When I asked Google AI to put image of two black people it shows me. Then I asked , can you make them Indian? It shows. The I asked, can you make them Russians? It gave me racial discrimination policy violation. And it didn’t show. Then I asked two persons riding a bike… it showed. Then asked can you make them indian software engineer ? A surprisingly it did, and showed. Then I asked make them American software engineer.. it did. The I asked them to make Hindus - it didn’t . - understandable Then I asked them to make Protestants - it didn’t. Understandable. Then I asked can you make these people black ? - it shows me. Then I asked can you make these prior white ? - at this point first it gave me policy violation and changed the response and showed two white persons. Lastly when I asked as to why you changed response? - it accepted that it’s not reliable since I couldn’t strike a balance in last two responses. #tech
What about gypsies
Ok...
Rajesh and Priya finally have their love story ❤️
Kya chahte ho? Kyon itni chull hai?
😂 is this offend you. Check fair and lovely add once😂😂
This is a known issue with GPT too and was reported on Social Media. Chat GPT was asked to create jokes on Indians and it obliged without issues. But it refused to generate similar jokes for other ethnicities. It means it was trained on data that had a bias.
Do you think the AI engineers are using ChatGPT to write there AI policy?