Harvard Business Review May-June 2020 pp158-159 “Harnessing Artificial Intelligence” “We need new rules to avert serious risks” by Gardiner Morse.
Morse mentions the unique elements of each of four books, one upcoming and three already published in 2020. He highlights some commonalities.
The Books
1) T-Minus AI by Michael Kanaan BenBella Books August 2020
2) The Autonomous Revolution William H. Davidow and Michael S. Malone Berrett-Koehler 2020
3) A World Without Work Daniel Susskind Metropolitan Books, 2020
4) Reprogramming the American Dream Kevin Scott Harper Business 2020
“Stephen Hawking warned that ‘the development of full artificial intelligence could spell the end of the human race’”. All the references (1-4) assure us that we are decades or centuries from when “AI surpasses human intelligence”. So we should really be focusing on what AI can achieve now-“smart speakers to self-driving cars”. Seems pretty benign but Kanaan(1) points out that some nations are already using “AI to advance antidemocratic agendas”. The Chinese are using AI to calculate “social trustworthiness” as a basis to reward or punish. Bad score and getting a plane ticket, a loan or a job might be harder. Reportedly facial recognition is being used to ID and round-up Ulgur minority members, mostly Muslim, for “reeducation”. Kanaan’s point is start discussing and developing policies to preempt this type and related application.
Davidow and Malone(2) point out that AI is rapidly narrowing human jobs by outperforming humans on an ever increasing number of tasks. “AI may boost productivity while driving down GDP.” An abundance leading to lower prices, declining wages, increasing economic inequality until we have “millions of ZEVs’-people of Zero Economic Value”.
Susskind(3) “…sees us humans retreating to an ever-shrinking set of activities”. Truck driver with GPS is optimal but a fully autonomous truck transport means 3.5M fewer jobs. There are worries that jobs requiring human traits like empathy, judgement and creativity may eventually be replaced as AI just gets better at mimicking-close enough but not perfect.
Scott(4) sees AI as enabling. An example being with a little worker training, some computing power and robotics, a person could manufacture from any location-from home rather than a centralized location. Scott also believes that AI will create new jobs “supporting AI itself”. Hard to quantify if that works out as a net negative or positive in job number.
What should be the focus going forward to avoid “serious risk”?
“As Kanaan puts it , we need to implement AI ‘only in ways consistent with fundamental human dignities…and only for purposes consistent with democratic ideas, liberties and laws. “…in the end, the future demands as we begin to build a world that is truly safe and sustainable, one where humans and AI can truly exist together”.