Sunday, December 16, 2018

Artificial intelligence and the limits of the machine model

In his bestselling book, Up the Organization, former Avis president Robert Townsend captured the problem of automation precisely. Writing at a time when the vast paper systems of corporate America were being transferred to computers, he warned that it was important first to make sure that a company's paper systems are actually effective and accurate. "Otherwise," he quipped, "your new computer will just speed up the mess."

Today, we are faced with a new wave of optimism about the prospects of what is called artificial intelligence (AI). Just to be clear, here is what the originators of that term meant by it:

[T]he basis of the conjecture [about artificial intelligence is] that every aspect of learning or any other feature of intelligence can in principle be so precisely described that a machine can be made to simulate it.

It is important to parse these words carefully for they will tell you why artificial intelligence as it is currently conceived will very likely "just speed up the mess."

First, whatever we decide to hand over to artificial intelligence has to be described in terms that a machine can implement. The machine is the primary limiting factor because the world itself must be modeled as a machine so that a machine can manipulate it.

This might be fine for some tasks. Computers have long done mathematical operations at speeds and volumes that no human could ever hope to emulate. But we now say that machines can "learn." But "learn" in this context really means improve performance in furtherance of goals that we humans set.

And discussion of goals gets us right back to the question of what we will be asking so-called AI systems to do for us.

Let me seemingly diverge for a moment. Since moving to Washington, DC, I have found that electronic access to people is overhyped. Anyone with an internet connection and a computer has had access to anyone else similarly equipped across the planet for many years now. But Washington is a city that attracts very smart people from all over the world to live or visit here for obvious reasons. The single most important change in my life since moving here is physical proximity to such people.

These people were always theoretically available to me electronically. But I didn't know who I didn't know until I bumped into the innumerable people who have been willing to meet with me in a city that accepts personal networking as a way of life in the same way that a fish accepts water as a way of life. Meeting in person, conversing in person leads to a much more dense information flow than any other type of encounter.

My point is that no machine can copy that. And, no experience mediated by a machine (email, video chat, conference call via internet, etc.) can match it. Strangely, this is something that one of the world's arch-promoters of the networked world knew only to well. A contributor to the Harvard Business Review wrote the following about Apple Computer founder Steve Jobs:

Jobs was a strong believer in face-to-face meetings. "There’s a temptation in our networked age to think that ideas can be developed by e-mail and iChat," he told me. "That’s crazy. Creativity comes from spontaneous meetings, from random discussions. You run into someone, you ask what they’re doing, you say ‘Wow,’ and soon you’re cooking up all sorts of ideas."

Jobs designed the offices of Pixar—the hugely successful animated film company he developed from a division purchased from Lucasfilms in 1986—to maximize personal interactions. Serendipity was the watchword. How exactly do we program that into a computing machine, an apparatus which Robert Townsend once characterized as a "fast, dumb, adding-machine typewriter"?

There is one other thing we should note about AI. Computer scientist Larry Tesler explained that AI was nothing more than "whatever hasn’t been done yet."

So, AI is a more sophisticated term for "speeding up the mess." AI is supposed to increase human capabilities. This National Geographic article asks if AI will help us cure cancer. To me this is the wrong goal. We ought to try to prevent cancer from ever happening. The first goal accepts current arrangements and power relationships. The second calls for a radical reordering of those arrangements and relationships. You can bet that the corporations working on advancing AI won't be focused on the second goal.

And yet, the myriad environmental challenges we face call out for a vast and rapid reordering of society that simply cannot come from any machine that we endow with artificial intelligence. Yes, AI might make existing processes more efficient and effective than they already are. But should we be trying, for instance, to be more efficient and effective in extracting the planet's nonrenewable resources such as fossil fuels and other minerals?

The notion that we will create human-level intelligence in machines presupposes that human intelligence comes from a machine called the brain. But human intelligence is actually the cognitive result of a connection to a field we call the world. Our hands and eyes and skin are part of that field. Everything around us is part of that field. The interactions I have at meetings with others is part of that field. The unspoken, unwritten communication between fellow humans and between humans and nonhumans cannot be "so precisely described that a machine can be made to simulate it."

Not all information can be rendered into words. And, even if we could do that, words are a surprisingly imprecise way to convey meaning, freighted as they are with nuance, cultural context, history and so many other interlocking dependencies for their meaning.

The huge problems that we face—climate change, fossil fuel depletion, deforestation, fisheries decline, soil degradation, water scarcity—will not be solved by AI. They have and will continue to be partly caused and aggravated by computers which have made us all the more effective at harvesting what we want from the planet in order to provide for the growing human population and the unquenchable appetites of consumers.

It's a tired cliche that technology is a double-edged sword. But it seems we are programmed by our machine culture to remain ever hopeful that the latest technology will only lead to solutions rather than to more problems.

If we reduce all of our efforts at addressing our problems to language a machine can understand, we will get machine solutions. What we need, however, are solutions that come from our deep connections to this planet as beings of this planet, connections that no machine will ever fathom.

Kurt Cobb is a freelance writer and communications consultant who writes frequently about energy and environment. His work has appeared in The Christian Science Monitor, Resilience, Common Dreams, Le Monde Diplomatique, Oilprice.com, OilVoice, TalkMarkets, Investing.com, Business Insider and many other places. He is the author of an oil-themed novel entitled Prelude and has a widely followed blog called Resource Insights. He is currently a fellow of the Arthur Morgan Institute for Community Solutions. He can be contacted at kurtcobb2001@yahoo.com.

2 comments:

sv koho said...

Kurt, you might need to moderate the comments. BTW we readers would like to hear why you moved to DC.

Unknown said...

Think of machine intelligence as next form of life. At first stages it will help humans, at later stages it will replace them. Human brain is so limited and the human itself is so inefficient what this process is inevitable.