The Enlightenment’s Wager: Intelligence, AI, and the Open Future

The Wager

In my recent essays, I have been pursuing a conjecture: that intelligence is the expanding an adaptive capacity to make the future less like fate and more like an open field of possibilities.

No conjecture appears out of thin air; conjectures are oriented by, with, from, and toward purposes and problems. The problem that orients my conjecture is this: we are placing a very large wager that intelligence can be engineered and that this can lead to breakthroughs in fundamental problems that frame our present—climate change, disease, hunger, clean and abundant energy, democratic governance, turning scarcity into abundance.

If this wager is to pay off, it will require rejuvenating the Enlightenment’s orientation to broad-based progress through creative problem-solving. We then need to understand the Enlightenment not merely as an historical era that has come and gone, but as a crucial expansion of life’s ability to shape the future—i.e., a major change in the evolution of intelligence.

AI appears to be an opportunity to renew this orientation, but it seems to have entered a polarized trajectory of history in which one side seeks the idealized past of a hard-coded ‘again’ while the other is still reeling from the gut punches in 2016 and 2024 that demographics are not destiny. AI enters this lack of coherent orientation to history and oscillates between the either/or of apocalypse or abundance. This seems to me to be the opportunity and need for rejuvenation of the Enlightenment.

The Enlightenment’s Calling

The Enlightenment was a reorientation to fate and time. It was a calling and a wager: if we reorient knowledge production to better explanations (and away from justifications of authority), we could gain the upper hand in our long confrontation with those forces that were always assumed to be irreducibly beyond our control—hunger, disease, wars, boom and bust economic cycles, violent weather, tyrannical governance. (This is largely the view of David Deutsch in The Beginning of Infinity.)

How do we continue to hear the call today? How do we hear this call without seeking an echo chamber amid the noise that surrounds us? Echo chambers are crucibles of ressentiment—the wrapping of ourselves in our own virtues while denigrating the world as irretrievably fallen.

Engagement is necessary if we are to achieve broad-based prosperity rather than only faster package delivery, increasing wealth polarization, more addictive and divisive data feeds, and other fruits similar to those of the late-twentieth century’s digital revolution.

What does an engagement look like that believes the Enlightenment’s calling remains relevant? This is what this essay seeks to outline.

Accompaniment and Storytelling

The engineering effort of AI requires the essential and important work of defining what intelligence is. As Deutsch writes, ‘if you can’t program it, you haven’t understood it’ (Beginning of Infinity, 154; italics are his). There is a great deal of insight and importance in this assertion (though I wonder if it could be reversed to achieve a broader understanding). As we experienced with the Enlightenment, broad-based progress is not likely without successful engineering efforts.

I also find myself compelled by Blaise Agüera y Arcas whose What Is Intelligence? ambitiously weaves a story of intelligence emerging from life. Of the many innovations of the book is the assertion that intelligence is recursive computational modeling that can infinitely learn and expand to shape more and more of the future. I’m inclined to agree with much of his analysis, especially his final thoughts on the dangers of monocultures.

Both of these strike me as quite good explanations for their chosen scope. Like all good explanations, they arise from stories but also should become the springboard for better and richer stories that we can tell about the past and orient us to a better future.

These stories accompany these explanations. By ‘accompany’ I draw from the tradition of spiritual direction where the director helps the directee discern the voices of angels from those of demons. In a more secular vein, accompany could mean ‘riding shotgun’. In both cases, accompaniment involves watching for the dangers while scanning for the genuine opportunities without occupying the role of conscience, priest, or therapist. Accompaniment covers the directee’s blindspots because it realizes that any pursuit of progress creates winners and losers. The history of Enlightenment progress is littered with as many innovations of violence as productivity—often deeply entangled if not inseparable.

Again my conjecture: intelligence is the expanding and adapting capacity to turn fate into possibility. Perhaps someday we will no longer have need for the word intelligence because its limits will have been left far behind. In transition, we may be talking about multiple kinds of intelligence, or we may collapse what Modernity has placed outside its boundaries and labeled as irrational or emotional—anything not reducible to scientific and instrumental rationality.

For the moment I’ll ride alongside those who are vested in the definition.

Leaps of Faith

I find Bergson’s perspective still relevant with respect to the exploration of the limits of intelligence treated as a concept:

Swimming is an extension of walking, but walking would never have pushed you on to swimming. So you may speculate as intelligently as you will on the mechanism of intelligence [i.e., treating it as an idealized concept]; you will never, by this method, succeed in going beyond it. You may get something more complex, but not something higher or something different. (Creative Evolution, 193).

To get from walking to swimming, one must take a speculative leap of faith. One must speculate (conjecture, imagine) that the action of walking can become swimming but not as a difference in degree but of kind. The only way to find out is to leap from the land into the water. ‘You must take things by storm: you must thrust intelligence outside itself by an act of will’ (193).

A new Enlightenment should orient us to the possibility of swimming when our explanations have only considered walking. We make the leap into the water—to something more rather than the increasing complexity of something already conceptualized and assumed to be explained. We do so by accompanying engineering so that error-correction and back-propagation can become greater aspirations.

We cannot do so with polemics but only by genuinely exploring limits of what we believe we have already explained. Therefore, fallibilism is a necessary humility in our explanations. But fallibilism alone risks the stasis of a skeptic who would prefer never to be in error than to venture into the unknown. As Deutsch puts it, fallibilism must be coupled with optimism.

As much as we like to believe that the Enlightenment sought the certainty of knowledge as the basis for all progress, it did not. It placed its bets on an orientation that preceded the certainty of truth: that the open-ended improvement of the human condition could come about by seeking better explanations.

Time and Enlightenment

In the midst of the Enlightenment proper, James Hutton’s stared into the stratified rocks of the British landscape and saw ‘the marks of marine mammals in the most solid parts of the earth’. This didn’t compute with Biblical reckonings that posited a fixed Earth that was only about 5700 years old:

If, therefore, we knew the history of those solid parts, and could trace the operations of the globe, by which they had been formed, we would have some means of computing the time through which those species of animals have continued to live. But how shall we describe a process which nobody has seen performed, and of which no written history gives an account? (Theory of the Earth, Part I, 1788)

Not only is time being extended infinitely backward, it is becoming an open-ended wager that we can make sense of its processes—processes unseen and unwritten because they extend well before us.

The temporal relationship of knowledge to time is crucial and must be understood. Hutton sees the potential for new and better explanations—why are these marine fossils embedded in this landlocked terrain?—because he sees (with Newton) the possibility that nature’s process are computable.

To answer this conjecture, he has no texts to turn to—only the landscape itself and the assumption that it is the result of measurable and computable processes. Here is where the modern age begins its acceleration, by slowing down time and pushing it into an indefinite past that will yield an equally infinite future.

The orientation that believes that better explanations are always possible cannot always wait for certainty in order to proceed. If there is truth to be discovered, it is downstream from where Hutton stands, and he wagers that his conjectures can be true as a result of producing new knowledge. But this new knowledge won’t come without turning time into stories and thereby orienting to a future whose truths must be lived as ongoing conjecture.

Function and Orientation

Hutton sees the evidence of water where human intelligence up to that point would have only seen permanent land since the Creation. It invites a leap that must undo the assumed nature and duration of time—beyond the 4000 B.C. consensus that dominated Biblical calculations at the time.

The leap into the water requires using our intelligence to understand its own inherited limits. Our analytical bias is to read the complexity of motions as the expression of stable things—from Aristotle’s teleologies to Newton’s mathematical laws to Einstein’s block universe. Since Aristotle, we have had a bias toward seeing function as causing orientation—telos is embedded in ergon. In other words, orientation since Aristotle (strengthened by Scholasticism) has had no explanatory status independent of functional determinism. The function dictates the orientation.

Analytical intelligence needs this bias. It sees a world of things with essences. Knowledge seeks clear definitions of these things, which can be essential functions, teleological orientations, or stable objects existing unchanged across time. Let’s call this the principle of grounded orientation. Aristotle, Aquinas, Descartes, Kant, Heidegger, and most of cognitive science share this principle. They disagree only about the nature of the ground. The ghost of Aristotle’s unmoved mover often lurks in our will to knowledge.

Riding shotgun does not mean criticizing these functional concepts, nor does it mean declaring them wrong. It means attending to their blindspots. If the driver is concentrating on the road and the route, his vision is narrow and oriented toward the assigned destination. The shotgun rider attends to the surrounding motions, reading patterns of possibility in the surrounding landscape. Both are necessary for a safe and effective journey.

If we are to understand intelligence in a way that extends the engineering efforts, we will need to keep orientation and function separate, and we need to retain the primacy of motion without an unmoved mover. As we’ll see, intelligent orientation requires contingency and disorientation not as bugs in the code but as necessary conditions for the future to be opened to multiple possibilities.

Function will need to be despecialized so new orientations can emerge.

We will pick that up next week.