 |
Thiago Dias Oliva studies anti-LGBTQ+ hate speech online. He has conducted studies into the artificial intelligence tool Perspective which assigns text a 'toxicity' rating. His results found major flaws in the software which identified a Black, gay drag queen's tweets as more 'toxic' than a white nationalist, or Donald Trump.
Queer activists and creators have long maintained that their content is 'over-policed' while a whole host of considerably more harmful content is under-policed.
"Hate speech and harassment are all about context", says Jillian York, citing how the oppressed often turn insults into empowerment speech. AI possesses the nuance to account for this - but with the tech industry still struggling with diversity, it so often fails to accommodate these contexts. |
|