Jobs That Artificial Intelligence Will Create
sloanreview.mit.edu
sloanreview.mit.edu
OK, great, but...
a.) these jobs will only be "uniquely human" until AI becomes self-generating and sustaining. [1]
b.) as these jobs emerge, surely they don't expect all of the millions of displaced laborers, truck drivers for example, to find work in these roles.
AI is increasing overall efficiency (less human work => more productivity), while global population grows. This is a good thing, but I can't buy into the idea that these new jobs will be sustainable. IMO #1, we instead simply need to conciously decide to carve out jobs for humans to do, even though the work could be done (probably better) by a machine. After all, I mow my grass with a push mower (vice a gas mower) because it's clean and I enjoy it. IMO #2, a base living wage without work does not a happy life make.
1. https://www.technologyreview.com/s/603381/ai-software-learns...
Makework doesn't cut it, though. You can hire a guy to dig a hole and another guy to fill it in, but both of them will be utterly demoralized. And I have trouble believing garbagemen wouldn't be happier staying home even though they know they're doing necessary work.
And for point b), it's always truck drivers brought up in these instances. Yes, think of the poor truck drivers. I'm sure they love doing a dull job.
If a job is replaceable by machines, it's not worth doing.
I get the worry that people will be without livelihood, but it's not as if this world has a shortage of work to be done. If we had no more work to do, sure, bring on the UBI. But we do. Clean litter out of forests, plant new forests, visit the sick. Whatever it is, there will absolutely be more work for humans in the future. People not being forced to drive trucks means they could be doing something else more useful.
And if people lament losing a particular job and having to pick up a new one, well, sorry but that's reality. I could just as soon lament not getting paid to doodle cartoons.
Who's going to pay people to do the things that are more useful, but don't fit in to the standard capitalist system? You? Some benevolent billionaire? The government, if we raise taxes and increase social spending to an extent that makes plenty of Americans recoil in instinctive horror? Yeah, I do think of the poor truck drivers, because I don't want them to starve.
UBI has to be worked out "before" these become big problems, not after; now is the time.
You're right, in theory we could have a system where we put truck drivers to work doing great stuff. Losing lots of jobs could be amazing if we had the services and programs that made our society compatible with that. But we don't, they will take a long time to establish, and the political climate appears to be reluctant to even start building them. Thus: a disaster.
There's room for more than one social program in dire circumstances, but ideally we should have a social structure which incentivizes doing good. Help those who can't help others, and encourage those who can to do so.
I do advocate at least modicum of empiricism and caution, an approach of testing ideas on city level before country.
I have met some truck drivers who DO love their job. They have spoken about the opportunity to travel (that from a husband-and-wife driver team who claimed to enjoy living on the road). They speak of the opportunity to own their own business (although, to be fair, independent truckers are becoming less common).
Real AI would create mass unemployment, 50% or more of the population unable to find someone to pay them enough to live on because robots will be doing all the jobs they're capable of. How are these people going to survive? That's the issue being discussed.
It's that there is a huge population that relies on that work, and if you haven't noticed, our culture is pretty terrible about offering alternatives.
So should this large swath of people suddenly be on their ass, or even not so suddenly but over time, with no alternative, shit will get real.
Your hard cold facts approach forgets humans are emotionally volatile, and as we're seeing, our social structures not as stable as we tend to believe.
You seem to believe that should the bottom fall out, your cold hard facts about reality will spare you the wrath of a hungry populace.
Yes, they're clearly doing that for the personal achievement that it brings them and not for, maybe, money. That tiny thing we all kind of need.
Jesus Christ, the arrogance of HN sometimes.
It's not about truck drivers. If your job consists entirely of stuff that goes in and out over a wire, expect to be replaced by computers.
If you are a software engineer, and don't understand this - then you don't understand software engineering.
The whole point of software engineering is to automate everything - including software engineering.
What happens afterward...? Who knows...
Beats not doing any job at all. Especially given the shape of society that we're going to have for the foreseeable future.
Another candidate is environmental management, like environmental reviews, construction permitting, and pollution remediation. Again, there is almost no limit to the complexity you'll encounter when you want to model geological systems to make detailed predictions.
Technology usually leads to job creation when it creates new levels of abstraction upon which new careers will be built. These are most needed in areas where complexity overwhelms current technology. Biology and geology currently fit that bill.
It's possible that using AI to tune therapy techniques will thrive, like pacemakers and blood glucose monitoring. But diagnosis will take considerably longer to catch on, since the responsible party for making medical decisions can't be mere software.
I work at a pharma, and from what I've seen, the use of clinical data is often complicated and unrewarding unless your question is very precise and worth the extra cost/effort that all that regulation requires. Asking or acting on more general questions like those amenable to AI would be infeasable due to all the constraints against gathering a sufficiently critical mass of general personal data.
https://www.google.com/#safe=strict&q=http://sloanreview.mit...
Automation really got going at the start of the industrial revolution and people have been scared of it ever since.
Today I put together some Google Form that writes to a spreadsheet. This means no more excel spreadsheet email attachments or copying and pasting. It took me an hour to research and prototype. With a few adjustments all was done. In this task there was no A.I. to teach, just a few boxes to move around and ability to listen to requirements.
I handed over the form to be maintained and updated by internal client. So no more work needed. Someone normally spends weeks collating the spreadsheet data. Now they don't have to do that. Now they will do cool stuff in Fusion tables, interpreting the data. Plus the company is growing and therefore that role will not be an idle one.
Whole teams of people can be swept away from back office jobs from the simple non AI changes such as better forms. Or people can be empowered to move their job on to the next level, e.g. data analysis rather than data entry.
I know computers can do amazing A. I. things but the much feared change is already happening and it is mostly based on if statements rather than A.I. Some get swept away and others embrace and effect the change. Attitude of the individual matters and there are winners plus losers with change brought on by automation.
I 100% agree. Modern business still organize themselves like firms a century ago but with faster tools to accomplish largely the same set of tasks. We have only begun to realize the productivity gains/decreased labor demand of technology that was boring in the nineties.
- "trainers" to train the AIs,
- "explainers" to investigate the "black box" and get insights and knowledge out of it, and
- "sustainers" to monitor and make sure that the AI behaves appropriately (think, e.g., Microsoft's Twitter bot-turned-Nazi).
Yes, but for that to be economically a net win, way fewer man-hours will be spent training systems than now are spent by humans doing customer service. Net result will be a significant job loss.
Won't be remotely enough to replace the jobs it destroys leaving us with mass unemployment.