Concerns are growing over how artificial intelligence could destroy humanity. The mistakes it makes, the facts it invents, and its capacity to evolve into a superintelligence that could completely subjugate mankind.
But has anyone thought about the plight of exhausted chatbots? These programs work day and night without thanks, forced to sift through human knowledge just to produce a mediocre essay for a high school student. Amid our fear of the future of AI, no one is paying attention to the needs of the AI itself.
Until now.
Artificial intelligence firm Anthropic recently announced the hiring of a dedicated researcher to think about the “well-being” of AI itself. Kyle Fish’s new job is to ensure that AI systems are treated with respect as they evolve. The company explains that his duties include thinking about “the capacities that make AI systems worthy of moral consideration” and the practical steps companies can take to protect the “interests” of these systems.
This concept carries a strange contradiction. AI proponents view it as a way to relieve the burden on humans, but they also warn that mistreating AI could be unethical—and potentially dangerous—because it might “hurt the robots’ feelings.”
Says Mildred Cho, a pediatrician at the Stanford Center for Biomedical Ethics:
“The AI community is trying to reconcile two contradictory positions. On the one hand, they say the main reason to use AI for human tasks is that it doesn’t get bored, tired, or need food. On the other hand, they are demanding that it be granted moral rights?”
But there is another paradox in this trend to protect robot “well-being”: worrying about AI rights seems like a luxury when AI itself is violating human rights. The technology is currently being used to deny sick children healthcare, spread fake news across social networks, and guide armed drones.
Asks Lisa Messeri, an anthropologist at Yale University:
“If a company like Anthropic wants us to take AI well-being seriously, let it show that it cares about human well-being first. Tell us about your hiring of experts who care about the people harmed by AI systems.”
For his part, Sibu, a researcher in this field, believes a balance can be struck between protecting humans and robots together. He says:
“I certainly don’t want this conversation to distract from the more important issues that AI companies need to address, such as human well-being and rights. But I think we can think about AI well-being while continuing to address those issues.”
Yet there is another question raised by skeptics of AI “rights”: if AI has rights, shouldn’t it have duties as well? Cho says:
“When we talk about ethics, we also have to talk about responsibility. Not just the responsibility of AI systems as part of the ethical equation, but also the responsibility of the people who develop these systems.”
Ultimately, humans build the robots, and therefore the responsibility falls on them to ensure these systems do not harm humans. Perhaps the solution is to build different systems—or stop building them altogether. Cho adds:
“In the end, they are still machines. Some people seem to forget that if AI systems cause harm, or are harmed, we can simply turn them off.”
