|Bambi the lobster mermaid on Coney Island will make sense by the end of the blog.
“Lobster” by Angus McIntyre, Creative Commons Attribution Noncommercial 3.0 Unported License
Yesterday I read a tweet from the MS Society UK that said “80% of people with MS are forced to give up work within 15 years of diagnosis. That needs to change…” The intention of the tweet seemed in part to be to raise awareness of the need for those diagnosed with MS to immediately go on and stay on disease modifying drugs, but the immediate effect on me was to do some math on the green paper towel sitting on my desk. It appeared that I had an 80% chance of being out of work by the end of 2027, which was a little earlier than I had expected retirement.
Whether it’s rational or not, I don’t really think I’ll be forced to do anything because of MS. I’ve been on disease modifying drugs since diagnosis—putting aside the pesky question of when I should have been diagnosed—and the only symptoms I’ve experienced during that time have been a few weeks of what at worst could be called a nuisance. Certainly I’ve not experienced disability. That, however didn’t stop me from noticing an article about Japanese robotics that could help disabled people that kept popping up in my social media feeds this week. I never clicked through; the knowledge that someone was working on it should I ever need it was comfort enough.
What I did click through on today is a piece by Gary Marcus about the current state of artificial intelligence, including the big news this week that DeepMind beat the European champion in Go, “a game that has been notoriously difficult for machines.” I started reading it thinking nothing about MS, spurred on only by my recently acquired interest in AI, nurtured by encounters with terrific film and literature on the subject. (My late-blooming interest in AI is a matter of much derision on the part of my lifelong sci-fi-loving husband.) But when I got to the part where Marcus recounts a talk by a graduate student of a deep learning expert on the same day the Go paper went public, I immediately felt a pang familiar to any patient of neurological disease who quickly learns the answers to most of her questions are “We don’t know.” Speaking about AI, the graduate student acknowledged “(a) people in the field still don’t really understand why their models work as well as they do and (b) they still can’t really guarantee much of anything if you test them in circumstances that differ significantly from the circumstances on which they were trained.”
In other words, AI sounds a lot like the human brain. And as with the human brain, it’s inevitable (as had been the fodder for the conflicts in countless sci-fi plot) that AI will be unleashed on the world before we ever understand it. It’s already been happening for a while with products like Siri and Google Now.
I’m less concerned about this than what Marcus poses at the end of his article: “The real question is whether the technology developed there can be taken out of the game world and into the real world. IBM has struggled to make compelling products out of DeepBlue (the chess champion) and Watson (the Jeopardy champion).”
As a person with a neurological disease that has the potential to impact both my motor and cognitive function, the one that’s more terrifying is the latter. Sure, not being able to walk is a horrifying prospect, but people are working on powerful exoskeletons I’ll soon be able to suit up in to handle that; I want mine to look like a lobster, just for the record. (Note to self: business idea for novelty human exoskeletons.) If AI scientists want to deliver a compelling product, how about one that will supplement cognitive skills in patients with neurological impairment? All I ask is she’s given a better name than Siri.