•   over 2 years ago

looking for AI teammate

I'm interested in the online reputation of reliable content to prevent misinformation. I know how to do a Chrome Extension that can match the URL against a database of trusted sites. I need an AI partner who can calculate reputation scores with ML on all the covid-19 links.

  • 6 comments

  • Manager   •   over 2 years ago

    Tom! Welcome to our Hackathon! This is a really cool area of focus. I think one can scrape data about the sources (authors, publications,) use the scientific journals that have been released, and others to then use to compute some credibility score.

    I will be monitoring this to try to make sure you find a teammate.

    Miguel (Lumiata CTO)

  •   •   over 2 years ago

    I think in this case, less is more. Like wear mask or not wear it. Chloroquine or not. Stop overloading people with new information as there's prob not a lot of reliable ones. Most are experiments. Scientific journals included. As someone who is pretty informed, frankly, I'm still unclear on Chloroquine. So maybe manually curate from top leaders links to sources and use AI as backup to surface interesting topics.

  • Manager   •   over 2 years ago

    I think there is something about combining the "expert curation" with ML approach. The simpler the solution, the better if it yields the desired results. Regardless of the specific solution, I think this is a super important area given the vast amounts of misleading, and biased (not in a good way) information.

  •   •   over 2 years ago

    I like this idea. And there are bigger efforts that I know of that are trying to attack this problem. Happy to discuss that. One thing you have to keep in mind and appreciate is that even if you have a source/publications because so much of this is being done without proper study design principles the validity of those results are highly questionable. I cannot tell you how many publshed/pre-print papers I have come across in the last three weeks that wouldn't pass muster.

  •   •   over 2 years ago

    I mean even the US President is misinforming others. There’s no way to prove absolute “truth” these days. We can only show the patterns. Who say what when and how many support it. Let viewer make their choice.

  •   •   over 2 years ago

    I don’t think we need to worry about setting any benchmarks on information based upon Trump. It can’t be as simple as who says and how many support it, because that just translates into the amplitude of the signal. Relying on picking high amplitude signals alone will not be enough. Happy to discuss offline or through Slack.

Comments are closed.