# Jeepers Creepers Batman! The Singularity is Near.

One of the most recent episodes of Radiolab reports on recent work by Cornell scientists. A robot called Eureqa (you can download her brains here) is capable of taking in large amounts of data and then summarizing laws that govern that data’s behavior. For example, without any prior knowledge of physics or geometry, this robot – equipped with a camera capable of recording data – takes ultra-fine data points and then “learns” what the fundamental laws of the system are. In just a few hours this program derived Newton’s law that Force equals mass times acceleration, or F=ma. When watching a double pendulum it derived Hamilton’s equation with the appropriate Hamiltonian (Energy function) describing the double pendulum. See a great Guardian Article here.

In the above Radiolab podcast they describe how a biologist from Texas then asked the Cornell group to analyze a huge batch of data tracking single-cell dynamics for a specific organism (A Cambridge group did the same with Baker’s Yeast). The machine started printing the “laws” that governed the single-cell dynamics and they were extremely accurate and simple. The trouble is that the scientists can’t figure out what the equations “mean”. This presents a problem as our current paradigm for doing science and mathematics requires that our results be understandable *by humans*. (See Thurston’s article “On Proof and Progress in Mathematics”.)We seem to be moving quickly into a time where we can describe phenomena without comprehending the significance of our results.

This raises several interesting points of discussion. For one, what would a version of this robot carried to its logical extreme have to say about our universe? Given that our current Eureqa has covered centuries of human intellectual endeavor in a few days, just imagine what deep insights it will have in the future. Will it come to the conclusion that our universe is too complex to be described by mathematical laws and that it must be the product of a creator? Perhaps, but regardless of the answer the exciting fact is that we are poised to create our own Douglas Adams’ Deep Thought and this machine might resolve the great dispute once and for all. Alas, it seems that if we built an actual Deep Thought an answer as meaningless as 42 is a plausible outcome.

An Aside: As a youth I was very interested in politics and the structure of an “ideal” society. I then read the article “Seeing Around Corners” (PDF) and summarized that it truly was an analytical problem whose solution didn’t lie in philosophy, but rather in mathematics. I decided to pursue the hard sciences from that day forth. Now as a PhD student studying mathematics I listen to the above podcast and panic. Oh God! Science is dead! I haven’t decided to leave mathematics in part because it is inventive and probably not going to be subsumed by the machines, but if all that is left for future thinkers to do is inject meaning into life Philosophy is looking pretty good!

This Radiolab got me thinking too, but I don’t see it as being quite as profound. I don’t see scientists chasing around after the results that the computers churn out. A result is only of utility if it’s understood, so whilst it could provide insight into why natural laws (as a concept) exist and how they are found, it strikes me that there’ll still be a parallel scientific process going on. They may well influence each other, but I don’t see science dying!

An interesting question. Can a computer program come to the conclusion that there is a creator? Well, humans have. What are our brains other than (very) complex combinations of simple physical processes? How is this different. I guess it’s not too big a stretch to suggest that one day a computer could give rise to similar conclusions.