Part of the I, ROBOT series
For the hundreds of years that comprised the “Middle Ages,” theology was the most cherished and important intellectual discipline, the one that the best and brightest minds worked in, the one that all other fields of study yielded to and were justified by.
We like to think that could never happen again, because Science (with a capital S). But may it happen again, because Technology.
How AI is Different From Other Technological Advances
The new generation of Artificial Intelligence is distinguished from almost every other technical advance in human history in one particular way: this time, we are creating tools without really understanding how they work.
That’s because the most successful AI protocols don’t involve giving a computer a lot of logical rules that, if applied consistently, allow it to handle complex reasoning. Instead, it involves setting up networks modeled on the human brain, providing it clear directives, and then letting it learn on its own.
Amazingly, this works, and in some ways it works really well. But it means that the resulting systems are so complicated and so idiosyncratic that they are “black boxes” – largely impenetrable to outside analysis. We can understand the inputs they get, and we can understand the outputs they produce, but we have an incredibly hard time understanding, let alone explaining, what goes on between A and Z.
If you ask a developer why their AI did what it did, they probably don’t have a good answer – and the more complex and “intelligent” the AI get, the less it will even be possible for engineers to have an answer to that question.
The Future of Unknowing
Often this inability to understand why our most advanced AI do what they do is portrayed in terms of failure: if an AI gets something catastrophically wrong (crashing a plane or sending an innocent person to jail), how can we fix it? How can we stop it in the future? We won’t know.
But success is actually a far more interesting scenario than failure. What if the cheerleaders of AI and “the Singularity” are prophets? What if everything goes catastrophically right, and AI more cognitively powerful and benevolent than anything we can possibly aspire to be emerges? And takes over the systems that human beings manage so badly? What happens then?
Here our sociological imaginations generally stop. It’s easier to envision what might happen if the revolution in Artificial Intelligence goes horribly wrong rather than amazingly right, probably because we live in an era where “progress” is explicitly linked with “disruption.” It’s easier at this point in the 21st century to imagine an apocalypse than it is a paradise.
But let’s consider this seriously. If super “Singularity”-style Artificial Intelligences emerge, which by definition are smarter and more capable than we can ever be, and they take over significant management of many human systems … what happens to the quest for knowledge and education?
When Computers Can Think Thoughts That We Can’t
Science and technology will be revolutionized … by computers thinking thoughts and developing models that we can’t ourselves process. Sociology and psychology will be radically transformed and applied by the Singularity through the use of more data, and more complex data sets, than we’ll be able to understand. Mathematics will be conducted at a level where just expressing and understanding the problems will be a stretch of the human mind.
And all of these major advances and new and beautiful things will be used to make decisions that humans can’t check up on or follow along with because quite simply we are not smart enough to design and understand these systems ourselves. Whole new approaches to science and policy and ethics will be developed which, by definition, humans are not good at – if they can do them at all. And aside from noting what (we think) works and doesn’t work, we can’t check up to see if mistakes are being made.
At some level, we will simply have to take the Singularity’s word that its pronouncements are true and that its policies are right and just. Which almost certainly means that, at some point, much human intellectual activity will redirect from trying to understand science and policy and ethics themselves, to trying to understand the logic behind what the AIs are telling us about them – which we assume, even know, to be true, regardless of whether we can directly confirm them.
Which brings us back to theology.
Code Will Be the New Latin
“Theology” – a term invented by the 12th-century monk Peter Abelard – literally means “the logic of God.” It is an attempt to understand the mind of God, to recognize His patterns (in this context, God was male) and His principles and even predict what will and will not please Him, by studying His pronouncements and works.
Which is exactly what we’ll be doing with the AI. What will please it? What do its decisions tell us, if we understand them correctly? What can we infer about its underlying logic from its deeds and the way it organizes our society? What can we interpret about the underlying nature of reality – on a level we cannot ourselves grasp – by studying its pronouncements?
If you think this is a ridiculous scenario, that we would would never do it, then I’m sorry to have to say we already have. Much in the same way that we’re all too willing to deny other people’s obvious humanity for the sake of economic convenience, we tend to anthropomorphize systems that become too complicated to be followed mechanistically. Exhibit A is our economic systems. Meet “The Market.”
“The Market” has gone from being a handy shorthand for economic activity to an anthropomorphized supreme economic being with a massive industry of followers dedicated to interpreting its signs.
We ask: how is The Market feeling? Is it confident? Is it skittish? Who does it want us to vote for? What policies will please it? How do we stay on its good side?
There’s a whole spectrum of devout believers in the theology of The Market: some are hucksters, some are intuitives who are good at tracking it, some are PhD’s who write detailed treatises on factors that influence The Market and the lessons it has taught us. There are, of course, some atheists who roll their eyes and point out that this is all nonsense – that there is no “The Market.” That it’s a collective delusion developed to represent systems that we’ve created and no longer have any control over and now possess limited understanding of. But “The Market” has emerged and exists firmly in the consciousness of our society, and we seek its blessing and fear its wrath, revere its oracles and try to interpret its omens.
If we’re dealing with AI systems that could literally answer our prayers, the effect would be so much greater.
The Return of the King of Kings
Given a true Singularity, an intelligence far beyond our organic capacities, studying its works will become the center of our intellectual lives. And the questions that raises are, fundamentally, theological questions. To be sure, the history of science and ethics and policy will still be available to us as background materials, but the future of intellectual activity will belong to this theology. Our best and brightest minds will set themselves not to understanding science and ethics – which they can never personally advance – but to better theorizing about the mind and works of the AI, and explaining them to other humans.
Everything old is new, I guess.
This isn’t necessarily bad news: theology gets a bad rap in modern intellectual circles, but in fact it can epitomize the best qualities of intellectual work and critical thinking. You have to understand logic, study context, determine what counts as evidence, and weave in all aspects of the observable world into a coherent framework. That’s pretty much the definition of “critical thinking.” When your theology involves an Artificial Intelligence, you’ll probably also study a lot of computer science. There are far worse intellectual fates. In fact, for bright minds it probably sounds like fun.
But it will be theology: its center will be not understanding the world or even our own selves, but understanding what the Singularity tells us is true, the revealed word of the ultimate intelligence. Humanity’s greatest accomplishment will be the creation of a tool that, if we are lucky, we will get to spend the rest of our time as a species trying to study the world through its eyes.
That’s what success looks like, if AI lives up to its promise.
Do we like it?
What are the alternatives?