Artificial intelligence is one of my strongest interests: the ongoing research in the topic is achieving extraordinary results, and it would seem like we are now approaching a turning point that might completely revolutionize this incredibly exciting field.

But when I first heard that NASA is planning to land an AI on Mars, I was just astonished. As technologically advanced as NASA can be, I didn’t believe we were already in the position to talk about something quite like this, as a complex solar-powered AI would mean that the American aerospace agency is well beyond that critical turning point.

Honda's Ashimo

The current algorithms we use for artificial intelligence are essentially NP-complete, which means they belong to a class of problems for which we don’t know an efficient solving algorithm just yet: however, if someone was to find the algorithm to efficiently solve even one of such problems, the resulting algorithm could be applied to the entire NP-complete class of problems which, given their numerous applications, would in turn mean a major technological breakthrough.

In simple terms, the typical solving time for a NP-complete problem increases exponentially or binomially in the number of inputs. Since most AI need to operate in real time, giving a timely response to human interaction, this puts an extremely tight limit to the number of inputs a CPU can process within an acceptable time frame.

While NASA scientists can probably do without real-time response and let their robots crunch the numbers for a relatively long amount of time, they still need to deal with the huge power consumption AI algorithms entail: and solar radiation on Mars with our current technology (up to 28% efficient) isn’t quite enough to allow for a decent AI to compute anything useful… at least, that is what I thought until I heard about this.

This is from a NASA article from some time ago:

NASA is planning to add a strong dose of artificial intelligence (AI) to planetary rovers to make them much more self-reliant, capable of making basic decisions during a mission.

Scientists at NASA Ames Research Center, in the heart of California’s Silicon Valley, are developing very complex AI software that enables a higher level of robotic intelligence.

Later on, the article makes the very valid point that keep instructing our robots from Earth is definitely not the way to go, given the long reaction times that ultimately don’t allow us to exploit these multi-million dollar machines to their full potential. And the waste is going to increase even more with time, once we reach destinations further and further away from Earth.

Developing an AI so sophisticated and efficient that would be able to make the robot completely autonomous would be a huge step forward. There was a very interesting “hypothetical documentary” from some time ago examining how something like this could greatly help us finding an extraterrestrial form of life. It’s very “science-fictiony” but realistic in many aspects.

Mars Phoenix - Artist's Concept

One of the leading companies devoting to AI is Novamente, which is pioneering a new approach to the problem of testing and developing what they define as an “artificial general intelligence” — something capable of deduction, abstraction and other kinds of human-like reasoning.

Their approach is to test their AI (which is, by the way, being developed in C++) in virtual words such as Second Life rather than in the real world, in order to cut production costs and not to waste time solving mainly sensor- and motion-related issues instead of concentrating on the actual problem.

On their site, you can watch videos that demo their work. The papers section is also pretty interesting. What will first come to mind watching videos of the AI in action, though, is that the response time is currently very far from being real-time even for relatively simple tasks, and a powerful computer cluster is needed to keep it at least acceptable.

This is therefore my question: how can NASA be so confident that they will be able to develop artificial intelligences capable of elaborating such complex algorithms, including figuring out how to let robots decide where and when to move, and power them with something as inefficient as solar panels? I hope I’m wrong, but it would seem that we will have to wait quite some time to witness something like this happening.