AI and the Future of Work: a discussion with Irving Wladawsky-Berger
Irving Wladawsky-Berger is an expert in emerging technology, a regular contributor to the Wall Street Journal’s CIO Journal, and a Visiting Lecturer at MIT's Sloan School of Management, as well as Adjunct Professor at the Imperial College Business School. He was born in Cuba and came to the US at the age of 15, and was named Hispanic Engineer of the Year in 2001. He has an M.S. and Ph. D. in physics from the University of Chicago. To learn more about Irving, see - http://www.irvingwb.com/ - on that site, you can subscribe with your email address to his blog.
The discussion below is in relation to a conference that took place at MIT with various AI professionals and researchers, and the question of the impact of AI on the job market. If you’re interested in the question, “What’s Going to Happen? And When?”, I highly recommend reading the related blog post: http://blog.irvingwb.com/blog/2017/12/ai-and-the-future-of-work.html
--
Todd: Irving, I appreciated your blog post on AI and the future of work about the conference at MIT. It was really interesting to hear about that meeting, it feels like a watershed moment of some kind, especially with the people who were there. It has really come closest of anything I’ve seen so far, to providing solid evidence, perhaps the first I’ve seen that is convincing to me at least, of the potential for AI to have flat or eventual net job increase. I think this particular quote from one of the attendees was striking:
No invention in the 250 years since the first industrial revolution has caused mass unemployment, and that though jobs are constantly being destroyed, they are also being created in even larger numbers
But in spite of the evidence and data discussed at the meeting, it still feels like something may be different, meaning AI could end up replacing more jobs than it adds, and not just assisting people, but replacing them. I think what comes to mind is just the extent to which jobs and industries are virtualized these days, making them potentially easier to replace, and much faster than in past industrial transformations. In relation to the quote above, I’m also not sure there has ever been an invention like digitization, which began as a tool and has now been at the core of so many industries, not just speeding them up but actually absorbing them. If the computer was at first a tool and then the Internet was at first a set of tools, then it seems that virtualization has been a novel phenomenon.
Irving: Many people think that this time is different, that our technologies, including virtualization, the Internet, AI, etc are different from anything we've seen in the past. But, many believe that the changes around the turn of the 20th century may have been even bigger, with electricity, indoor plumbing, cars, airplanes, telegraphs, radio, TV, etc. If you look at what life was like in advanced countries like the US, UK, etc in 1880 and compare it to what life was like in 1930 it was truly a gigantic difference, perhaps bigger than comparing today with life 50 years ago - 1967.
Todd: That seems worth taking into account. I did read the very recent McKinsey study that came out, and I recognize that it has fairly deep data, and suggests more of an optimist view. Like many other optimist sources, it recognizes there could be significant job lost, and I appreciate how it urges government leaders to make policy decisions in support of workforce transition. I want to believe its conclusion that in the end, based on the data they see, that more new jobs will be created, eventually, but I’m not quite convinced.
Irving: The MIT AI conference I wrote about in that blog post, and the McKinsey study discussed at the meeting, are some of the best explanations why we should be optimistic. There is no question that if managers can get the same work done with fewer people they will do so. But remember, new technologies and innovation will lead to new jobs and industries over time. So, as McKinsey said, there will be jobs lost, jobs transformed, and jobs gained, but in the end, McKinsey and most speakers at the AI conference felt that over time we'll be fine.
Todd: Ok, so if you can indulge me, please address the remaining doubt I have; let’s say I want to be an optimist, but I keep thinking how in the business world and many other fields, so much of it is already digitized, which seems to argue that disruption could happen a lot quicker, and that it may be a reason why more jobs could be lost. If a great deal of jobs use digital tools, making it easier to automate them, then it seems possible. Even in the McKinsey report, with all due respect, they mostly seem to be doing predictive analysis based on past trends, and they don’t seem to address the phenomena of how digitized work is. Maybe even more simply, it seems like if something can be automated, it will be, and because of digitization, it seems like the speed and scale could have a more significant impact on the past.
What do you think about the issue of digitization? Would you agree that in spite of the data in the McKinsey report and similar ones, that the impact of digitization may be a wildcard, that could potentially make this disruption more impactful? Do you think that the virtualization and automation of tasks could make the disruption from AI different than the inference and projection made by Mr Gordon?
Irving: Todd, let me try to address your very reasonable doubts based on research on the topic at MIT and McKinsey. The McKinsey study you referenced above explained that most occupations involve a number of activities, and just because some of the activities in a job have been automated, does not imply that the whole job has disappeared. Based on a detailed analysis of over 800 occupations, McKinsey concluded that less than 5 percent, consist of activities that can be entirely automated. However, in about 60 percent of all occupations, at least one-third of their constituent activities could be automated. This means that many more occupations will be significantly transformed than will be automated away by 2030.
Another piece of evidence comes from the MIT conference, where MIT professor Erik Brynjolfsson explained that AI is a general-purpose technology, like electricity, and such technologies require complementary innovations and investments that take quite a bit of time to play out before their benefits are felt across the economy and society. Realizing the benefits of AI, will probably require reinventing our organizations and institutions. History suggests that eventually things work out, but there can be long, painful periods of transition.
Todd: Fair enough. I suppose I lean towards pessimism about net job loss, and if that is possible, it seems to have implications for individuals and policy makers that we could be in for some serious trouble. And if a greater danger is perceived, perhaps more people will wake up to it and take action. But I’m not an alarmist, because I see a lot of good coming from AI, and I can see the possibility of new jobs arising. But I guess we just don’t know how big the impact will be and how fast.
Irving: We don't really know what the future will bring. There are techno-pessimists who think that there will be little innovation, mass unemployment, social disturbances, etc. There are techno-optimists who feel that we will once more adapt to technology advances just as we have for the past 250 years. I'm in the techno-optimist camp, but in the end, we don't really know.
Todd: The other part early in your article that I wanted to believe, that I want to believe, is the message I hear in various readings about the notion that only portions of jobs will be automated. It reminds me of the announcement earlier this year with big fanfare by Eric Schmidt, and a conference, that “90% of jobs are not fully automatable”. But this feels more like wishful thinking, as opposed to grounded in corporate reality. It makes perfect sense if upper managements’ priority was to protect full time jobs - but in a publicly traded company, I can hardly see a manager keeping 100% of full-time staff in situations where a significant portion of their tasks can be automated, especially if faced with economic uncertainty, declining revenues, or simply the constant need to increase profit. It seems more likely that they would reduce headcount.
If a CEO has the ethical shield of duty to shareholders, why would they not do what is best for the shareholders, which would be to be as profitable as possible, which would be to reduce waste as much as possible, and therefore automate as much as possible?
Irving: Todd, clearly companies will do what they have to, to increase productivity, revenue and profits, including as much automation as is feasible. But, as I mentioned earlier, the prediction that over 90% of jobs are not fully automatable in the next 15 to 20 years, is based on research at McKinsey, MIT and other universities and think tanks. The longer we have to adapt to AI and other transformative technologies, the more time we have to create new jobs and industries. But, remember that the fact that things work out over the long term does not mean that there isn’t considerable pain in the short term, as many workers who don’t have the skills to keep up with technology will be left behind. Being a techno-optimist does not mean that we shouldn’t acknowledge that the transition will not be easy for many workers.
Todd: I guess the most important thing is not the question of whether there is net job loss or net job gain, but how do we respond, as societies and individuals. I’m guessing it would benefit government leaders to take a close look at your blog post, at the McKinsey report and other similar ones from Oxford, PwC, etc., as well as the work of Dr. Carlota Perez, mentioned at the meeting, which seems to strike a good balance and strongly recommend investment in infrastructure and renewable energy as one of the areas of job gain. The consensus them on a macro level seems to be that the more passive we are as individuals and societies, the negative impact of AI will be much greater. And conversely, we have a choice – the more active we are as individuals and societies in responding to AI, the better.
Irving: Precisely. Governments, business and individuals have a responsibility to do everything possible to adjust to this changing future, so that technology advances benefit everyone. As the Mckinsey report said in conclusion:
“Automation represents both hope and challenge. The global economy needs the boost to productivity and growth that it will bring, especially at a time when aging populations are acting as a drag on GDP growth. Machines can take on work that is routine, dangerous, or dirty, and may allow us all to use our intrinsically human talents more fully. But to capture these benefits, societies will need to prepare for complex workforce transitions ahead. For policy makers, business leaders, and individual workers the world over, the task at hand is to prepare for a more automated future by emphasizing new skills, scaling up training, especially for midcareer workers, and ensuring robust economic growth."