Launching the right stuff: who will make the better space explorer: robot or human being?

More than a year has passed since the space shuttle Columbia broke into pieces over central Texas. This past January President Bush announced a long-term program of space exploration that would return human beings to the Moon, and thereafter send them to Mars and beyond. As this magazine goes to press, the twin Mars Exploration Rovers, Spirit and Opportunity, are wowing the scientists and engineers at the rovers' birthplace--NASA's Jet Propulsion Laboratory (JPL)--with their skills as robotic field geologists. JPL's official rover Web site (marsrovers.jpl.nasa.gov/) is being stampeded by visitors.

The confluence of these and other events resurrects a perennial debate: with two shuttle failures out of 112 missions, and the astronomical expense of the manned space program, can sending people into space be justified, or should robots do the job alone? Or, given society's sociopolitical ailments, is space exploration something we simply cannot afford to pursue? As an astrophysicist, as an educator, and as a citizen, I must speak my mind on these issues.

Modern societies have been sending robots into space since 1957, and people since 1961. Fact is, it's vastly cheaper to send robots: in most cases, a fiftieth the cost of sending people. Robots don't much care how hot or cold space gets; give them the right lubricants, and they'll operate in a vast range of temperatures. They don't need elaborate life-support systems, either. Robots can spend long periods of time moving around and among the planets, more or less unfazed by ionizing radiation. They do not lose bone mass from prolonged exposure to weightlessness, because, of course, they are boneless. Nor do they have hygiene needs. You don't even have to feed them. Best of all, once they've finished their jobs, they won't complain if you don't bring them home.

So if my only goal in space is to do science, and I'm thinking strictly in terms of the scientific return on my dollar, I can think of no justification for sending people into space. I'd rather send the fifty robots.

But there's a flip side to this argument. Unlike even the most talented modern robots, a person is endowed with the ability to make serendipitous discoveries that arise from a lifetime of experience. Until the day arrives when bioneurophysiological computer engineers can do a human-brain download on a robot, the most we can expect of the robot is to look for what it has already been programmed to find. A robot--which is, after all, a machine for embedding human expectations in hardware and software--cannot fully embrace revolutionary scientific discoveries. And those are the ones you don't want to miss.

In the old days, people generally pictured robots as a hunk of hardware with a head, neck, torso, arms, and legs--or maybe some wheels to roll around on. They could be talked to, and would talk back (sounding, of course, robotic). The standard robot looked more or less like a person. The fussbudget character C3PO, from the Star Wars movies, is a perfect example.

Even when a robot doesn't look humanoid, its handlers might present it to the public as a quasi-living thing. Each of NASA's Mars rovers, for instance, is described in JPL press packets as having "a body, brains, a 'neck and head,' eyes and other 'senses,' an arm, 'legs,' and antennas for 'speaking' and 'listening.'" On February 5, 2004, according to the status reports, "Spirit woke up earlier than normal today ... in order to prepare for its memory 'surgery.'" On the 19th the rover remotely examined the rim and surrounding soil of a crater dubbed Bonneville, and "after all this work, Spirit took a break with a nap lasting slightly more than an hour."

In spite of all this anthropomorphism, it's pretty clear that a robot can have any shape: it's simply an automated piece of machinery that accomplishes a task--either by repeating an action faster or more reliably than the average person can, or by performing an action that a person, relying solely on the five senses, would be unable to accomplish. Robots that paint cars on assembly lines don't look much like people. The Mars rovers look a bit like toy flatbed trucks, but they can grind a pit in the surface of a rock, mobilize a combination microscope-camera to examine the freshly exposed surface, and determine the rock's chemical composition--just as a geologist might do in a laboratory on Earth.

It's worth noting, by the way, that even a human geologist doesn't go it alone. Unaided by some kind of equipment, a person cannot grind down the surface of a rock; that's why a field geologist carries a hammer. To analyze a rock further, the geologist deploys another kind of apparatus, one that can determine its chemical composition. Therein lies a conundrum. Almost all the science likely to be done in an alien environment would be done by some piece of equipment. Field geologists on Mars would schlep it on their daily strolls across a Martian crater or outcrop, where they might take measurements of the soil, the rocks, the terrain, and the atmosphere. But if you can get a robot to do the schlepping and deploy all the same instruments, why send a field geologist to Mars at all?