Veritas (सत्यं)

We must ask ourselves what is true, if anything. At the limit of our understanding,

there is no forever in this material plane. Our best understanding of physics suggests that our universe has an end date, albeit still quite far out. Humans have long sought eternal life. I do not believe that any of us can speak clearly as to how we might exist in the spiritual or transcendent realms, so I will focus my analysis on what is true on the material plane, this immanent realm.

Harvard’s motto is Veritas, or “Truth.” In Parallel Ontologies, I acknowledge that there can be many truths, in line with Jain epistemology. But an urgent order of business is aligning humanity and our beliefs. This is an extraordinary challenge with such a diversity of epistemologies competing amongst us, but there are some truths we ought to agree upon now before it is too late.

The first truth is that capitalism will eat itself. I am in no way an anti-capitalist. My previous job of a decade had capitalist in the title, and I am in no way ashamed of that. A popular axiology in academia wants to cast capitalism as good or evil, but I think is neither. It is simply a system with emergent features, some of which are good and others bad. Capitalism has proved extraordinarily effective at organizing human means of production, which has improved the health and prosperity of many. Capitalism has proven equally blind to its externalities and any morality of profits, pushing humanity to the brink of survival. Too often, the long-term analysis of capitalism is rooted in our biases. Those who like capitalism are irrationally convinced it will last forever, and that infinite growth is plausible on the argument that humans have infinite demand. Those who dislike capitalism are convinced it is already dead. Both are wrong.

As I already noted in Real Talk on Markets, Money, and Labor, the mechanism that will cause capitalism to fail is the value of labor asymptotically approaching zero due to near-total automation driven by advanced artificial intelligence. The result of this will be wide-scale unemployment that is destabilizing to countries, depending on the emergent societal equilibrium resulting from wide-scale economic cooperation in a capitalist economy. This inevitable destabilization will prove an extraordinary threat to any democracy. We should assume this will happen relatively soon. The terminal value of most publicly traded businesses, especially those driven by information assets and knowledge arbitrage, will asymptotically approach zero. As most of the savings of the wealthiest people in the world are tied up in these assets, with a solid majority of them being based in the United States, the work will soon experience an unprecedented disaggregation between money and power. In short time money will lose the meaning it has today, but we should not naively assume equality: power will still be a highly relevant force. We must prepare for a world that could have anything from equality to hierarchy and assume that the balance of power in the world might pivot in wildly unpredictable ways, depending on who controls what, especially AI.

The second truth is that we will be able to extend the human lifespan indefinitely. That is not to say that our fleshy bodies will be immortal but to say that they will be amortal. One might still die from being hit by the proverbial bus, but a single human could live thousands or millions or billions of years, up to the limit that our physical universe allows. If we can upload our consciousness, which I am less totally convinced will carry our true being but could at the least represent a copy of perfect fidelity, a la the discussion oft-posed in an introductory undergraduate philosophy course, we might extend that life by some orders of magnitude, to the utter end of universe with some exotic time dilation to boot. Obviously, with so much of the human experience being tied to thecycle of life and death, this poses an extraordinary set of design questions for future human societies. Will we choose to die even though we don’t have to? How will we deal with birth? With indefinite lifespans and continuing birth, the Malthusians would finally have their day. With each new possibility will come new tradeoffs, and we will have to shift our values accordingly. These will not be easy transitions as we will be forced to let go of elements of the human experience we have been deeply attached to for as far back as our collective memory goes.

The third truth is that our knowledge will likely reach an asymptotic peak. I suspect superintelligence will reveal this to us relatively soon. I personally believe that there is an infinite amount of knowledge to be discovered. This is wild speculation on my part, but I suspect matter is infinitely indivisible. Physicists will thus always have something to do. As we zoom in, we have an infinite. But as we zoom out, I suspect we will find the universe has a limit that we can keep approaching but never cross. That implies that all problems that we can conceive of are solvable with sufficient intelligence, but that there may not be many more solvable problems left. Superintelligence may not yield any greater potential consciousness than we already have, so we must let go of our attachment to the infinite expansion of knowledge. Just as the infinite growth of the economy is implausible, so too might be the infinite growth of wisdom, though on this point I am admittedly not one hundred percent certain.

The fourth truth is that we will not be able to know with certainty whether the artificial intelligence we create has a consciousness like our own. Per Susan Schneider’s talk at our very own Humanity Meets AI, the way that we escape solipsism in understanding that humans are also conscious beings like ourselves is through parallelism.³⁰ It is not just that other humans pass the Turing Test, but also that we see that their physical structures are like our own. Today, the argument against transformer models having consciousness is largely rooted in structure. They do much simpler calculations than our own minds, and do not mechanistically look to be aware in the complex and dynamic ways that our own minds are.³¹ Today, it is easy to argue that an AI that can pass the Turing Test is not conscious. But in the future, not only will it not be easy; it will be impossible. We will have to decide how to treat beings that will seem to be just as conscious as we are but that have god-like superpowers relative to us; beings of our own creation, as if we, rather than god, have created Aquinas’ angels and demons.³² Though if we are god’s creation, then perhaps so too are our creations.