AI and Technological Anxiety: Paranoia , or are the robots out to get us Comments on
1. Business School
AI and Technological Anxiety:
Paranoia, or are the robots out to get us?
Comments on
“What can Machine Learning Do and What Does It Mean for the Economy?”
by Erik Brynjolfsson
Kevin Fox
OECD Global Forum on Productivity
Sydney
20-21 July 2019
2. A Significant Body of Work is Emerging
Including:
• Brynjolfsson, E. and T. Mitchell (2017), “What can machine learning do?
Workforce implications,” Science Vol. 358, Issue 6370, 1530-1534.
• Brynjolfsson, E., D. Rock and C. Syverson (2017), “Artificial Intelligence
and the Modern Productivity Paradox: A Clash of Expectations and
Statistics,” NBER Working Paper 24001, Cambridge MA.
• Brynjolfsson, E. T. Mitchell and D. Rock (2018), “What Can Machines
Learn and What Does it Mean for Occupations and the Economy,”
AEA Papers and Proceedings 108:43-47.
3. Artificial Intelligence and Machine Learning
Brynjolfsson and Mitchell (2017):
• Machine Learning is a subfield of artificial intelligence that studies
the question:
“How can we build computer programs that automatically improve
their performance at some task through experience?”
• They argue that this is a General Purpose Technology: pervasive,
improves over time and generates complementary innovation
(Bresnahan and Trajtenberg 1995).
But what is “artificial intelligence”? I asked Siri.
9. Two Views on Artificial Intelligence
1. A potential source of productivity growth, which is a key determinant
of long-run living standards.
2. A threat to our living standards, as it will replace us as workers.
Each contribute to “technological anxiety”:
• Too little technological progress will result in low productivity growth
and low growth in living standards
• Technological progress will make us obsolete
Or as a scientist arguing for government support for
manufacturing asked me:
“What are we going to do in the future Kevin? Serve each other
cappuccinos?!”
10. Two Views on Artificial Intelligence
• Start by looking at the productivity slowdown
• Then look at possible technological obsolescence of humans
11. faculty of science
Trend Labour Productivity Growth in G7 Countries
Average Annual Rate, OECD Productivity Compendium 2016
12. faculty of science
Trend Labour Productivity Growth in G7 Countries
Average Annual Rate, OECD Productivity Compendium 2016
14. Australian Multifactor Productivity Slowdown
Market Sector, Annual Averages
-2 -1 0 1 2 3 4
Arts and Recreation Services
Mining
Accommodation and Food Services
Construction
Manufacturing
Electricity, Gas, Water and Waste Services
Information, Media and Telecommunications
Market Sector (12)
Retail Trade
Transport, Postal and Warehousing
Financial and Insurance Services
Wholesale Trade
Agriculture, Forestry and Fishing
Average percentage growth
1989-90 to 2003-04
2003-04 to 2016-17
15. “The History of Technological Anxiety and the Future of
Economic Growth: Is This Time Different?”
Joel Mokyr, Chris Vickers, and Nicolas L. Ziebarth (2015), Journal of Economic Perspectives 29(3), 31–50.
Have we run out of useful things to invent? That’s been a cause of
anxiety many times in the past.
Alvin Hansen’s1938 book Full Recovery or Stagnation?
“Hansen drew on the macroeconomic ideas of John Maynard Keynes
in fearing that economic growth was over, with population growth and
technological innovation exhausted.”
16. Productivity Slowdown
• If AI is so great, then where’s the productivity?
Mismeasurement of productivity can’t be the whole story:
Syverson (2017).
Lags probably play a role; diffusion, learning, re-organising,
development of complementary technologies (Brynjolfsson, Rock
and Syverson 2018)
Historical evidence of new GPTs suggest that there is still hope
that a big burst of productivity is forthcoming.
17. Workers, Arise!:
Luddite Riots of the early 1800s
“Luddites objected primarily to the rising popularity of automated textile
equipment, threatening the jobs and livelihoods of skilled workers as this
technology allowed them to be replaced by cheaper and less skilled
workers.” (Wikipedia)
18. Workers, Arise!:
Karl Marx (1937), “The Strife Between the Workman and Machine,” Capital
Vol. I, Ch. 15, Sect. 5:
• “The instrument of labour, when it takes the form of a machine,
immediately becomes a competitor of the workman himself.”
• “Abbé Lancellotti, in a work that appeared in Venice in 1636, but
which was written in 1579, says as follows:
‘Anthony Müller of Danzig saw about 50 years ago in that town, a
very ingenious machine, which weaves 4 to 6 pieces at once. But
the Mayor being apprehensive that this invention might throw a
large number of workmen on the streets, caused the inventor to
be secretly strangled or drowned.’”
20. Workers, Arise!:
What has Uber got to do with AI?
“AI is so central at the company, whether or not it should be used is not
even a question anymore.”
• “machine learning and neural networks are now core to just about
every business process. Uber uses AI for fraud detection, risk
assessment, safety processes, marketing spend and allocation,
matching drivers and riders, route optimization, driver onboarding,
and just about everywhere else it's possible to apply.”
John Koetsier, Forbes, 22 August 2018
https://www.forbes.com/sites/johnkoetsier/2018/08/22/uber-might-be-the-first-ai-first-company-which-is-why-they-dont-
even-think-about-it-anymore/#4971ec145b62
21. Workers, Arise!:
• Weavers protested new technologies in the 1800s, as they saw their
investment in skills, which are (intangible) assets, being devalued by
technology.
• Taxi drivers are protesting new technologies in the 2010s, as they
see their investment in licences, which are assets, being devalued by
technology.
23. Workers, Arise!:
“Yes, excessive automation at Tesla was a mistake. [...]
Humans are underrated.”
Elon Musk, 13 April 2019
Quoted in: The Human Side of Productivity: Setting the Scene, Background
Paper, OECD GFP Team, June 2019.
24. The Future of Work:
“Are we going to serve each other cappuccinos?” There are alternatives!
Stand Up Paddle Board Yoga Instructor
25. The Future of Work:
Some takeaways from Erik’s work:
• Tasks, not jobs, will be taken over by AI.
• Different jobs will be affected differently, as they’re combinations of
tasks with different susceptibility to being replaced.
• Economic considerations will be key.
• Jobs can be categorised by their “suitability for machine learning”
(SML)
• Workers will be impacted to different extents, maybe differently than
with previous rounds of technological disruption.
• New types of jobs will emerge.
• That is, the “robots” will take away tasks, but not usually whole jobs.
26. The Future of Work:
Few occupations will be immune to disruption from AI