ACT-W: UX and Machine Learning

Transcript

1. Solving What to Solve UX + Machine Learning @ashleydzick
2. The 20K Foot View
4. Input Output Hidden Layer variables result
5. correct!
6. not so correct..
7. Machine Learning is Everywhere
10. medium.com/netflix-techblog/
 artwork-personalization-c589f074ad76
11. https://algorithms-tour.stitchfix.com
12. Machine Learning Your $$$
14. Solving What to Solve
15. If you aren’t aligned with a human need, you’re just going to build a very powerful system to address a nonexistent problem.
 
 - Google’s ML Team
16. Can you solve your problem without machine learning?
17. mail.google.com says It seems like you forgot to attach a file. You wrote “I have attached” in your message, but there are no files attached. Send anyway?
18. Is this problem worth your time? money, effort
21. Why does the problem need to be solved? Impact Advantage ROI
22. Positive user impact Machine learning requirements
23. Positive user impact Machine learning requirements Forgot Attachment 
 in Gmail Amazon 
 ResumeBot Product 
 Recommendations Self Driving Cars No UX, No ML
24. Positive user impact Machine learning requirements Forgot Attachment 
 in Gmail Amazon 
 ResumeBot Product 
 Recommendations Self Driving Cars No UX, No ML
25. When in doubt, just go ask your users.
26. Setting Yourself Up For Success
27. Well-defined problem Realistic scope Validated with users Meaningful and impactful
28. Describe it in simple terms your ML problem
30. List assumptions
32. for ________ users in ________ situations, we assume they’ll prefer ________ and not ________
33. For Google Photos users searching we assume they’ll prefer “playgrounds” and not searching for “children”
34. You’re ready to start solving! slides will be on twitter, @ashleydzick