Most routine tasks that everyone can do are impossible to automate with current ML technology, even after spending millions of dollars to collect training data
Show this thread
-
Although "impossible to automate with ML" doesn't mean "impossible to automate". I'd say most things are still automatable -- by having human engineers handcraft complex software to handle each task. Which can be enhanced with ML, if need be
Show this thread
An example of this is Facebook discarding its M assistant project. The plan was to start with a Wizard of Oz, and collect tons of queries (with the steps to answer them), to eventually train a supervised model to answer the most common queries. Sounds clever, but it doesn't work
-
-
Another example is self-driving. You can automate driving, but not with an end-to-end ML system. The variability of real-world situations just cannot be handled by a system that merely pattern-matches with the past
Show this threadThanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Thank you for the insight. Consider writing an article.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Do you happen to know what the particular failure mode was for M? I imagine their approach would have never led to proper state tracking but optimistically, you could still have hoped for M to be useful à la Amazon Alexa.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.