real information to estimate their credibility.
E.g. in a theme similar to Page Rank, you could have an AI that has some sites as a source of objective truth (Wikipedia, science journals, reputable sources of news etc), and then use that as a basis of estimating trustworthiness of a material.
Also, AI could find, for a given subject, opposing opinions, and estimate which ones are possibly fake, and which ones are real.
In essence - do what current fact-checkers do, but for every single website and comment in existence.
Very interesting. How do you think about navigating the misinformation problem? I would imagine that the AI could result in a fair amount of bias in the writing over time.
Seems like a sensible policy to me. instead of trying to work out if content is AI generated (which isn't possible in general) just check if it's true. not sure why this is a problem. truth is the only thing that matters in this case
1) It would have to be a human. Existing "AI" solutions aren't close to being as good as humans, at least when they're only fed articles without context.
Even humans aren't exactly infallible when fact-checking, and truth is a debated term. The training data for any AI would be "biased" to some people and not to others.
2) HN is already great at this. That's why I read comments first.
I think that's kind of the problem with these tools lol, there is no obvious solution to this. Automatically fact checking an AI model would probably require a bigger and more sophisticated AI model.
But that is true also of lots of content not generated by AI! Fact checking always needs to be done, AI generated or not. But does it matter that it was generated by AI?
This is probably the only possible answer to fake news. With human censors one can always cry “bias!” but a publicly published open source AI would be irrefutable.
what I really want is a true AI to search through all that and figure out the useful truth. I don't know how to do this (and of course whoever writes the AI needs to be unbiased...)
We are using AI and GenAI to get better at detection, training and policy.
However, There can never exist a tool that can correctly detect harmful content. You would need the intent of a submission to get to that level of accuracy.
Everything else is: AI filters -> Human review.
Your offenders are 2 types: Bad actors and Adversaries. Both evolve to find ways around your detection processes.
That might work in a narrow set of circumstances where data can be published to trusted sources for one to read and say yes this information is true. But in much broader situations AI can spit out disinformation in many locations and it will be information that is not testable like celebrity news and it will be nearly impossible for one to verify truthfulness.
At first I thought the article was going to be about human-led misinformation but I wonder whether with both hallucinations and human-fed misinformation (AI-helped or not!) whether we can use AI to fact/self check results (both AI generated and human ones) and prompt us about potential misinformation and link to relevant sources? That way AI could actually help solve the trust issue.
reply