Philosophical Progress in an AI Era
An Opportunity
Philosophers don’t solve problems. People expect us to. But we don’t. Philosophy is abound with philosophical problems. And we appear to try to solve them. But we don’t succeed. Ever. And we don’t really seem to care.
What care about is exploring solution space. We think about possible solutions to the problems that vex us. Failing to reduce them absurdity with our mighty logics, we weigh their merits against their drawbacks. Sometimes we learn things. We can learn about the connections between, and limits of, our starting assumptions. But that’s about it. We study the structure of assumptions. These assumptions arise in an effort to solve problems — to wit, our exploration of solution space. But simply making assumptions does not solve problems. It does not produce unconditional knowledge. It does not produce unconditional truths.
Why? Because to produce such a truth, we must have some understanding of ‘true’. And there is no consensus on what ‘true’ means. We are forced to glob on to what others mean by it, if we opt to speak with them. Once we say what we mean by ‘true’, every truth we acquire is true only conditional on that, or some other, meaning. And we could have meant something else by ‘true’. And some of those alternatives are no worse than what we in fact mean; our ways of speaking of “true” are no better than some alternative meanings.
It is no wonder that so many people dismiss philosophy as a failure: as science’s loser little brother. Big brother sometimes mocks us: “Still working on the Ship of Theseus problem, huh? What has it been, a couple of thousand years?”
Such mockery rests on a mistake. The public has been led to believe that progress in philosophy is like progress in science. Science progresses when scientists come to a consensus (or something near enough). Philosophers very rarely come to a consensus over any issue worth philosophizing about. And it’s a good thing too.
Consensus is the death of philosophy. When consensus is reached, the machinery of philosophy grinds to a halt. The issue is no longer worth discussing; it has lost its value. And if it’s valueless now, one wonders why we ever thought it had value. Consensus breeds disinterest. The value of philosophy is the value of interest. It’s interesting! Philosophical value is not derived from problem-solving ability. It’s derived by its ability to evoke wonder and puzzlement of a most pleasurable sort.
Philosophy thrives on problems and dies by its solutions. Science thrives on solutions and dies by its problems. For science, the solutions are what’s valuable; the solutions are what’s interesting. Scientific value is derived from problem-solving ability. In this way, philosophy and science are very different kinds of intellectual enterprises. We should not confuse one for the other. They are not contiguous. They are more like opposites. They pull the epistemic state of the world in opposing directions: science fills the holes in our knowledge, while philosophy pokes holes.
Philosophical progress is made when problems are discovered, not solved. The problems are interesting. And the interest is indicative of value. It’s the intrinsic value of philosophy; the instrumental value is that studying it improves one’s clarity of thought like nobody’s business.
Problems aren’t the only source of interest in philosophy. Philosophy is full of a variety of philosophical objects: problems, paradoxes, questions, theories, arguments, objections, rebuttals, methods, logics, distinctions, thought-experiments, heuristics, etc. These are the things for which we read and discuss philosophy. These are the things which generate interest. They are wherein the value of philosophy.
Science progresses when it discovers the things it values: the solutions that generate consensus. Philosophy progresses when it discovers the things it values: the philosophical objects that generate controversy. Philosophers, after all, are those who ask and answer philosophical questions: questions whose answer is not obvious, appear to be appropriately addressable using our faculties of reason, and which generates controversy despite that a correct answer has no obvious practical application. The controversy indicates interest. Interest, I say, is the value of philosophy. Philosophical progress, therefore, is the discovery and production of philosophical objects: objects that attract philosophical interest; objects that generate philosophical value.
Artificial intelligence, some say, has no philosophical value. Some say that it poses an existential risk to philosophy. Philosophy, as it is currently practiced, largely consists in the orderly spewing of words in response to a philosophical question or remark. Good philosophy amounts to putting those words in exactly the right order. And AIs — specifically, large language models — are becoming increasingly adept at doing exactly that. I predict that, not too far in the future, AI models will produce, in minutes, a work that could have taken a professional philosopher months; or a book that would have taken years.
But, contrary to what many people think, this is not the death of philosophy. Philosophy only dies when consensus is reached. And so, AI is not an existential threat to philosophy. It is only a threat to the philosophy publishing industry, and to academic philosophy as currently practiced and as its practitioners are currently incentivized. Human authored philosophy books and papers, I expect, will be valuable only anachronistically: because they were written by a human, like in the old days. More comprehensive works, and deeper insights, (I expect,) will arrive to us by, or assisted by, AI philosophers in a setting — using a medium — that, unlike paper, is equipped to handle their power.
This hardly sounds like a threat to philosophical progress. Not to me. I think it will enhance philosophy. AIs, whether performing the role of philosopher or philosophical assistant to a human philosopher, can hasten the discovery and production of philosophical objects. With the help of AI, we can create more problems, more objects, more theories than ever before.
In this way, ThePhilosopher.AI intends to make more philosophical progress than any other time in history: we intend to scrape every discourse — as much of human philosophical corpus (or corpse?) as we can legally scrape — of its philosophical objects, and then to use a series of AIs to recombine those objects to generate new ones. Endlessly.