This column is part of Globe Careers' Leadership Lab series, where executives and experts share their views and advice about leadership and management. Follow us at @Globe_Careers. Find all Leadership Lab stories at tgam.ca/leadershiplab.
For most people, it is not easy to picture the buying and selling of cognitive capabilities that have traditionally been embedded in humans – things like judgment and decision-making. Yet, thanks to recent advances in machine learning, we face the very real possibility of precisely such a 'market for intelligence'. Given the potential of this market to transform the entire global economy, we must all begin preparing – now – for its emergence.
To be clear, machine intelligence is still in its infancy, and while some of the current applications are remarkable, none are transformational. For example, the recommendation engines employed by companies such as Amazon and Netflix – which learn our preferences and recommend which books we should buy or which movies we should watch – are a common application of machine learning. Although they may increase the sales of books and movies – and may even enhance social welfare to some extent, by increasing matches between consumers and products – they do not represent a transformation to the economy.
Similarly, in the healthcare sector, applications of machine intelligence that identify and classify tumours from medical imaging data – with a higher degree of accuracy than the best human technicians – will surely enhance the productivity of doctors; but they will not transform the broader economy.
I suspect the reason why the driverless car has had such an effect on people is because it shows them something truly transformational: most people imagined a bigger gap between the machine's recommendation system and the cognitive requirements of a human driver. They are shocked to learn that, in fact, this new machine does not actually need human input at all.
How will advances in machine intelligence transition from providing simple productivity enhancements in individual, narrow markets, to transforming the overall economy? Will such a transition happen gradually or suddenly?
To the extent that the last technology shock – the Internet – offers guidance on this question, the answer is both.
In the case of the Internet, the development of the technology and associated infrastructure occurred gradually, but the transformative impact on the global economy occurred relatively suddenly. The economic transformation began abruptly in 1995. In 1991, the High Performance Computing Act passed; in 1992, Network Solutions took control of the domain name system and the Internet Society was founded; in 1993, the Mosaic Browser was launched for Unix and Windows OS; and in 1994, the cookie was invented at Netscape and the World Wide Web Consortium was founded. Then, in 1995, Bill Gates wrote his famous 'Internet tidal wave' e-mail, Microsoft launched Windows 95 and Netscape went public with a market capitalization of $3-billion – without displaying a nickel of profit.
So, when will machine intelligence experience its own '1995'?
As with any early-stage, general-purpose technology, there is much speculation and debate. For example, last fall, Tesla CEO Elon Musk remarked: "AI is much more advanced than people realize." To which deep-learning pioneer Yann LeCun responded via Twitter: "No, it's not. Quite the opposite in fact."
Within this fog of uncertainty, futuristic depictions abound. Not only are fiction writers featuring AI in stories like Transcendence, Her and Ex Machina, but governments – notably Japan, Germany, China and the United States – are featuring machine intelligence in their industrial strategies.
In the midst of all the speculation, real companies and investors are making real capital allocation decisions – today. In 2014, Google acquired the pre-revenue AI startup Deep Mind for approximately $500-million and created AlphaGo – which this year famously beat the world's top human player at the ancient Chinese game, Go, demonstrating what appeared to be 'machine intuition'. In March 2016, General Motors acquired AI startup Cruise for more than $1-billion to help turn regular vehicles into self-driving cars, and the following month, Salesforce acquired AI startup MetaMind to automate and personalize marketing and customer support.
Companies such as these are betting on how the future will unfold, and their bets are endogenous, in that they will influence the rate and direction of technological development – as well as where and how it occurs.
The downfall of once-mighty corporations like Barnes & Noble and Blockbuster provide ample warning for those contemplating a 'wait-and-see' strategy with respect to AI. It is incumbent upon investors, governments and firms in every sector to have a thesis regarding how their industry will be transformed when the necessary technological and regulatory pieces snap into place, and the world is suddenly confronted with a functioning market for intelligence.
While there is plenty of disagreement as to when AI will come to fruition, one thing is clear: decisions of significant consequence lie ahead.
Ajay Agrawal is the Peter Munk Professor of Entrepreneurship, Professor of Strategic Management and Academic Director of the Creative Destruction Lab at the Rotman School of Management.
A version of this article appeared in the Fall 2016 issue of Rotman Management, the magazine of the University of Toronto's Rotman School of Management. Reprinted with permission.