Competing against robots

Some scholars are trying to discern what kinds of learning have survived technological replacement better than others. Richard J. Murnane and Frank Levy in their book “The New Division of Labor” (Princeton, 2004) studied occupations that expanded during the information revolution of the recent past. They included jobs like service manager at an auto dealership, as opposed to jobs that have declined, like telephone operator.
 
The successful occupations, by this measure, shared certain characteristics: People who practiced them needed complex communication skills and expert knowledge. Such skills included an ability to convey “not just information but a particular interpretation of information.” They said that expert knowledge was broad, deep and practical, allowing the solution of “uncharted problems.”
 
These attributes may not be as beneficial in the future. But the study certainly suggests that a college education needs to be broad and general, and not defined primarily by the traditional structure of separate departments staffed by professors who want, most of all, to be at the forefront of their own narrow disciplines. But this old departmental structure is still fundamental at universities, and it is hard to change.
 

Full article here from Robert Shiller.

A few random thoughts. Disciplines which are purely about knowledge accumulation are risky if the type of knowledge acquired is that which computers can accumulate in a fraction of the time. Lots of Ph.D's seem unlikely to be economically worthwhile considering the cost of higher education.

Watch the virtual assistant on your phone. Siri or Google Now are good benchmarks for what skills are becoming obsolete, and which are still of great value.

Most humans still prefer a bit of entropy and warmth from those they interact with, especially in the service sector, and indexing high on that still commands a premium.

A cookbook from IBM's Watson

Robots taking all the jobs, cooking edition:

Steve Abrams, the director of IBM’s Watson Life research program, told Quartz that Watson scanned publicly available data sources to build up a vast library of information on recipes, the chemical compounds in food, and common pairings. (For any budding gastronomers out there, Abrams said Wikia was a surprisingly useful source.) Knowledge that might’ve taken a lifetime for a Michelin-starred chef to attain can now be accessed instantly from your tablet.
 
What separates Watson from the average computer (or chef) is its ability to find patterns in vast amounts of data. It’s essentially able figure out, through sheer repetition, what combinations of compounds and cuisines work together. This leads to unusual pairings, like Waton’s apple kebab dish, which has some odd ingredients: “Strawberries and mushrooms share a lot of flavor compounds,” Abrams said. “It turns out they go quite well together.”
 

The researchers are publishing a cookbook with recipe ideas from Watson, and it releases this Tuesday: Cognitive Cooking with Chef Watson: Recipes for Innovation from IBM & the Institute of Culinary Education. I have not read the book, but some of the recipes sound intriguing (“Belgian bacon pudding, a desert containing dried porcini mushrooms”) while others sound, at best, like clever wordplay (“the shrimp cocktail, which is a beverage with actual shrimp in it”). Regardless, I'm purchasing a copy just out of sheer curiosity. Let's hope they turn this resource into an app or service instead of a book, I blame Watson's vanity for wanting this in the outdated format of a book.

To the extent that standout recipes and flavor pairings are a matter of pattern recognition, there's no reason a computer, with its infinitely more scalable hardware and software for that purpose, couldn't match or exceed a human. And, so, a variant of the infinite monkey theorem: given enough time, a computer will write the French Laundry cookbook (and win a third Michelin star).

To be clear, I'm okay with this. I just want to eat tasty food, I'm fine with employing computers to come up with more amazing things to feed me.

For now, however, the computer still requires a human to actually prepare the recipe. In a true demonstration of how far artificial intelligence has progressed, no sufficiently advanced computer wants the drudgery of life as a line chef. Better profits in cookbooks than restaurants anyway.

A new cooking show concept already comes to mind: Top Freestyle Chef. Like freestyle chess, in freestyle cooking competitors would consist of a human or a human consulting with a computer. I am ready to program this into my DVR already, as long as they don't replace Padma Lakshmi with a robot host. I'm as big a fan of artificial intelligence and robots as the next guy, but I think we're a long way from replacing this.

Robots taking all the jobs, cont.

By studying the brains of drivers when they were negotiating a race-track, the scientists were intrigued to find that during the most complex tasks, the experts used less brain power. They appeared to be acting on instinct and muscle memory rather than using judgement as a computer programme would. 

“It looks as if the skilled race car drivers are able to control their cars with very little cognitive load,” said Prof Gerdes. 

Mr Vodden agreed saying in difficult manouvres experience kicked in. "If you're thinking you're going too slow."

You'd think from that excerpt that the human driver remains superior, but it turns out the driverless car beat the track champion by 0.4 seconds on a track in Northern California.

One race track, the worlds' greatest driver (whoever is the Michael Schumacher of the moment) versus the best computer driver. I don't enjoy watching auto racing on TV, but I'd watch one that pits man and machine against machine and machine.

One more wrinkle for AI to learn: how and when to cheat.

In the race between Shelley and Mr Vodden, the racing driver left the track at a sharp corner, rejoining the race ahead of the robot car. 

“What we’re doing as humans we’re weighting a number of different things,” added Prof Gerdes. 

“We’re not driving within the lines, we’re balancing our desire to follow the law with other things such as our desire for mobility and safety. 

“If we really want to get to the point where we can have a car that will drive as well as the very best drivers with the car control skills and also the judgment it seems to me that we really need to have a societal discussion about what are the different priorities we place on mobility and safety on judgement and following the law.”

Robot handwriting

When humans write by hand, our margins are — you guessed it — irregular. On the left-hand side of a paragraph, we try to keep a nice straight margin, beginning each line in the same location. But we usually can’t keep it very precise. We drift inwards, producing a left margin that slowly slopes towards the center of the page. Here’s an example of a human-written letter Maillift composed to send to customers:

See how the left-hand side drifts inwards? The lines beginning “choose”, “color” and “logo” all move to the right.

But the activity on the right-hand margin is even weirder.

Curliss and Jurek have noticed that their human handwriters often produce a “rounded” right margin: It bulges outwards and then tucks back inwards. In the example above, “showcase” sticks out further than the line above, “option”. But then the lines begin rounding inwards, with “size” and “own” moving further and further into the center of the page.

Why would humans do this? Probably because we aren’t terribly good at judging how many words fit on each line. If we accidentally overpack a line — trying to cram too many words into it — we immediately overcompensate by becoming too cautious, and putting too few words on the next couple of lines. You can see that dynamic at work in the example above. The composer wrote a nicely-kerned first line, but in the next line crammed in too many words, such that “showcase” comes uncomfortably close to the edge of the page. So he or she pulled back sharply, and the next three lines shrink in length.

A robot, in contrast, knows precisely how much space each word takes up, and doesn’t make these mistakes. You won’t see that rounded margin on the right.

Until, of course, robots are programmed to mimic that, too.

From a fascinating Clive Thompson post on the recent adoption of robots to generate hand-written letters to try to improve response rates on marketing offers. As Thompson notes, it's actually the exception, not the rule, to receive hand-written communications nowadays. A hand-written note actually may be the clearest sign that a letter is just marketing collateral.

I don't have kids; do they even teach handwriting in school anymore?

Optimal robot personality

They gave them distinct personalities. It was an experiment: would humans react to the robots differently based on how they carried themselves? They tried out two different personalities on each robot. One version was extraverted; the robot would speak loudly and quickly, use more animated hand gestures, and start conversations instead of waiting to be spoken to. The other personality was more reserved, speaking much more slowly and quietly, moving around less, and letting the user initiate communication.

What the researchers found, as they described in a recently published paper, was a striking difference between the two. When it came to the nurse robot, people preferred and trusted it more when its personality was outgoing and assertive. What people wanted in a security guard was exactly the opposite: the livelier, extraverted version clearly rubbed people the wrong way. Not only were they less confident in its abilities, and dubious that it would keep them away from danger, they simply liked it less overall.

...

What researchers are finding is that it’s not enough for a machine to have an agreeable personality—it needs the right personality. A robot designed to serve as a motivational exercise coach, for instance, might benefit from being more intense than a teacher-robot that plays chess with kids. A museum tour guide robot might need to be less indulgent than a personal assistant robot that’s supposed to help out around the house.

A growing body of research is starting to reveal what works and what doesn’t. And although building truly human-like robots will probably remain technologically impossible for a long time to come, researchers say that imbuing machines with personalities we can understand doesn’t require them to be “human-like” at all. To hear them describe the future is to imagine a world—one coming soon—in which we interact and even form long-term relationships with socially gifted devices that are designed to communicate with us on our terms. And what the ideal machine personalities turn out to be may expose needs and prejudices that we’re not even aware we have.
 

More here. How many of our personality preferences for robots will we inherit from the human analogues we're most familiar with? We may wish for a robot personal trainer to be tough, forceful, while we may prefer a calm, almost flat affect from our robot therapist.

Regardless, I'm excited to see the first generation of robots or AI's with personality roll out to the world. It feels like one of the most likely vectors of delight in user experience design when it comes to AI.

More lessons in human-computer team-ups

"I don't know if you know anything about Six Sigma," Coombs asked rhetorically. "a human being is at best a 2-sigma machine. Which means that humans get things right 92 to 93 percent of the time."

 

From Alexis Madrigal's piece on telemarketing “robots.” It turns out the robotic Samantha West who telephoned a Time reporter was a telemarketer choosing pre-recorded audio clips from a sound board using call assist software that is increasingly popular in the industry.

Madrigal raises two other great points in his investigation of this industry that Samantha West brought into the limelight.

"The impact on the people was really dramatic. It was one of the things we didn't expect," Bills told me. "In outbound sales, it knocked our turnover from 400 percent a year to 135 to 140 percent. And it dramatically changed the characteristics of employing people."

To be clear, 140 percent turnover is about on par with the fast-food industry. The paragons of employee retention keep their numbers in the single digits. These are still hard jobs. 

But maybe this technology makes it a little bit easier. 

"It creates detachment," Bills said. "What we see is that our employees, when they have a successful outcome of the call, they take pride in operating the system effectively. When it doesn't work, they say, 'Ahhh that wasn't me.' It doesn't beat people up in the same way."

The machine absorbs some of the "emotional wear and tear" that comes with the job. CallAssistant can even employ people full time because the "shift fatigue" that hits other outbound telemarketing firms doesn't set in in quite the same way. 

* * *

Though no one quite puts it this way, the number-one selling point for the soundboard technology is obvious to Filipino telemarketers: Americans' xenophobia. We want to hear from people who sound just like us. 

 

Anyone who's ever worked in telemarketing (I've made fundraising calls in college and for the Obama campaign) knows you already work off a script, so it's entirely shocking that the industry would transition to using pre-recorded audio clips.

One thing I thought when watching the Spike Jonze movie Her is how humans have a finite amount of love and attention to give and how computers can increase the supply in the world by a near infinite degree. For humans to accept it, though, we have to shift our conception of the definition of love. How much do you value someone's love because you know it's finite and they've chosen to give that precious resource to you? It sounds selfish but it may be wired in our DNA. [SPOILER ALERT for those of who haven't seen Her, skip the rest of this paragraph]. When Joaquin Phoenix's Theodore finds out his operating system Samantha is carrying on love affairs with hundreds of others, he doesn't rejoice at the amazing leverage and increased supply of love in the world, he reacts like a jealous lover, to no human's surprise. An economist might be disheartened, though.

I'm reminded of Joe Pantoliano in The Matrix: “You know, I know this steak doesn't exist. I know that when I put it in my mouth, the Matrix is telling my brain that it is juicy, and delicious. After nine years, you know what I realise? Ignorance is bliss.”

Comedic variant of the Turing test?

Someone has developed a robot stand-up comedian (h/t Marginal Revolution).

Katevas developed an algorithm for comic timing: tell a joke, wait two seconds to measure audio feedback from the crowd, and pause for laughter, holding for no more than five seconds. If the audience responds positively, encourage them; if not, RoboThespian might say “Hmm” or “Take your time.”

RoboThespian was also embedded with software called SHORE (Sophisticated High-speed Object Recognition Engine) to detect faces in the audience and identify their expressions. The program lets him know whether the crowd is enjoying themselves. If not, RoboThespian could look at them, point, and tell a joke at their expense. “If the whole show is bombing and everything is going terribly wrong,” Jackson said. “Should the robot change course, or should it just keep going like a dumb machine?”

Comedy is an art of precision. “The difference between an amateur and a professional is that it feels off the cuff, but it’s something I’ve worked very hard on,” the comedian Rob Delaney, the author of an eponymous new book, told me. “I have a narrative arc that I want to adhere to. Sure, I’ll make changes, but it’ll be eighty-per-cent similar.” He added, “I do a thing that a robot could do, which is: I listen to the room. That, I think, could be learned.”

Below is a YouTube video of RoboThespian performing live at a comedy club.

Okay, let's be honest (the robot shouldn't have any hard feelings, right?), Louis C.K. has less to worry about from RoboThespian than Gary Kasparov or Ken Jennings did from Big Blue and Watson.

Still, how and why the robot falls short is fascinating and instructive as to both the art of comedy and what it means to be human. A couple observations:

  • The vulnerability of the comedian is often critical to a joke. Since a robot can't really empathize with human emotions, it's difficult for us to buy that the robot really understands the pain of human situations he might discuss in a joke.
  • I still felt uncomfortable for the robot when some of his jokes fell flat. Maybe I was projecting my empathy for the programmer onto the robot? Perhaps a robot comedian can only be successful if it can first establish a persona or believable personal history. Maybe that can be as simple as making light of how badly he had bombed early in his career?
  • As outlined above, a big hurdle for robots which also applies in comedy is the ability to read other humans. What if all the humans in the crowd were fitted with bio-sensors that fed data up to the robot in real-time?
  • It might be easier to build a credible cartoon or animated comedian than a robot comedian. The stiff movements of the robot, its severely limited facial expression, and its lack of vocal inflection seem to leave it best suited to deliver deadpan jokes. It would also be helpful if those deadpan jokes were either really intelligent or naive. A robot of average intelligence is not interesting. Maybe feed it from the joke library of Mitch Hedberg?

The impact of robots on the household corporation

As feminist economists have long pointed out, households are factories in function and corporations in identity. They are factories because they apply human labour and tools to convert inputs like groceries, nappies, houses, etc. into things worth having, like meals, children, homes, etc. They are corporations because they are unified economic units, separated from the individualistic competitive market that operates outside its walls. That is, the individuals who make up a household, like the employees of any firm, are supposed to work together as colleagues to advance the success and prosperity of the corporate 'family' as a whole, rather than to advance their own individual material interests as actors in a market would.

Organising production inside the household - outside of 'the market' - makes economic sense in many circumstances, which is also why we have business firms. Using the market comes with significant transaction costs associated with establishing trust and quality assurance between self-regarding strangers. For lots of household work - like washing the dinner dishes - the costs of contracting with someone else to do it are so high that even though you have much more productive things to do with your time you are still better off doing it yourself.

More significantly, in addition to minimising transaction costs, corporate structures permit positive efficiency gains from coöperation. In particular, many projects - child-rearing for example, or soccer matches - are most economically achieved by team-work. A team works together on many-hands problems and thereby achieves much more than the same number of individuals operating by and for themselves could. One can't organise team-work through the market because it is impossible to identify and directly reward the marginal contribution of each worker to the final outcome (whether producing thriving children or winning a soccer match). The corollary of this is that team-work requires not only suspending the individualistic 'homo economicus' logic of the market, but also inculcating an ethic of self-abnegating commitment in which individuals adopt the common good or goals of their 'family' as their own, and do not shirk the sacrifices it requires of them. There are different psychological routes to establishing this disposition to self-less coöperation, including viewing the work itself as sacred or feeling bound by honour not to let down one's co-workers. But in the family it is generally achieved through love.

From a long and fascinating piece speculating on the impact of robots on family by the always interesting Thomas Rodham Wells.

It strikes me that we already have some directional test cases in the potential impact of robots on the family household because many wealthy people already outsource a lot of their childcare and home care to other people. This article observes the same but foresees a surprising forecast for the impact of decreased reliance on each other:

The arrival of cheap robot-servants will revolutionise the political economy of households. We will be able to produce consumption goods like meals and child-care much more efficiently since the number of human hours involved will be much less. That means the standard 'team' of two adults will no longer be required. There may not seem anything fundamentally new about this, since machines have been replacing human labour inside the home for a 100 years (e.g. washing machines). Such technologies have supported the social emancipation of women: less household work to do means more opportunity for higher status paid external employment. But they have also permitted the rapidly increasing number of single adult households. It turns out that when people can afford not to be mutually dependent on another person, not to have to love another, fewer of us do so.

Look, I've seen Wall-E, I know how this ends, with all of us as obese hedonists, fattened on a life of overflowing leisure, all physical drudgery having been offloaded onto an army of robots.

I suspect we're still a ways off from feeling a massive overabundance of leisure due to robot labor, though. Until my Roomba figures out how to cook me dinner, wash the dishes, and do my laundry, I have enough chores to fortify my moral fiber and keep me loyal to the household corporation.