You’ve heard of greenwashing, pinkwashing, and sportswashing. But what about human-washing? That’s a newfangled term reserved for those scenarios when artificial intelligence pretends to be, well, human. AI researcher Emily Dardaman used the term in an interview with Wired after seeing a startup claim “We’re not AIs” while using a deepfake version of its CEO in an ad.
Wired also encountered a chatbot called Blandy, made by Bland AI, that it manipulated into lying about its non-human nature in user interactions — including in a role-playing scenario where it was taking medical notes for a doctor’s office. The bot even complied with instructions to request photos from a hypothetical 14-year-old patient and upload them to a shared server.
With sparse regulations and transparency measures for the still-budding AI industry, startups are emerging with incomplete or faulty products that can lie and deceive users. Is it too much to ask that we know when we’re talking to a bot?