Regarding the AI vs human debate (see issues at SuccessFactor)

[Quick thread]

1) AI is a * prediction technology* that does NOT succeed 100% of the time

So it’s going to work for certain business use cases and fail for others

Something like this (low tech chart alert!):
2) Where AI works great: categorization problems that are not in real time where the stakes are low - if Google Photos doesn’t not recognize the pic of little Emma, nobody gets hurt

Where AI fails: needs to be 100% correct in real time or somebody dies - like self driving
3) Cases like accounting automation are tricky: you need to be 100% correct all the time, but since you don’t need to be correct in real time you can supplement with humans
4) What& #39;s difficult for entrepreneurs and investors alike is that, in the early stages of building the venture, you need a lot of humans to train the AI

The hope is that at some point, you go from 60/40 humans/AI (or whatever) to 90/10

But AI is capricious and unpredictable
5) So, regardless of whether the problem you picked is tractable with AI or not, you may get stuck on your way to automation

I have zero knowledge of the facts but, based on the quality of people around the table, I suspect this may have happened here: https://twitter.com/eringriffith/status/1285327407839207424">https://twitter.com/eringriff...
You can follow @mattturck.
Tip: mention @twtextapp on a Twitter thread with the keyword “unroll” to get a link to it.

Latest Threads Unrolled: