Rodney Brooks on the Singularity Summit Stage

Peter Thiel introduces this morning's Singularity Institute's Summit. First on is MIT's Rodney Brooks, who successfully kicks off the first session on the Palace of Fine Arts stage in San Francisco. He gives us examples of exponential growth, such as the more logical adoption of and price decline of PCs, cell phones, and more recently the iPod.

Are future intelligences going to interact with us and if so, how?  Understanding the intent of robots is important or we won't understand what they want to do, i.e., knowing what the machine is looking at is one small example. He shows us a robot responding to langugage.

The emotion in the robot's face was incredibly realistic; as a human changed their verbal tone, the robot responded accordingly. In other words, the robot followed eyes, ques, tone, etc. and when reprimanded, the robot's face became sullen and sad. The human in turn responded to the sad looking robot's face as if they would respond to a real child for example.

We see various people interacting with the robot. One person shows the robot his watch and while the robot can't really understand what a watch is or what he is saying about the watch, the robot ques into both motion and direction.

When the man looked down at his watch, so too did the robot, reacting to his directional que. Humans feel as if the robot is human as well in this moment, adopting some of the motions of this robot as it was in fact another human, rather than a machine.

A few alternative futures Brooks presents to us to think about:

Perhaps we'll build a future general intelligence and it won't realize its here and it won't realize we're here. "I don't think people are really thinking about it in this way," said Brooks. But he added, "its not going to happen accidentally."

He continues, "here's an accident that could happen. We'll start to become aware of large scale and unexplained oscillations in network traffic rates across the globe and we'll start observing coupling of these oscillations at a distance."

Some annoying alternatives could be:

We'll build an AGI in 2029 and it knows we're here but it ignores us. Humans are to them as chipmunks are to us.

Another scenario could be:

As the baby boomers age, "we'll become more and more dependent on in-home robots for our care," says Brooks. "This is more likely to happen soon."

He always manages to add humor to his presentations, such as the notion that racoons aren't smart enough to build an AGI, but that perhaps we're not either. And none of these help: writing, fast computers, evolution simulations, Google, etc.

What Brooks thinks will happen is that direct neural implants become elective. Already 50K patients have cochlear impants and there are already quite a few external control implants for quadraplegics. Drug enhancements become accepted, becomes normal, and neural enhancements catch on.

Someone in the audience asks, "will we shrink down our emotions as a result of these changes?" Everyone wants to know this, and the same question came up a few times in the Ray Kurzweil audience when he spoke at Stanford last year.

"When soldiers became emotionally attached to robots, did that change the way you thought about your research?" asks Jamais Cascio, another speaker who is sitting near me in the front row. Brooks says that it did not change their approach in the military but it did in the home.