Caleb Ontiveros

How Philosophy Will Influence Technological Development

What philosophical work should you do if you want to influence the rate of technological development?

This question is related to more fundamental questions. Like:

  • What sort of philosophical work should be done at all?
  • If you want to positively influence the world, what sort of philosophical work should you do?
  • What philosophical research program will most influence the trajectory of civilization?

I'm ultimately more interested in questions like the first two. But one way to make progress on them is to ask descriptive questions, leaving the normative aside. That's what I'll do here. Of course, there is obviously important work for philosophers to do other than influencing the rate of technological development. I see asking the question, “what philosophical work should you do if you want to influence the rate of technological development” as being useful for other goals. Influencing the rate of technological development isn’t all that by itself.

I’ll start by clarifying the question, given examples of influential philosophy, and then speculating on what kind of work is the most promising.

Let's clarify the question.

By influence I mean that there should be a causal connection between the philosophical work itself and a change in technology. The causal chains for technological development are complex and speculative, but in order for something to be influential we need to be reasonably confident that it played a non-trivial role in the chain. Note that it should be insights from the philosophical work itself, not other considerations that play the causal role.

By philosophy, I'm talking about the kind of philosophy that solves descriptive problems. I'm not talking about these kinds of philosophy:

  • Performative philosophy, where the point of philosophy is to ask questions or express some aesthetic.
  • Historical philosophy, with one caveat that I talk about below.
  • Moral and ethical philosophy.

A good first pass of what I mean by philosophy is expressed by Scott Soames:

Philosophy is the partner of every serious intellectual discipline. It appears when enough is known about a domain to make new progress conceivable, even though it remains unrealized because new methods are needed. Philosophers provide new concepts, new interpretations, and re-conceptualized questions that expand solution spaces.

Philosophy provides the conceptual foundations for theoretical knowledge. Examples below will make it clearer what I mean.

You can influence the rate of growth in two directions: faster or slower.

By technology I mean the kind of things that take us from 0 to 1. Artifacts and processes, from computers, vaccines, microscopes, bombs, and more. I'm not talking about "social technologies" or "cultural technologies", such as institutional designs or group decision procedures like companies and democracy.

By development I mean technological change. It need not be change for the better, though it would be better if it was!

Influential Philosophy

Some philosophical research programs have influenced the rate of technological development. How much influence they had is unclear, but I'm confident that some philosophical work has mattered. I'll give three examples: computation, probability, and predictive philosophy.

Computation

Many people were involved in the creation of modern computers. Babbage, Lovelace, Turing, Von Neumann and many more. The founding text of the architecture of modern computers is “First Draft of a Report on the EDVAC” written by Von Neumann and the EDVAC design team. However, it’s been asserted that Von Neumann "repeatedly emphasized that the fundamental conception was Turing's."

Alan Turing's Turing Machine is essentially a conceptual analysis of computation. What is computation? From the Stanford Encyclopedia of Philosophy, "A mathematical problem is computable if it can be solved in principle by a computing device." What Turing did is give us a useful abstraction for a computing device. In addition to being a philosophical success, this played a non-trivial role in the development of computers.

As an aside, how much of a role this kind of work played is unclear. It could be that it only sped up the development of computers by a few days. Von Neumann himself may not have been influenced by Turing’s work. Note however, that the intellectual waters that Turing, Von Neumann, and others breathed were made up of work from a rich philosophical tradition going back to Pascal, and including Frege, Godel and Hilbert. The cumulative results of this intellectual tradition likely sped up the development of computers by years, maybe even  centuries. It's not clear. For many inventions, like calculus, it seems like the inventors may have only sped up the development by a few years, likely less. Yet there are many artifacts that have taken a surprising amount of time to invent -- which suggests that invention and intellectual discovery are usually very hard and that humanity can go centuries without discovering something that will eventually be taught to preschoolers of future generations.

Coming back to computation, someone may push back on this work and suggest that it's not philosophy, but logic or computer science. This objection is semantic, but worth addressing. First, Turing’s work is captured by the earlier definition of philosophy. Turing did a conceptual analysis of computation. Conceptual analysis is at the bedrock of what philosophers are up to. Second, it’s pretty close to what a lot of philosophers who specialize in logic do. Of course, one can object that what they are up to isn’t philosophy either but at this point this objection becomes too semantic to me.

Probability

Our concept of probability has impacted several intellectual programs, influencing financial markets, to scientific models, to artificial intelligence. Many philosophers have done important work here. Here are two. Bayes’ Theorem from our friend Reverend Bayes. If you're reading this essay, you likely know what that is. If you don't, go here. It’s mattered for a number of fields, both in terms of establishing epistemic standards, and also proving useful for the work itself, as in artificial intelligence and neuroscience. Another example is work done Frank Ramsey. In the early 19th century Frank Ramsey applied probability to the question of what it takes for an agent's beliefs to be rational. His analysis has impacted how we've modeled agents in economics, psychology and computer science.

Where work on computation verges on the logical, philosophical work in probability verges on the mathematical.

Predictive Philosophy

Finally, one can impact the rate of technological development through predictive philosophy. Nick Bostrom's work, like Superintelligence, is a paragon of this. Way back in 1997 Bostrom wrote a paper called "Predictions from Philosophy? How philosophers could make themselves useful". In this paper, he argues that there's a role for a philosopher-polymath. That is, someone who can bring together work from a number of fields, like philosophy, economics, physics, psychology, math and more. In the essay, Bostrom highlights anthropics, the Fermi paradox, superintelligence, transhumanism, uploading, and cosmology as areas for potential philosophical research -- areas he's worked in since writing that paper.

Wilfred Sellars said that:

The aim of philosophy, abstractly formulated, is to understand how things in the broadest possible sense of the term hang together in the broadest possible sense of the term.

Roughly, the aim of predictive philosophy is to understand how things in a narrower sense of the term hang together and in order to make useful predictions. You could also think of this kind of philosophy as “generalist philosophy” or “polymathic philosophy”, but I prefer the term predictive philosophy since it says more about the methodology.

Though it verges on the normative side of things, Toby Ord's The Precipice is an excellent recent example of this work. Ord is concerned with existential risks to humanity. Here’s a question: what’s the probability that human civilization will be destroyed by a natural extinction risk? On the surface, this may appear to be a strictly empirical question, not something that philosophers would be well suited to answer. This perspective overlooks that is that there’s a lot of uncertainty here and non-trivial questions about how to assess the probability. Do you look at the rate of species like us going extinct? Do you only include mass extinctions or should extinctions of any species like us contribute to the rate? Do you extrapolate the rate from the fact that we haven’t become extinct yet?

In a way, I see this work as similar to philosophical work on the likelihood of the existence of God. Empirical inputs matter a lot (facts about the fine tuning of the universe for example), but there's so much uncertainty that a lot of these issues depend a lot on debates over how priors should be set, reasoning about anthropics, and the nature of probability and explanation. To be clear, I’m talking about probabilistic work by people like Swinburne and Oppy, not ontological arguments. It’s not entirely by coincidence that John Leslie was interested in both existential risk, cosmology, and traditional religious debates.

Promising Philosophy

These are the kinds of philosophical work you should want to do if you want to influence the trajectory of technological development:

  • Formal philosophy
  • Predictive philosophy

The examples of computation and probability are instances of formal philosophy. Superintelligence, The Precipice, and The End of the World are instances of predictive philosophy.

I’ll say more about the methodology and domain selection for these two kinds of philosophy.

Formal Philosophy

With formalizations, the work is more obviously philosophical. Though it has tended to be, and will likely continue to be, on the borders of the logical, mathematical, and conceptual.

When philosophers successfully formalize a concept, sometimes that formalization will directly contribute to the development of technology (as, say, in the example of computers and Bayes’ theorem) or it may indirectly contribute by improving our cognitive capabilities (as,say, work on probability has).

There are two things I'm interested in here: methodology and domain selection.

The methodology is informed by the examples of computation and probability. Philosophy here will often look more like math, logic, and computer science than some philosophers may be used to. The area of philosophy most closely approximating this style of philosophy right now, unsurprisingly, is formal epistemology.

Formal methods aren't enough, one needs to select promising domains. What you're formalizing matters. A large portion of formal epistemology will not influence the development of technology, or likely anything else for that matter, because it's aiming to answer questions whose answers will make no difference to other fields.

There are a few features of this kind of philosophy and its domains that make it more likely that work is promising:

  • The work is mathematical or logical. The historical record suggests that this work matters more for technological development.
  • The work answers foundational questions of a given field. One can imagine a graph of concepts, philosophers want to analyze the ones with the most edges.
  • The work is problem-focused. If we knew the answer to a question or had a refined notion of a given concept it would be useful for solving non-philosophical problems.
  • The work is positive. It puts forward a way the world is, rather than falsifying philosophical theses. There are some potential counter-examples to this, like Godel’s work, but by and large, the philosophy that matters is offering a positive vision of the world that we can do something with, not shooting down an existing framework.
  • The domain itself is speculative and relatively nascent. There’s a lot of uncertainty.

Domains where formalization could be useful are:

  • Modeling agents and decision theory

Work on newcomb-like problems may matter for the development of artificial intelligence.

  • Causation

Most of what academic philosophers have said here isn't useful. Computer scientists and statisticians have done most of the important work to date, but there's no reason philosophically oriented computer scientists or statistically oriented philosophers can't make important contributions here.

  • Philosophy of physics

Most work I’ve seen here just looks like the metaphysics of physics, but I could see work here being promising. I don’t know enough about this physics or the philosophy of physics to have a strong view here.

  • Cultural evolution

This work is unlikely to make a significant difference to technology, but is novel enough that philosophers can continue to make important contributions here. There are sociological questions here around why philosophers' work on cultural evolution hasn’t been more influential.

  • Evolutionary biology

Unlikely to make a significant difference to technology, but worth mentioning since it is an area where scientists should pay more attention to philosophers. Unfortunately, the same philosophers who make important contributions also get caught up in issues that don't make any difference to scientific practice.

  • Information theory

There may not be that much work to do here anymore, I don’t know.

  • Consciousness

Work on the hard problem of consciousness likely doesn’t matter for the development of technology. However, thinking about how to formalize conscious states, debates over global workspace theory and its competitors, or the meta-problem may matter.

  • Neuroscience

Carl Carver has done interesting work here. One aspect of neuroscience that makes it promising for philosophers is that no one has any idea what is going on.

Some of these domains are orders of magnitude more promising that others. The probability that any philosophical research program will result in contributions that impact that development of technology is small. That, of course, doesn’t mean that the effort isn’t worthwhile.

Predictive Philosophy

I’ve sketched out the methodology for predictive philosophy when sketching it out above. There’s more to say about it, but I’m not sure if being more precise would be that useful.

There are a few features of this kind of philosophy and its domains that make it more likely that work is promising for philosophers:

  • The work verges on the logical, mathematical, and scientific. It’s the sort of thing you could almost get away with outside of a philosophy department.
  • The domain takes inputs from a number of different fields. Being a polymath or generalist is required to make a contribution.
  • The domain is relatively neglected.
  • The work is problem focused. If we knew the answer to a question or had a refined notion of a given concept it would be useful for solving non-philosophical problems.
  • There’s a lot of uncertainty. Positions may be determined by nuanced views about probabilities and reasoning under uncertainty.
  • The work has both philosophical and non-philosophical questions at its center.
  • The work analyses the history of technological and scientific development for useful lessons.

There are a number of promising issues here:

  • Emulations and digital uploading

These issues have philosophical and non-philosophical questions at its center, making the work pretty promising.

  • Genomics and gene editing

The uncertainty here makes it potentially promising.

  • Existential risk

See The Precipice. It includes a section for useful future research, many of which are better answered by scientific specialists, but a few that philosophers can contribute to.

  • Artificial intelligence

See Superintelligence and related efforts.

  • Thinking about thinking well

Philosophers and non-philosophers have tackled this, but there’s still so much work to do.

  • Progress Studies

Progress studies are explicitly focused on economic and technological progress. Philosophers of science are especially well suited to contribute here, see below.

  • Philosophy of science

Much of philosophy of science like the metaphysics of science or descriptive history won’t influence science of the trajectory of technology. However, some it can, and potentially has been useful. Mining for applicable lessons, analyzing institutional designs, and coming up with novel systems that change the development of science and technology are quite promising.

  • Philosophy of technological development

Yes, the meta matters. What kinds of philosophy have pushed the peanut of technology forward or backwards? What kind of philosophy should we do if we want to have an impact on technology?

It’s worth reiterating that I’m not suggesting that slowing or speeding up the rate of technology is what philosophers should do. Most of the work above is worth doing, some of it is much more important than others and there’s valuable kinds of work that I’ve left out.

The main idea, that philosophy can and has influenced the development of technology seems right to me. Philosophers, as conceptual engineers, can be more ambitious.

If this is something you'd like to do, reach out to me.

Thanks To Evan Gaensbauer and Arya Sabeti for reading and commenting on earlier version of this post.