AI Bias Pt 4

Non-binary people and people from the LGBTQ+ communities have suffered exclusion and toxicity for many years. With AI, it goes way back to the 1950s when Alan Turing, recognized as one of the godfathers of AI, took cyanide poisoning to avoid being jailed for life due to his sexual orientation. And from the early AI chatbot days, LGBTQ+ communities have been ridden with all manners of toxicity, hate speech, and misrepresentation. For example, Microsoft's early AI chatbot called Tay, which lived for about 16 hours only, was trained on Twitter chats and generated highly offensive tweets, including explicit anti-gay slurs and homophobia. Now fast forward to the 21st century and the current times that we're in, non-binary communities are still facing various challenges with AI systems. There is a risk of erasure and algorithmic exclusion representing the real-world discrimination that affects people from these communities. Gen AI in particular has shown how image generators reinforce non-binary stereotypes. When prompted to generate images of non-binary individuals, Midjourney returned cliché results. It produced a gay man with earrings and colorful outfits, and a purple-haired woman as a bisexual woman. Now transgender individuals are also problematic and misrepresented. ChatGPT also reflected biases when asked for media recommendations for a straight person versus a gay person. Now, these misrepresentations drive further hate and hate crimes in society. It also excludes and misrepresents people, leading to structural inequalities. So what should you do? Make sure you build and work with AI, making sure that you represent the different underrepresented groups. Make sure to include them in your conversations, listen to their feedback, and simply ensure that appropriate representation is given in all your AI-generated outputs.