Gahh,
NLP AI conference.
mah head hurtz ...
Your head will soon be replaced by NLP AI. Pointy-haired bosses won't be able to tell its output is rubbish, and will resort to "because the computer says so".
Read comp.risks to see how such things appear to "work" for spurious "reasons", e.g. https://catless.ncl.ac.uk/Risks/32/80/#subj4.1
e.g.
"Some AIs were found to be picking up on the *text font* that certain
hospitals used to label the scans. As a result, fonts from hospitals with
more serious caseloads became predictors of covid risk."
I was first aware of that in the early 80s, when Igor Alexander's WISARD correctly distinguished between cars and tanks in the lab, but failed dismally when taken to Luneberger Heath in north Germany. Eventually they worked out the training set was tanks under grey skys and car adverts under sunny skies.
The problem with AI is that it's not artificial intelligence. It's machine intelligence. Assuming that it learns and thinks like (or trying to make it emulate) us leads to all manner of problems.
People and machines excel at different things and will do so for quite some time.
I’m not going to use the term AI because that’s just a lie. It’s not artificial and it’s not intelligence.
I always like to point out that actual meat shaped intelligence usually can’t resolve all facets of an algorithmic problem or any decision making process so why should an ML model be able to do it with any level of reliability or determinism that doesn’t put your business at mortal risk.
Ha! Just saw this. Yep, and error-prone people are creating the ML and feeding the machine. As often quoted in this thread, "What could possibly go wrong?"
Of course, not all machine intelligence is terrible. Just use the right thing for the right task and don't get carried away believing it'll solve everything. Right tool for the job...blah, blah, blah.