That "Machines will be much better at estimating JIRA ticket resolution times than humans" is quite a take.
I mean if they could do that, wouldn't we just ask "AI" to in natural language to "make the service do X" and eliminate the entire ticketing process? Why even create tickets? Just submit voice commands directly.
Much of my EM time is spent understanding what's _not_ in the description or connecting that to other non-existent but potential work. Literally non-existent training data.
Mastodon Source 🐘
That we could exhaustively and precisely, without ambiguity, instruct a machine to produce a behavior via spoken language seems...like a solution which the margin is too small to contain.