CGP Grey recently put out a video along very similar lines. Worth 15 mins: https://www.youtube.com/watch?v=7Pq-S557XQU&list=UU2C_jShtL7...
I'm so confused why everyone thinks the least expensive to purchase(low skill labor) will be replaced first when it requires the most expensive version of AI(robots).
Some more insight on this is research into why Wal-Marts did not cause as much employment loss as feared, but instead removed "community leaders" from the area. Because Wal-Mart did all of it's professional services back at corporate headquarters, the area it moved into lost all of those people. The first line of true AI will be very similar to a WalMart moving into town.
If we do look at this from an energy spending point of view.. this robot/AI revolution is gonna have a few interesting twists and turns along the way. One has to remember, the evolution of life has been a constant "economic" struggle for limited resources, with energy being the most limiting factor. Right now our economic development is largely based on a temporary liberation from that. However, that won't last, and energy-wise, life is far more energy-efficient than machines for most tasks[1] (I'm not entirely sure how this translates to robotics and computers, but for now I doubt they win). Energy spending also scales sub-linearly for living things[2], and I'm not sure if that holds true for machines.
[0] http://en.wikipedia.org/wiki/Embodied_energy
[1] http://www.lowtechmagazine.com/2014/03/how-sustainable-is-di...
[2] http://peakenergy.blogspot.se/2011/07/geoffrey-b-west-why-ci...
I'm not promoting this; I'm just predicting it. The sustainability of fossil fuels will be reflected in prices. More demand, political restrictions, and fewer new discoveries: prices go up; conservation, carbon sequestration/removal tech, and growing efficiencies supplemented by new sources such as fracked gas: prices relax. At some point (you say "very near", I say no one knows), hydrocarbon prices will go so high that the developed world will either have to voluntarily undevelop (not going to happen politically) or pay what it costs to solve some nuke problems and live with others, and we will continue to grow our use of energy.
The spread of robotics will proceed apace.
> (you say "very near", I say no one knows)
Given that our energy spending grows exponentially, it's certainly going be sooner than later.
I really have to question this. Humans are notoriously inefficient at converting nutrition to usable energy.
The third link I posted references Geoffrey West who has an interesting high-level physics approach to this issue, but it doesn't really do justice to his points. But I figured posting this 40 minute one, which is much more comprehensive, would be a bit much to ask:
https://www.youtube.com/watch?v=DFFVSvAr7Wc
(also has a lot more phycisist's arrogance in the form of cursing, which I find entertaining myself)
Actually, given that city energy spending also grows sublinearly, robot and AI networks probably would as well. So that argument likely won't hold against it.
I'll be blunt and say that that is just bullshit for quite obvious reasons. Your [1] does not support the claim.
"A human-controlled punching machine is 25 to 35 times more efficient than a laser cutter when producing a standard part"
"Fibre laser cutters are more efficient than CO2-laser cutters, but their power consumption is still 10 times larger than that of a slower but equally powerful human-controlled nibbling machine"
... and has tables like:
http://krisdedecker.typepad.com/.a/6a00e0099229e8883301a73d8...
(which is per item, by the way)
... is not supporting the claim that manual labour is more energy efficient for many tasks.
There might be some manually controlled machines that are more energy efficient than some automatic machines, but that would be a very very different fact, and I don't see any reason to believe that it would hold in general. For example the table you linked to doesn't say how much that is produced per energy unit, which would be the relevant measure, and it doesn't take into account that the operator uses a lot of energy. With your kind of arguing I could show that a hand saw was infinitely more efficient than a chainsaw because the hand saw in itself doesn't use any energy at all and we don't count the operator.
A human in standby (resting) consumes about 80W power, and a human that does any kind of work much more than that. Humans use extremely inefficient fuel that takes many times more energy to produce than it makes available to do any kind of work. For most physical work it is easy to find a machine that does the same work using much less energy. In those areas where computers can replace humans they also do the same work using much less energy but also much faster. And then we haven't even taken into account that the human only works for 40 hours per week, need a house that keep her warm and uses a shitload of energy every day just to get back and forth to their workplace.
For example, within the first few minutes they show this graph: http://i.imgur.com/IDXYI9Y.png
But the U.S. Bureau of Labor Statistics shows this: http://i.imgur.com/BPNyjM8.png
They've excluded other industries in forming their basis and it's clearly cherry picking to fit a narrative. I like the idea of a fully-automated utopia as much as the next person, but jobs are simply not going to disappear - they're going to shift industries.
I see your objection that to consider agriculture's decline indicative of the future of labor in general could be considered cherry-picking, but I think the video at least attempts to respond to this charge, explaining why its author does not feel other industries will "pick up the slack" in the same way in the future as happened in the past for jobs lost to automation.
[For what it's worth, while enamored of the dream of leisure-society utopia, I am quite skeptical of the author's assessment of the current state of progress on various forms of automation (and thus, a fortiori, of the projected speed at which such automation will take off as ubiquitous in the near future). But I did want to defend them against the charge of not even considering the idea that jobs will simply shift to other industries.]
At best he could write about his ideal of an utopia or his vision of a dystopia.
The US has major political dogma against communism and socialism. The only way out of this is going to be with major socialist programs. The only way to avoid civil unrest over unemployment rates will be literally paying people off. Basic income is inevitable.
The issue is, how long will companies be allowed to hold on to control. What happens when we start automating out jobs with AI. What happens when even the CEO is replaced, and investors are just AIs representing a portfolio. I find it highly doubtful the public will allow billions of dollars and control of corporations to sit in the hands of AIs with the intent of profiteering.
So how long until companies are taken over by the government. You said what happens when companies possess more power than the government, but I don't know any corporations with 10 aircraft carriers in operation. Eminent Domain seizures will be the way the government retains control.
How long after basic income would it take to see grocery stores become seen as a public utility? Or housing construction? When everyone can afford a house, we can't afford not giving people houses.
It's going to be interesting to see how much of our traditional practice's remain.
In corporations?
I have a feeling that this is an issue is going to creep up on us quickly, and I worry that our governments have proven themselves completely incapable of driving the discussion and making the decisions that need to be made.
Self-service checkout – “What used to be 30 humans is now 1 human overseeing 30 self-service check outs” – At all the grocery stores I’ve been to in my area, there are still ~30 human-staffed checkout lanes and 4 – 6 self-service machines, which replaced maybe two rows of check outs and still employs a human for oversight. However, these self-service checkouts allow the stores to operate 24/7, because after 10:00 they are the only lanes open. Now the store is open longer so it can make more money, whereas before they closed at 10:00. More money for the store means perhaps a whole new store opening, employing more people than would have otherwise been employed without self-service lanes.
Self-driving cars – “They are not the future, they are here and they work” – For narrow definitions of "here" and "work". There's a big difference between being developed in a lab and driving up and down the coast in sunny Cali, and being deployed to millions of consumers around the globe in multitudinous climates. He lists some things self-driving cars do better than humans, but fails to mention that self-driving cars lack the complex decision making, cognition, and communication abilities that are required in every-day driving. Self-driving cars are low hanging fruit for roboticists because we have a set of rules governing transportation, and that’s prime material for automation. Take the rules, code them up, done. Problem is, the rules are broken constantly, and in a way that driving a million miles up and down the west coast won’t solve (How many miles exactly has the Google car driven in snow?). Self-driving cars will come, but as long as there are still human drivers and weather, they will not live up to the hype they are currently generating. Take it from someone who actually engineers these types of robots.
Law Bots – He claims that bots will replace the bulk of lawyering because they can sift through a million e-mails quickly. And what happens when the bot is done? The lawyer gets back to work. It sounds to me like the bot is making the lawyer more efficient. Instead of spending weeks going through e-mails, he can use a computer to help him, and do his job more efficiently. So maybe we need fewer lawyers? The only reason we needed more in the first place was because of the amount of e-mail and electronic documents created by computers. To say we shouldn’t use a computer to fix a problem caused by computers is beyond ridiculous.
Doctor bots – He says that knowing the reaction of every drug between every other drug is beyond the scope of human capability… sounds like a great job for robots then! Again, this is an instance of a robot augmenting the capabilities of a human, not replacing one. He goes on to say that the doctor bots will be able to read the latest research, understand it, and then go on to implement the knowledge. First, one has to wonder how the research is being generated if human doctors are being replaced by robots. Second, as long as we’re assuming an AI advanced enough to read a research paper, generate inferences and implications based on that research, and act on that new knowledge, we might as well assume that all disease has also been cured, eliminating the need for doctor bots in the first place.
Creative bots – Finally, and most egregiously. Creativity is right up there next to sentience in the totem pole of AI capabilities, and is well into the realm of science fiction. The (only) example he gives, Emily Howell as a source of machine creative music is a complete joke. Here’s what the inventor of the bot has to say about it:
> Professor Cope argues that Emily Howell’s music is still predominantly created by humans. “Computers are not separate things,” he said. “The computer is human-made. The program itself is human-made. The music in the database is human as well. There’s so much about this that is human. There’s just a lot more humans involved in making this than usual.”
Emily Howell is a program designed by a human to output a specific kind of music. It may be music that has not been written before by any human, but it is not creative. In the end, the music is the result of an algorithm, designed by a human. In my view, a creative machine is one that performs a task, on purpose, it was not programmed to perform.
So those are the big problems I have with what he showed. What is even more aggravating is what he left out. What about robots that do things humans can't do. What about a robot that can fit into a 1/2" pipe, or one that can fly (we call them drones), or one that can go into a nuclear reactor or the bottom of the sea. Those robots are taking away 0 jobs, and create hundreds (if not thousands) of jobs for those that design, engineer, manufacture, repair, sell, program, manage, and use those robots.
Finally, robots are great at replacing some tasks that can be automated, but not so great at others. We've had robot receptionists for a while now. They are so terrible, that having a human receptionist is now a competitive advantage for some businesses. I think the general point that this video misses, is that robots are terrible at interfacing with people. Any human-facing occupation, even if it has aspects that can be automated, will probably still exist no matter how pervasive robots become in our lives. When robots can interact with humans well, we're probably talking about a future that is too far out for us to accurately predict.
I think you may have misunderstood the purpose of the video. The video doesn't say, "hey, we need to turn off these computers because they'll make us obsolete."
He's arguing that we'll inevitably use computers to solve a problem created by computers, and that a lot of people will be unemployable as a result, and that's something we need to think about as a society.
The video doesn't suggest any concrete solutions at all. (Basic income/negative taxation seem like obvious candidates; walking back technology itself is probably not a good candidate, though some people will certainly try.)
Regarding the doctor bot portion, I found the claim that knowing the reactions of a cascade of drugs is beyond the scope of human capability to be an absolutely ludicrous position to take. That right there is a huge portion of internal medicine. Pharmacists, nurses, doctors and respiratory therapists, etc. all spend years learning these matrices before even entering practice. Maybe knowing the totality of all drug combinations is difficult, but it ignores the fact that most—if not all—drugs are prescribed under known, common patterns, often now with the assistance of programs to make sure they don't accidentally prescribe something that will interact poorly with other meds. Medical professionals end up specializing and work with only subsets of the matrix anyway, but I really doubt a robot will ever be able to replace a doctor in this area, let alone all of medicine. If recent history is a decent guide, robots and software will only augment a doctor's capability to work accurately and efficiently.
With the creativity bot, I also wondered about this. With creativity, a bot creating sound based on rules isn't a musical genius, let alone "creative." They're a performer. They'll have difficulty 1) creating new genres of music and 2) creating music that is popular or meaningful to people. Technical performance and creativity are two very, very different things. We tend to think of the rule breakers as the geniuses of creativity. Hard to do that when the act of telling a machine to break the rules is itself a rule. How boring.
Lastly, we've found that a robot that plays Jeopardy! against humans to be an interesting gimmick, but nobody would suggest a game show where robots continually compete against other robots. Who would find it entertaining?
I believe I read that stores don't encourage self-checkouts very much because having humans around significantly decreases shoplifting. This may be less necessary if we get better AIs for monitoring security cameras.
I've never read anything about self-driving cars being impossible in bad weather, only that no has done it yet. Machine vision systems are not as vulnerable to this btw and that has been improving in leaps and bounds over just the last 10 years.
Lawyers still searched through documents tediously before computers. It was probably even harder then, not easier.
It's at least theoretically possible create a robot that can diagnose and prescribe treatments for a patient just like a doctor. All the video claimed is that they can do statistics and search the relevant research a lot better. Which is definitely true, and see IBM's Watson.
Arguments about what "creativity" means are never useful or productive. But AI that generates music well already exists and is going to continuously get better.
The main reason for them is reducing workers necessary during slow sales periods (like overnights, or weekday mornings).
And anyway, the job wasn't replaced by a robot in this case. The job was replaced by a customer.
In Sweden there has been quite a big reduction in number of people working at the checkouts. And even if they are freed up to do other things in the stores, there has been a net reduction in number of employees.
Since the takeover is apparently accelerating, humans aren't significantly improving, and non-humans are, it's going to be a difficult race.
Also, let's not forget that we now have a human needing to supervise these 24/7 bots. Graveyard shift work in any industry regardless of pay is not a pretty thing to witness with really high turnover rates and the health ramifications for encouraging it are probably not positive in the aggregate. Of course, we could employ remote workers to oversee machines in their regular time, but this usually results in the loss of the on-site job and the creation of lower-value jobs in developing markets due to geographic limitations for one of the obvious end results.
But good criticisms overall about the robot revolution alarmists on many common arguments. I can't help but agree given I'm a strong AI skeptic given what I've learned about AI through research and the tendency of the AI community to overestimate itself on achieving human-like behaviors grossly for decades with no exception while so much progress (and more importantly for continued investments in modern economies, immediate value) has been achieved with weak AI. The comparison that's most accurate is that we achieved human flight, but nobody would say that an airplane is "superior" to a bird's capabilities. A Jetsons like future has been dreamed of in various capacities since human being realized they could make machines to do work, but the nature of work just adapts with all of these inventions.
Or maybe it's because they've had a federal law for 60 years that regulated open hours: https://en.wikipedia.org/wiki/Shopping_hours#Germany
Receptionists are an odd case, as telephone access to people has either been removed (use a web interface), or goes direct to where you want. Redesigning the interface.
Law bots will only change if the law changes to be more computer friendly.
The presenter isn't attacking robotics, but outlining the increases in capabilities.
I have concerns about this myself, but I don't find that video remotely convincing.
In the past, new technological changes lead to the demise of horses. There were no "new" jobs that horses could fulfill. So it will be with humans.
We didn't react to horse unemployment by finding new things for horses to do because, aside from the horse owners, we did not value the employment of horses and the horses did not create new jobs for themselves because... they are horses.
Without necessarily disagreeing with the conclusion, the analogy comes off as extremely silly.
I do in fact agree that we need to think about what society we are building as we approach machines that rival humans on all levels, so it's frustrating to see a video that pretends that the employment of horses is at all similar to the employment of humans.
Dark as it may seem, many people would say the same about most human consumers in this country. Looking at the most successful creations in the past century, most are made by corporations. While people have been spearheading those efforts so far, the whole novel creation process would be automated if and when it can be. (The "if" in this case is weather or not you believe in "the singularity")
> We do not as a society place value on the employment rate of horses.
.. but a society of horses sure would. We, as a society of humans, are worried about the future of human jobs. Our human society and social structure is in for one hell of a shock, but the world will move on.
> Our economy does not depend on horses purchasing goods and services.
And our economy barely relies on humans anymore. As long as there is something with money to spend, the economy doesn't really care. Look at the banking industry. Look at all these B2B partnerships.. who cares about humans in those relationships? Robots are getting better and better at deciding how to spend money.
Is the analogy really that silly? I love it because it helps wake people up to the reality we live in: we are the horses in this new age, growing ever more irrelevant. The world economy couldn't care less about humans.
This is getting depressing. I'm gonna go listen to "Dog Days are Over" by Florence and the Machine...
Do you have any idea how the banking industry and B2B operations work? People. Lots and lots of people. Way less automation than low-margin retail junk, which itself is way less automated than most people think it is.
You have it exactly backwards: It's not that the economy is run by robots that are not interested in you. Robots are not interested in the economy, but the economy is very interested in you -- money is just a way of keeping score.
The issue is not that there will not be new jobs, the issue is that the automation of these new jobs is accelerating - so people cannot adapt fast enough to be able to competently take these new jobs as their old jobs become obsolete.
So then, yes, he listed a lot of jobs that will go away, insinuated it's basically all of them (with which I agree, but no argument was made), and then took the strawman "well maybe you think creative jobs won't go away" and then defeated that strawman by incorrectly assuming creativity is limited to artists, poets, directors, and actors and said "there can't be such a thing as a poem and painting based economy", which I think is what you are referring to. That's just as dishonest as the horse analogy. It assumes without argument that creative people won't look at the new robotic world around them and find new things to create, or at the humans around them eager to work and find new ways to utilize them, all with the assistance of AIs and robots.
Over a very long period of time, animals developed cars. Now that same development process is remarkably faster, and robots are being developed. The horse analogy is perfectly valid.
Just as cars replaced horses for the benefit of human masters, robots may replace humans for the benefit of corporate masters.
You may still disagree with the analogy but I think this is at least the proper analogy to be arguing about.
No one flagged it.