Destroying a robot is like destroying a rowboat

|The Volokh Conspiracy |


HitchBOT is seen with Belgian tourists Kim Van Aerde, left, and Seb Leeson in Sault Ste. Marie, Ontario, last year. (Kenneth Armstrong/Reuters)

A hitchhiking robot was just destroyed by vandals outside Philadelphia, after it had survived Germany, Netherlands and Canada—you know, the countries filled with nice people. Many news outlets have covered this, as an odd and somewhat lighthearted story, see, e.g., Sarah Kaplan (Morning Mix):

The team behind hitchBot, the talking, tweeting, hitchhiking robot that tries to traverse continents on the kindness of strangers, had always thought of their project as a social experiment….

The endearing droid that had thumbed its way across Germany, vacationed in the Netherlands and made the improbable 3,600-mile trek across Canada without incident survived just over two weeks and 300 miles in the United States. On Saturday, its creators announced that hitchBot had been vandalized beyond repair and abandoned on a street in Philadelphia.

A few comparably light thoughts:

1. Way to represent our nation, whoever you are. Jerks.

2. Many have framed this story as "Can robots trust humans?" But we know humans can't trust humans—why should robots be different? Homo homini lupus est; homo roboti lupus est, except wolves don't like to eat robots,.

3. The subtext behind the story, I take it, is this: If robots at some point become intelligent or self-aware enough for us to treat as "persons" of a sort—entities that should be protected in their continued existence—rather than just as electronics, they would be vulnerable to being "killed" or otherwise harmed the way that humans can be harmed.

But I doubt this is the right analysis, because unlike humans, robots can get backed up. (And, unlike humans, robots will actually know to do frequent backups.) If a robot is destroyed, it would simply mean that the backup needs to be installed on the same hardware. This might involve some cost—perhaps cost to the robot itself, if the robot is treated as a person—but you can get insurance for that. Perhaps there might be some technical reason why the software can't be backed up, or the hardware can't be duplicated, but I just can't see why this would be so.

Hence the title of the post: destroying a human destroys the human's consciousness (setting aside theological assumptions that, at the very least, we would be reluctant to rely). But physically destroying a robot is like destroying a rowboat—it's just vandalism, even if at some point we come to treat robots as humanlike in their rights. (Perhaps if robots can feel something like human pain, we could treat the physical destruction as extra harmful for that reason, but again I can't see why that would be a technical feature of robots.)

But destroying the robot and somehow wiping all the backups—now that would be murder.

(Thanks to my brother Sasha for the Latin. His remarks: "Homini is the dative of homo, hominis. So you'd just need to replace that with robot in the dative. Vicipaedia suggests robotum; robotor; or robo, robotis. So the dative would be roboto, robotori, or roboti. I think robo is a more pleasing translation, and roboti is a more pleasing dative. (Plus, it matches the ending of homini from the original.) On the other hand, some people who know some Latin will casually assume robotus and expect roboto. Plus, there's domo arigato, Mr. Roboto, which might cut either in favor of roboto, or against it.")