A little over two years ago, I wrote an article called Male Chauvinist Machines. At the time, men outnumbered women in artificial intelligence development roles by about eight to one. A more recent report suggests the ratio is now about three to one.
The problem is not just that men outnumber women. Data mining also presents an issue. If machines mine data from the past (what other data is there?), they may well learn to mimic biases from the past. Amazon, for instance, recently found that its AI recruiting system was biased against women. The system mined data from previous hires and learned that resumés with the word “woman” or “women” were less likely to be selected. Assuming that this was the “correct” decision, the system replicated it.
Might men create artificial intelligence systems that encode and perpetuate male chauvinism? It’s possible. It’s also possible that the emergence of AI will mean the “end of men” in high skill, cognitively demanding jobs.
That’s the upshot of a working paper recently published by the National Bureau of Economic Research (NBER) titled, “The ‘End of Men’ and Rise of Women In The High-Skilled Labor Market”.
The paper documents a shift in hiring in the United States since 1980. During that time the probability that a college-educated man would be employed in a
“… cognitive/high wage occupation has fallen. This contrasts starkly with the experience for college-educated women: their probability of working in these occupations rose.”
The shift is not because all the newly created high salary, cognitively demanding jobs are in traditionally female industries. Rather, the shift is “….accounted for by a disproportionate increase in the female share of employment in essentially all good jobs.” There seems to be a pronounced female bias in hiring for cognitive/high wage positions — also known as “good jobs”.
Why would that be? The researchers consider that “…women have a comparative advantage in tasks requiring social and interpersonal skills….” So, if industry is hiring more women into cognitive/high-wage jobs, it may indicate that such jobs are increasingly requiring social skills, not solely technical skills. The researchers specifically state that:
“… our hypothesis is that the importance of social skills has become greater within high-wage/cognitive occupations relative to other occupations and that this … increase[s] the demand for women relative to men in good jobs.”
The authors then present 61 pages on hiring trends, shifting skills, job content requirements, and so on. Let’s just assume for a moment that the authors are correct – that there is indeed a fundamental shift in the good jobs market and an increasing demand for social and interpersonal skills. What does that bode for the future?
We might want to differentiate here between “hard skills” and “soft skills” – the difference, say, between physics and sociology. The job market perceives men to be better at hard skills and women to be better at soft skills. Whether these differences are real or merely perceived is a worthy debate – but the impact on industry hiring patterns is hard to miss.
How will artificial intelligence affect the content of high-wage/cognitive occupations? It’s a fair bet that AI systems will displace hard skills long before they touch soft skills. AI can consume data and detect patterns far more skillfully than humans can. Any process that is algorithmic – including disease diagnosis – is subject to AI displacement. On the other hand, AI is not so good at empathy and emotional support.
If AI is better at hard skills than soft skills, then it will disproportionately displace men in good jobs. Women, by comparison, should find increased demand (proportionately and absolutely) for their skills. This doesn’t prove that the future is female. But the future of good jobs may be.
What’s harder: physics or sociology?
We tend to lionize physicists. They’re the people who send spacecraft to faraway places, search for extraterrestrial life, and create modern wonders like virtual reality goggles. In short, many of us have physics envy.
On the other hand, we tend to make fun of sociologists. What they’re doing seems to be nothing more than fancified common sense. I once heard a derisive definition of a sociologist: “He’s the guy who needs a $100K federal grant just to find the local bookie.” Few of us have sociology envy.
But is physics really harder than sociology? I thought about this question as I listened to an episode of 99% Invisible, one of my favorite podcasts. The episode, titled “Built to Burn” is all about forest fires and how we respond to them.
We take a fairly simple approach to forest fires: we try to put them out. But this leads to unintended consequences. If we successfully fight fires, the forest becomes thicker. The next fire becomes more intense and more difficult to stop. As one forest ranger puts it: “A fire put out is a fire put off.”
A general question is: Why do we need to put out forest fires? The specific answer is that we need to protect homes and properties and buildings. We assume that we have to stop the fire to protect the property.
But do we? Enter Jack Cohen, a research scientist for the Forest Service. Cohen has studied forest fires intensively. He has even set a few of his own. His conclusion: we can separate the idea of stopping wildfires from the goal of protecting property.
Cohen’s basic idea is a home ignition zone that stretches about 100 feet in all directions around a house. By spacing trees, planting fire resistant crops, and modifying the home itself (no wood roofs), we can protect homes while letting nature takes its course. We no longer have to risk lives and spend millions if our goal is to protect homes.
Cohen has done the hard scientific work. So, can we assume that his ideas have caught on like … um, wildfire? Not so fast. People seem to understand the science but are still reluctant to change their behavior.
Cohen relates a conversation with a friend about the difference between fighting fires and saving homes. It goes something like this:
Friend: “Modifying homes to make them fire resistant isn’t rocket science.”
Cohen: “No. This is much harder. This is social science.”
Friend: “Oh, jeez. We’re screwed.”
Cohen has done the hard science but the hard work remains. As Albert Einstein, the most famous physicist of all, said: “It’s easier to smash an atom than a prejudice.” Perhaps it’s time to develop some sociology envy.
Julia and Elliot recently went to a wedding in Eureka, Colorado, a ghost town situated high in the San Juan Mountains. To say that Eureka is isolated is a vast understatement. Here are some things that the town doesn’t have: landlines, television, internet, wi-fi, mobile phone access, cable, newspapers, radio, and paved roads. When you’re there, you’re there.
Just before the wedding ceremony, ushers collected everyone’s cameras and mobile phones. The couple seemed to be saying to their guests: “We’re glad you’re here. We hope you’re with us fully and completely. Don’t fuss with your cameras and phones. Engage with us in a profound experience. Be here now.”
The place and the process reminded me of Daniel Kahneman’s definitions of the experiencing self and the remembering self. We can focus, engage our senses, and fully experience an activity. What we remember, however, is far different from what we experience. We typically remember two things: 1) the peak experience – the high point of the activity; 2) the end state – how things ended up. (For more on memory traps, click here).
The difference between the two selves has many implications. We remember things differently than they actually happened. This calls into question such things as eyewitness testimony and historical accounts. It may also be why we argue with our spouses – we simply remember things differently. It’s another good reason not to argue in the past tense. (For other reasons, click here).
The difference also affects how we plan our activities. We can plan to: 1) enhance the experience; or 2) enhance our memory of the experience. Let’s say you go to your favorite restaurant. If you want to enhance the experience, you should order your favorite dish. You can enjoy the anticipation and the experience itself. However, you won’t create a new memory. It will simply blend in with all the other times you’ve ordered the same dish. If you want a new memory, order something you’ve never had before. It may be great or not – you can’t anticipate – but it will be more memorable.*
FOMO, of course, is the Fear of Missing Out, which seems to be an increasing concern in today’s society. Everything is accessible and we don’t want to miss any of it. Technologies such a mobile phones, video chat, and instant messages democratize our experience. We can share anything with anyone at any time. We won’t miss a thing.
But, of course, we do miss things. In fact, the very act of inserting technologies into our experiences makes us miss some of the experience. We’re fussing with our cameras rather than experiencing the action. We’re checking baseball scores rather than engaging with others. The desire to miss nothing causes us to miss something: the intensity of the present moment.
FOMO shifts our attention from the experiencing self to the remembering self. We take pictures, which helps us remember and share an experience. But the act of taking pictures insulates us from the experience itself. We’ve inserted technology between our experiences and ourselves.
As you can probably guess, I’m not the first person to suggest that FOMO mania actually causes us to miss much more than we realize. The tech and culture blogger, Anil Dash, coined the term JOMO — the Joy Of Missing Out – more than two years ago. Christina Crook wrote a book called The Joy of Missing Out and popularized the idea of Internet fasts. Sarah Wilson points out that FOMO has eradicated traditional boundaries that separated public time from private time. It used to be easy to spend a quiet evening at home. Now we need to declare an Internet fast to get some alone time.
Though it’s not a new idea, I suspect that the JOMO message needs some more evangelists. As a famous American once said: Be Here Now.
*I adopted the restaurant example from an episode of the Hidden Brain podcast called Hungry, Hungry Hippocampus.
Daniel Kahneman is rightly respected for discovering and documenting any number of irrational human behaviors. Prospect theory – developed by Kahneman and his colleague, Amos Tversky – has led to profound new insights in how we think, behave, and spend our money. Indeed, there’s a straight line from Kahneman and Tversky to the new discipline called Behavioral Economics.
In my humble opinion, however, one of Kahneman’s innovations has been overlooked. The innovation doesn’t have an agreed-upon name so I’m proposing that we call it the Kahneman Interview Technique or KIT.
The idea behind KIT is fairly simple. We all know about the confirmation bias – the tendency to attend to information that confirms what we already believe and to ignore information that doesn’t. Kahneman’s insight is that confirmation bias distorts job interviews.
Here’s how it works. When we meet a candidate for a job, we immediately form an impression. The distortion occurs because this first impression colors the rest of the interview. Our intuition might tell us, for instance, that the candidate is action-oriented. For the rest of the interview, we attend to clues that confirm this intuition and ignore those that don’t. Ultimately, we base our evaluation on our initial impressions and intuition, which may be sketchy at best. The result – as Google found – is that there is no relationship between an interviewer’s evaluation and a candidate’s actual performance.
To remove the distortion of our confirmation bias, KIT asks us to delay our intuition. How can we delay intuition? By focusing first on facts and figures. For any job, there are prerequisites for success that we can measure by asking factual questions. For instance, a salesperson might need to be: 1) well spoken; 2) observant; 3) technically proficient, and so on. An executive might need to be: 1) a critical thinker; 2) a good strategist; 3) a good talent finder, etc.
Before the interview, we prepare factual questions that probe these prerequisites. We begin the interview with facts and develop a score for each prerequisite – typically on a simple scale like 1 – 5. The idea is not to record what the interviewer thinks but rather to record what the candidate has actually done. This portion of the interview is based on facts, not perceptions.
Once we have a score for each dimension, we can take the interview in more qualitative directions. We can ask broader questions about the candidate’s worldview and philosophy. We can invite our intuition to enter the process. At the end of the process, Kahneman suggests that the interviewer close her eyes, reflect for a moment, and answer the question, How well would this candidate do in this particular job?
Kahneman and other researchers have found that the factual scores are much better predictors of success than traditional interviews. Interestingly, the concluding global evaluation is also a strong predictor, especially when compared with “first impression” predictions. In other words, delayed intuition is better at predicting job success than immediate intuition. It’s a good idea to keep in mind the next time you hire someone.
I first learned about the Kahneman Interview Technique several years ago when I read Kahneman’s book, Thinking Fast And Slow. But the book is filled with so many good ideas that I forgot about the interviews. I was reminded of them recently when I listened to the 100th episode of the podcast, Hidden Brain, which features an interview with Kahneman. This article draws on both sources.
As a marketing executive, I’m accustomed to the process of segmenting markets. My software companies divided markets by company size, SIC code, geography, and installed software. By doing so, we could identify and communicate with companies that were most likely to buy our products and services.
The Pew Research Center has done something similar with America’s body politic. Rather than two major political parties, Pew identifies eight different clusters of political thought and action: four for the Left and four for the Right. Let’s take a look at each. (Click here for the complete report and here for a Q&A on the methodology).
Core Conservatives support traditional Republican positions including smaller government and lower taxes. They also believe that the nation’s economic system is fundamentally fair. Core Conservatives generally support globalization, believing that the global economy creates opportunities for American businesses. This segment accounts for approximately 31% of all Republicans and roughly 13% of the American body politic.
Country First Conservatives are deeply concerned about immigration and generally believe that America should withdraw from the world. They strongly believe that the global economy is deeply threatening to American interests. More than any other segment, Country First Conservatives agree with the statement “If America is too open to people from all over the world, we risk losing our identity as a nation.” This segment accounts for 14% of Republicans and 6% of the American public.
Market Skeptic Republicans share many conservative values but are losing faith in the American economic system. They believe the economy unfairly favors powerful interests and are skeptical of banks and other financial institutions. More than any other group, the disagree with the statement “The U.S. economic system is generally fair to most Americans.” This segment accounts for 22% of Republicans and roughly 12% percent of the American public.
New Era Enterprisers form the most optimistic segment of either the Right or the Left. More than any other group they believe that the next generation of Americans will be better off than the current generation. New Era Enterprisers strongly support business and generally believe that immigrants strengthen the country. This segment makes up 17% of Republicans and 11% of the general public.
Adding it up (with some rounding errors), the four Conservative segments account for approximately 42% of the American body politic.
Solid Liberals have liberal attitudes across the board. They overwhelmingly support government involvement in health insurance and government activism to ensure that blacks have equal rights with whites. They support government regulation of business and don’t believe one needs to be religious to be moral. Solid Liberals account for 16% of the general public 33% of Democrats.
Opportunity Democrats agree on many policy issues with Solid Liberals. They’re more likely, however, to be pro-business and to believe that most people can get ahead if they work hard. They comprise 12% of the general public and roughly 20% of Democrats.
Disaffected Democrats are positive about the Democratic Party but cynical about government in general. This is one of two minority-majority segments. They support activist government programs and believe that we should focus less on problems abroad and more on domestic issues. Disaffected Democrats account for 14% of the general public and 23% of Democrats.
Devout and Diverse is the second minority-majority segment. Like Disaffected Democrats, they face financial hardships. Among all the groups, this segment is the least politically engaged. They resist government regulation and strongly believe that one needs to believe in God to be moral. The Devout and Diverse account for 9% of the public and 11% of Democrats.
Adding it up, the four Liberal segments account for roughly 51% of the American body politic.
So if Conservatives account for 42% and Liberals for 51% of the general public, where are the rest of our voters? According to Pew, roughly 8% of the general public are Bystanders, described as a “relatively young, largely minority group” that is “missing in action politically”.
So where do you stand? Thanks to the Pew Research Center, you can now find out. Pew has published the Political Typology Quiz. Click here, answer a series of questions, and you can figure out which group most closely represents your worldview. I’ll be writing more about this in the future. You may want to know which group you’re in to follow along.