I was recently sent a copy of “Turning Point: Policymaking in the Era of Artificial Intelligence,” by Darrell M. West and John R. Allen, both of the Brookings Institution. It’s a high level overview of issues pertaining to artificial intelligence (AI), issues that impact both governments and industry. It’s an excellent book for middle and upper level management, even with the one significant flaw I noticed.
After the introductory chapter, there are chapters focused on five key business sectors being affected by AI: healthcare, education, transportation, ecommerce (though they still use the hyphen), and defense. The chapters clearly discuss both the opportunities and risks in leveraging AI. It’s no surprise that privacy repeatedly pops up as an important issue. What’s especially useful is that those discussions give both executives and government officials ideas for concern, suggesting potential regulations that might be needed and what actions companies might take on their own to alleviate risks.
It’s in that overview that the one big issue pops up. The chapter in healthcare is excellent except for one missing piece. Regulatory agencies such as the FDA have regular processes to approve new and updated technology. That’s easy to do for hardware and for procedural code, because it’s easy to specify exactly what has changed. In deep learning, it’s not so easy. The algorithms build in those systems are often not made explicit. More transparency is needed for agencies to create policy around changing AI. While the plethora of other healthcare issues is adequately surveyed in the chapter, this is a critical issue and was not mentioned.
The rest of the chapters are good, with wide coverage of issues. One small nit I have is in the ecommerce chapter, where the authors state “Already there are an estimated 31 billion e-commerce deliveries annually in China, compared with 13 billion in the United States.” Those numbers in a void don’t mean much, especially since the population difference is larger than the delivery difference. The percentage of each population are making those purchases would have been more informative.
In the other direction, I thought the chapter on defense was exceptionally good. What made it excellent was that it didn’t stick to they typical media focus on weapons and silly “Terminator” scenarios. It pointed out that, in the cyber age, defense means more than standard weapons enhanced with AI. Cybersecurity is a major defense threat, not only to militaries, but to governments, infrastructure (electricity, dams, etc.) and business. In addition, there’s a clear message to US focused readers, as it’s made clear that the US Department of Defense seems to be lagging adversarial nations in a focus on AI. That this chapter is wonderful should be no surprise given the background of retired General John Allen.
MORE FOR YOU
What Is The Artificial Intelligence Revolution And Why Does It Matter To Your Business?
7 Successful Ways To Use Artificial Intelligence To Improve Your Business Processes
10 Wonderful Examples Of Using Artificial Intelligence (AI) For Good
The last three chapters are devoted to overarching issues of AI’s impact on society. In the last one, they provide a list of eight principles that should be considered:
· Human control
· Human rights
While I have no argument with the list, I do have a problem with the ordering. Transparency shouldn’t be last. It is the cornerstone of showing that the others are being addressed. As mentioned above, deep learning is a black box. Without government regulations mandating transparency, the ability to provide the trust that the other principles are being followed will be lacking. Unfortunately, it’s the most technical of the issues, and technology is one thing that governments have consistently shown they don’t comprehend.
We can look back humorously at former Senator Ted Stevens declaring that the internet was a “series of tubes”, but managing AI is an even more complex challenge. Unless democratic nations work to ensure that governments are populated with people who understand that technology is critical, the ability to protect the freedom and concepts put forward in “Turning Point” will be difficult.
“Turning Point” is an excellent book that decision makers can use as a tutorial to understand the growing importance of policy related to AI, policy in both the public and private sectors.