And what the economy favors does not always overlap with the best interests of marginalized groups. Seems to me GPT answered the question well enough, better even, what is it supposed to do- lie?
IDK how people expect it to behave. No matter what the output is, someone somewhere is always going to be offended.
This is such a bizarre take. ChatGPT is not sentient. It's not doing some complex economic and social analysis that determines someone with a disability is less productive for a company. It's just biased by the prejudiced input it was trained on, which reflects the widespread ignorance of disabled people who exist and work in companies all the time. Do you really think someone in a wheelchair can't be a productive accountant?