Commentary

To Be, Or Not To Be? How Will AI Answer THE Question?

  • by , Op-Ed Contributor, September 7, 2017
When you follow artificial intelligence, you’re constantly reading about how this company or that one is developing systems that learn the way people learn, or that mimic the way the human brain works. So, no surprise: I’ve talked to prominent AI developers who study neuroscience.

Consequently, I started reading popular brain books (for lay people). I eventually worked my way up to the challenging and ambitious “Self Comes to Mind: Constructing the Conscious Brain,” by noted UC Berkeley neuroscientist Antonio Damasio.

I’ve written before about a seminal quote of his from a previous book: “Humans are not either thinking machines or feeling machines but rather feeling machines that think.”

In “Self Comes to Mind,” Damasio unpacks that statement in ways that are scientifically intense and deeply profound.

He hypothesizes that humans cannot think rationally without emotion. He explains that all emotions arise from “primordial feelings” created in the evolutionarily older brain components whose primary responsibilities involve autonomous processes of life regulation. Intensive, bidirectional information flows between each brain component, and the far-flung corner of your body for which it is responsible, generate these emotions.

The “state” of all parts of your body are reflected in these older brain components so that they can do their job of keeping you alive while you worry about other things.

Importantly, when scientists do brain scans of people thinking hard, those older brain components “light up,” in concert with components in the cerebral cortex.

The evolutionarily newer cerebral cortex is where most people think we do all our thinking, if we think about thinking at all. But no; the truth is, thinking involves an incredibly complex, orchestrated dance among disparate elements distributed all over our brains’ older and newer parts.

My simplistic description skips reams of neuroscience-y detail that makes Damasio’s hypotheses hard not to believe. I hope, though, it clearly establishes the scientific basis for mind-body connection, as well as the unavoidable role of emotion in rational thought. Damasio goes further, noting that the human brain doesn’t “think” all within itself; it thinks with your entire body.

One more key Damasio concept is “biological value.” He asks, “Why do we take virtually everything that surrounds us—food, houses, gold, jewelry, paintings, stocks, services, even other people—and assign a value to it? Why does everyone spend so much time calculating gains and losses related to those items?” His answer has to do with the will to survive that is coded into all living organisms’ DNA, into every individual cell of the human body, and ultimately into our human psyches.

Notably, in humans, that will to survive is expressed in a way that is a bit more sophisticated than, say, an amoeba’s. “The valuations we establish in everyday social and cultural activities have a direct or indirect connection to homeostasis,” he writes. In other words, “Value relates directly or indirectly to survival. In the case of humans in particular, value also relates to the quality of that survival in the form of well-being.” (Italics are his.)  

So, the human brain thinks with the entire body, and is possessed of an absolutely irrepressible will to live that colors (scientists say “biases”) everything we think and all our decisions.

Excuse me, but how are AI developers going to mimic that?

Suddenly, the idea of them trying to do so, without benefit of mind-body connections or biological value, scares me. What if they succeed? The resulting consciousness could be disastrously foreign to our own. (Maybe this is why Elon Musk and others are so up in arms.)

I reached out to tech investor/philanthropist Esther Dyson, who was once called the most powerful woman in computing and who I think of as the tech industry’s conscience. She connected me to an essay published last month, “Benevolent Artificial Anti-Natalism,” which ultimately concludes that such a disembodied super-sentient AI without evolutionary biases would likely decide it was in the best interests of homo sapiens to end the race, primarily due to all the pain and suffering life endures. Or, as Hamlet put it, “The heartache, and the thousand natural shocks, that flesh is heir to.”
    
Not exactly heartening. Yet a savvy commentator points out that the result comes directly from assumptions programmed into the AI by the essay’s author, Professor of Theoretical Philosophy Thomas Metzinger. Slight changes in those assumptions would dramatically change the resulting logic, and conclusion.

I guess one moral of this story is: be careful about your assumptions when you program AI.

What about simulating a body for the AI brain to think with? Says Dyson: “In the end, everything converts, at one end, into atoms and protons and electrical charges, and at the other end, into how I feel, which influences how I think. You can simulate a body as much as you can simulate a brain. Obviously, either one is still a ways off.”

But, she points out, “Consciousness is pretty much a miracle in the first place. So creating something that mimics neurons is probably no less realistic.”

Which brings me back around to “Self Comes to Mind: Constructing the Conscious Brain.” Damsio’s primary goal is to explain that miracle – how the human brain (and body) creates a conscious self, from the dual points of view of current biology and how it likely evolved, over millennia.

I encourage anyone who has read this far to read the book, too. You may not learn how to use AI to sell more widgets-as-a-service (though don’t rule it out, since understanding more about how minds really work could well end up producing insights into why such minds value more widgets). For myself, it was one of those life-changing reads.

Next story loading loading..