In some sense you must be right, or it wouldn’t be getting so much more involved.
If we ever develop strong AI, it will probably be in the fight between spammers and anti-spammers.
AI has already mastered picking titles for romance novels:
LOL…“The Surgeon’s Baby Surgeon”? I think it has a ways to go.
The Atlantic (my favorite magazine) has a new podcast called Radio Atlantic. I recommend it in general, but the third episode, "Ask Not What Your Robots Can Do For You"is what’s relevant to this thread. In fact, they essentially examine the exact same questions we have, and it’s a lively and interesting roundtable discussion. I was particularly intrigued by what the (human) Go champion said after he was wiped about by a Go AI.
Have you seen romance novel titles?
Sounds like a best seller to me.
And the world ends when the anti-spam AIs realize that total nuclear war is the only way to stop spammers.
LOL…I wish I could absolutely 100% dismiss that as a joke.
At least it’ll stop the spammers, right?
There is that.
Spammers are technically a species of cockroach, so probably not.
NPR is tackling this issue this week in a series they call “Is My Job Safe?” Apparently in 2004 MIT researchers predicted truck driving would never be automated? That seems rather myopic.
And then the one on radiologists is really interesting as a kind of psychological study. It seems quite obvious to me that they will be completely redundant within a decade or two–all the more so because they make an average now of $400,000, twice what a family doctor does? How did that state of affairs come about? Ridiculous. That provides an intense incentive for the coders to automate the practice. Yet while so many of them are properly worried, they quoted people who are whistling past the graveyard. If I had a child in medical school, I would advise them in the strongest possible terms to pick something else than radiology, that’s for sure.
At present, AI can do specific tasks and even sets of tasks pretty damn well, and this is true across all fields. AI is now being used by attorneys to predict the outcome of cases so that they can avoid the costs of taking a matter to court if the odds are against them (this could be good and bad). I don’t know much about radiology. It would seem we’d still want a radiologist who knows about radiology stuff for the sake of the patient’s safety. But in terms of administering and monitoring anesthetics, maybe that’s where AI comes in.
I think it’s more complicated than that when we’re talking about career strategy.
Radiologist is still a pretty sweet gig, and with how risk-averse the medical industry is, it may still make sense to choose that role.
If you need to retrain in 10 years…well, you’re probably in the same boat as a huge proportion of the workforce, except you also have $4 million in your pocket (if you didn’t eat within those 10 years).
Pretty good update on this kind of thing at VICE today: Robots Are Coming for Your Job Sooner Than You Think.
The thing that really caught my eye is that some people see what I (and a lot of y’all) see and are working on addressing the coming problem now rather than waiting until it’s too late.
Here are some numbers on job loss that has already happened:
Here is their cite for that 6100 number: Business Insider; Retail jobs lost US jobs report May.
And here is a projection:
Here is their source for that.
And here is Mr. Solomon’s website: http://thedayafterlabor.com
Great stuff! Thanks for posting this.
I remember not being sure we were ever again going to get unemployment as low as it is right now. Although I suppose it might stay low while the labor force participation rate continues to drop. Maybe the real question is whether that percentage will ever be as high as it was 10 or 20 years ago.
So in June we had ~6k fewer “retail trade” (as BLS likes to label them) jobs than a year before. But then in July the number of retail trade jobs went up by over 7k vs June. This is out of over 15 million retail workers. Jobs numbers fluctuate. Meanwhile in August we had over 2 million more non-farm jobs overall than we did a year ago.
Is there data on retail job numbers by year for the past 20 years? Can we compare numbers that way?
Any time someone links to jobs numbers they’re from BLS. So yes. BLS CES4200000001 goes back to 1939. That’s for seasonably adjusted numbers. CEU4200000001 if you want the raw numbers.
Ok; thanks. I had no idea how to find anything, but I put your CES and CEU numbers in Google and they turned up right at the top of the page.
I will not be surprised or unwelcoming if someone corrects my suppositions and/or math here; this isn’t exactly my area of expertise.
I’m using these numbers, which I got by googling for BLS CES4200000001.
In 1997 there were, for the year, an average of about 14,366 retail workers.
In 2016 there were, for the year, an average of about 15,825 retail workers.
In 1997 there were about 272,000,000 people in the US.
In 2016 there were about 323,000,000 people in the US.
That’s a little under a 20% increase in population, but only about a 10% increase in the number of retail jobs. Shouldn’t we also expect to see a corresponding increase in retail job numbers as population increases?