Tuesday, May 28, 2019

Some Amusing Musings About AI

Let us begin with a quote from Yogi Berra:

“It's tough to make predictions, especially about the future.”

Yes, round and round and round we go, and where we stop, nobody knows. The contentious AI discussions on Aunt Minnie continue ad nauseum.

We've been having this very same discussion for YEARS now, with no real change, no resolution, no obvious final answer. And every time Google or some other NON-physician comes up with an algorithm they CLAIM will outdo a radiologist in detecting something somewhere on someone's scan, the sky is immediately declared to be falling. I'm beginning to wonder if I'm caught up in some Groundhog Day remake, where the nightmare won't end until Andie MacDowell falls in love with me (don't tell Mrs. Dalai!) Alternatively, I wonder if I've been trapped in an HGTV marathon wherein every episode has the same 3-act play:
  • AI can identify a cat picture, so we're doomed
  • Self-driving cars kill people, who are just as dead as if they had been killed by a drunk driver but Progress!
  • You can't sue the computer...what would you get? Transistors? But the lawyers will find a way anyway.
I'll repeat what I have said from Day One of this discussion:
  • AI is a TOOL to be used to help radiologists help patients. This is the mantra of every HEALTHCARE company in that space. Google, by the way, is NOT a healthcare company.
  • We radiologists need to be at the forefront of AI development, so it will do what we need it to do, not what some geek from Google tells us it needs to do.
  • AI will NOT replace radiologists in the working lifetime of anyone reading this.
The rest is commentary, gibberish, or both.

Trust me, children, there are things out there that have a far greater likelihood of disrupting your livelihood, such as corporate takeovers, intergroup squabbling, intragroup squabbling, plaintiff whores, bean-counters, rogue administrators, VC entrepreneurs, gossiping physicians, and others too numerous to list. AI is so far down that list as to be a non-issue.

I have begged you all to look at AI through the lens of how it could be used to do what you do better. That would be far more productive than this constant pity-party.

Think about it.

1 comment :

stacey said...

The current 737 Max MCAS system is enough to make anyone fear AI. But it's been going on much longer in aviation as well.
SAS flight 751 back in 1993. Some ice flew off the wing and got sucked into the engines right at take off. Pilot did the correct thing, reduced power,which would keep the engine from destroying itself,(Automatic Thrust Restoration) "However, the newly installed ATR { My add here: blissfully unaware of the REASON the pilot chose to throttle back}prevented the pilots from successfully performing the normal remedial measure to halt compressor stall, i.e., throttling back the engines, as the ATR system – designed to prevent pilots using less than normal thrust when climbing out after take-off for noise abatement reasons – restored engine take-off power throttle settings in contrary to the pilots' reduced throttle commands. This damaged the engines, until eventually, they failed completely.