Cool, I wonder how troubleshooting speed with this would compare to lab members with different levels of experience with a given protocol, I’m guessing people with good “lab hands” have these probabilities encoded in their decision making processes at some level
Since it learns from data it can systematise the intuition that senior lab members have, freeing them up to focus on more important things.
This is very good argument for developing an open protocol for emitting telemetry from lab equipment: it'll be very easy to send that data into a system like this to update the posterior distributions of failure modes.
1 comment