Designing human experiences in the age of AI.
Links curated by Peter Polgar.
For a weekly newsletter of hand selected links, signup here.
Posted on 2018 January 21 by polgarp · Tagged in Research and Machine learning
Original link: https://www.wired.com/story/crime-predicting-algorithms-may-not-outperform-untrained-humans/
Interesting results. One way of coming up with ideas for new machine learning products is to think about problems where human intuition and experience solves the problem. Ofcourse the resulting product won’t be better then a human would be, only better at scale. What’s strange here is that the algorithm would be only as good as untrained humans. Issues + bias in data and no validation on results maybe?