“Tell me something that’s true that very few people agree with you on.”
My Answer: Artificial Intelligence is not going to result in a net destruction of jobs. AI will not result in a meaningful step-change in economic growth anytime soon.
I may be completely wrong but end of the day it’s just a point of view. It’s come about after a few years now of looking at some tech companies in my day job, the progress they are making, the current impact on society and then all the pundits brave enough to have an opinion on the net impact AI will have on society. So what follows is some of the more interesting viewpoints I’ve been reading recently —
“This time it’s different.” — Sir John Templeton
People forget artificial intelligence has been through booms and busts throughout history.
“Machines will be capable, within 20 years, of doing any work a man can do.” — Herbert Simon, The Shape of Automation for Men and Management, 1956
“Within a generation…the problem of creating ‘artificial intelligence’ will substantially be solved.” — Marvin Minsky, Computation: Finite and Infinite Machines, 1967
So people have essentially been saying the same thing for more than 50 years…mmm
A great conversation between Marc Andreessen and Dan Primack goes into great detail around two binary narratives that are prevailing in the context of how artificial intelligence will or will not change the world as we know it.
Narrative 1: AI is a bubble, worthless, over-rated and Silicon Valley is a bunch of kids running around creating a bunch of photo-sharing apps
Narrative 2: AI is the next revolution, overwhelming changing the economy, destroying jobs, automating every menial manual labour task
The four big questions to consider the pace of change in the economy
Would you expect that the current rate of productivity growth to be running at generational highs or generational lows?
Consensus: Generational highs
Reality: Generational lows — getting 1% annual productivity gains
Would you expect the rate of new companies in the economy to be at generational highs or generational lows?
Consensus: Generational highs
Reality: The rate of new companies being introduced to the economy has been falling for the last 40 years with no signs of inflecting
Would you expect the rate of the destruction of jobs to be at generational highs or generational lows?
Consensus: Generational lows — outside of the GFC era
Reality: Matches with consensus — at a generational low
Would you expect the time people spend in a job to be at generational highs or generational lows?
Consensus: Generational highs
Reality: Not at a generational low but it has been slowing for the last 20 years
What is the problem today?
- The problem is not too much technological innovation but not enough.
- The Cobb-Douglas function says that total output = to the product of total factor productivity (technology), capital input and labour input.
- Therefore if you had higher technology growth you would have higher economic growth and therefore faster job creation and rising incomes
Bifurcation of the economy
Andreessen argues that there is a bifurcation of the economy, and on the whole I agree with him on this point.
Fast productivity growth sectors
These sectors are less regulated and the most affected by technology. Technology has lowered prices which has been great for consumers but bad for businesses and workers resulting in accelerating bankruptcies and job losses.
- Media — the new distribution mechanism (internet) has accelerated productivity growth
- Retail — growth of Walmart then the arrival of Amazon (and peers)has fragmented consumer demand across more channels
- Financial services — investment management, stock trading and exchange traded funds have lowered the access costs for consumers to financial markets
So basically technology has created a consumer utopia with access to Netflix, Amazon and Vanguard with low prices and great products but resulted in a massive transformation in these industries
Slower productivity growth sectors
These sectors are more heavily regulated creating higher barriers for technological change and exhibit increasing prices and limited productivity growth
- Healthcare — Dr. Roger Perlmutter, Head of Merck Research Labs, recently said that the entire pharmaceutical industry is only capable of registering two dozen drugs a year and only a handful of those actually return their cost of capital
- Education — Consider the cost of private schooling in Sydney which has risen 20% in the past four years
- Real estate and construction — prices continue to rise particularly in high income areas
Overall the sectors where the prices are falling are shrinking as a percentage of the economy and those where prices are rising are growing as a percentage of the economy.
The Future of jobs and education
Historically we have seen the shifts created by changes in technology and it continues to progress — machines replacing various tasks involving manual labour — nothing new.
The argument for this next wave of AI, should it occur, will create new jobs like the ones courtesy of my imagination:
- 3D printer mechanics
- Drone air traffic controllers
- Self-driving car salesmen
“This is a cycle. The only difference with what we have been seeing throughout human history is that now, machines are coming after people with college degrees, political influence and Twitter accounts.” — Garry Kasparov, Chess grandmaster
The education system, particularly in the developed world, is largely back-ward looking. Yes, I hear that schools now have 3D printers but realistically being able to operate a 3D printer is akin to being able to operate a PC back in the 1990s — zero value add.
Personally, I think the greatest flaw the education system has today is rote learning. When I did the Higher School Certificate not that long ago I was told to memorise all my essays and just adapt them to the questions! Same experience at university.
We have to be teaching kids how more. They need to learn how to think critically and how to question effectively. What they learn in school needs to prepare them for a world that may look completely different once they enter the workforce.
There needs to be greater emphasis on:
- Developing emotional intelligence
- Risk literacy
- Financial literacy
- STEM subjects — particularly among young girls and women
The Extreme Consensus…
“The data point people always give is self-driving cars. The fact they always come up with the same example suggests that maybe there’s not that much to it. Even if you got — and I think self-driven cars would be significant — it might replace at most 1 per cent of the workforce, it might increase productivity a few percent in the economy. If you phased them in over a decade, it would not be that transformative.” — Peter Thiel
We do have job losses in some industries but on the whole there is job creation albeit part-time in nature (in Australia anyway). Productivity growth in sectors where significant pressures will be felt in the future from the ageing population are yet to exhibit any meaningful inflection in productivity growth.
We are yet to see 3D printer mechanics or self-driving car salesmen. In any event the threat of job-destroying artificial intelligence should be used as a catalyst to re-engineer an out-dated education system to prepare the next generations of kids for opportunities that may present themselves. Time will tell whether this time is different…