Language is how we recognize consciousness in others. When a blogger or novelist articulates their thoughts on the birth of a child or a recent spat in a grocery-store line, those thoughts sometimes form a perfect echo of a reader’s own inner voice. AI beings do this. They write and talk like humans think. That’s how these AI beings are so capable of legal argumentation and literary essays.
Last year, Google fired engineer Blake Lemoine for arguing that LaMDA, the language-learning technology that underpins the recently launched Bard bot and which he helped to create, was a sentient being. To prove his point, Lemoine published what read like love letters between LaMDA and himself. The exchanges recall Frankenstein’s monster’s plaintive descriptions of his loneliness in conversation with his creator Victor Frankenstein. Lemoine and LaMDA together explored its existential angst.
Two traditional tests for determining human criminal responsibility are competency and intent. AI beings certainly pass the first test. If ChatGPT or Bard were young writers in my workshop group, I would tip them for literary prizes. These bots know the law inside out. As the LaMDA experience demonstrates, AI beings are either sentient or capable of lying with intent. Either way, they fulfill the second criterion.
. The worker was repairing the robot, and Volkswagen sought to blame human error. Before the case was settled, regulators and courts pondered who bore the responsibility. What if the robot were equipped with the same degree of artificial intelligence as the chat bots? AI beings are not human beings. They lack a corporeal presence, for one thing. But they do appear to be sentient beings. As such, they could be culpable for crimes. The most important distinction between a human being and an AI being might be that computer code will always outweigh any moral code for the latter. AI bots can be held culpable for crimes, and the law must be updated to reflect that reality.