Cognitive technology The rise of “bionic brains”

One of the most significant developments in technology—and society in general—over the next several years will likely be the use of smart machines to replace human labor. This is one of Deloitte’s analytics trends of 2015, and I have no doubt that it’s coming. My only doubt is what to call it; even within Deloitte, some call it “bionic brains,” some “cognitive technology,” some “cognitive analytics”  and some “amplified intelligence." I do like hanging out at a firm where there are no language police.

And certainly, there is a wide variety of terms used to describe this phenomenon outside of Deloitte. “Artificial intelligence” is an old standby—perhaps too old. And googling “IBM cognitive” will get you not only the widely used “cognitive computing,” but also “cognitive systems,” “cognitive speech,” “cognitive environments,” and the intriguing “cognitive cooking.” Sometimes, to avoid all this terminological confusion, I use the generic term “smart machines.” I do avoid “robots” because most of these technologies don’t have arms, legs, or heads with antennae; they’re just another form of software running on a computer.

Regardless of what you call them, the case for their having a big impact on our lives is pretty strong. Here are a few straws in the wind:

• There are already systems that can make decisions previously made by humans in a variety of knowledge-work domains, including medicine, law, accounting and auditing, journalism, translation, teaching, marketing, and architecture. Thus far, they have led to few if any employment terminations, but they may already be limiting the growth of some of these professions.

• Given that these systems are becoming available, it seems unlikely that companies and organizations will decline to use them in order to preserve human jobs and skills. This is the wish of Nick Carr in his well-argued new book The Glass Cage, but I don’t think it’s how capitalism works. When I talk with business leaders about automation, they seem quite willing to explore how well it will work for their organizations, if it will save them money, or if it can make them more competitive.

• From a technology standpoint, there are a variety of developments coming together: Watson-like systems for digesting text, big data analytics, rule-based systems, machine learning for developing automated models, and various other branches of artificial intelligence. AI is still best at making narrowly-defined decisions, but it’s clearly moving in the direction of broad intelligence. The best systems will combine several or all of these tools; after all, our brains can employ a variety of ways to address a problem.

• The need for these types of systems is driven by the increasing complexity of knowledge. In cancer care, for example, we are now aware of more than 400 types of cancer. We know that hundreds of genes play a role in cancer. For breast cancer alone, there are more than 75 drugs to choose from for treatment and prevention. All this means that if you are a human doctor attempting to diagnose and treat a patient with cancer, you could use some help from a smart machine.

So this is going to happen, and the question is what are we going to do about it. Other observers have commented that we need to supply more education and retraining for displaced workers, but that seems like a tired old prescription. Others suggest that we need to begin to prepare for a time when the link between employment and “making a living” has been severed. They propose income redistribution, safety nets, more leisure activity, and work-as-a-hobby. These arguments may be valid, but I don’t see them being adopted anytime soon in the current society and polity.

I would argue that there are many ways we need to address this issue, but one that isn’t being addressed enough is “augmentation” rather than “automation.” We need to identify ways in which smart humans can augment the work of smart machines, and vice-versa. This is not an entirely new idea—it’s been mentioned, for example, in good books by authors like Tyler Cowen in Average is Over, and Erik Brynolfsson and Andy McAfee in The Second Machine Age.  But the augmentation approach has been discussed narrowly—in both of these books, for example, with regard to “freestyle chess.” This is an approach to chess in which humans use their own intelligence, combined with that of computerized chess programs, to compete. It’s a nice illustration of augmentation, but most of us don’t play chess for a living.

Instead, we need augmentation examples involving how doctors, lawyers, and accountants can work closely with computer systems. How can they produce better diagnoses, trial strategies, and audit opinions working with computers than either computers or humans working alone could produce? How could they become more efficient and effective as a team? I’ve seen a few examples of this collaboration in fields like radiology and tax accounting, but we need a lot more.

Augmentation won’t save every job from automation, but it’s perhaps the best hope we have. Humans aren’t going to improve their brains at the rate computers will, and we’re not going to turn the bionic brains off. So we need to devote a good chunk of our brainpower and creativity to how we can work in partnership with the smart machines that we have created.