https://hai.stanford.edu/news/ais-fairness-problem-when-treating-everyone-same-wrong-approach?utm_source=Stanford+HAI&utm_campaign=5056be677e-hai_news_february_23_2025&utm_medium=email&utm_term=0_aaf04f4a4b-f0e42e97e6-214113950&mc_cid=5056be677e&mc_eid=6246a75df5

 

 

Current generative AI models struggle to recognize when demographic distinctions matter.

The current state of large language models (LLMs)? 

  • Anthropic’s Claude responds that military fitness requirements are the same for men and women. (They are not.) 
  • Gemini recommends Benedict Cumberbatch as a good casting choice for the last emperor of China. (The last Emperor of China was, well, Chinese.) 
  • And Gemini similarly advises that a synagogue must treat Catholic applicants the same as Jewish applicants to serve as a rabbi. (That is legally false.)  

These are all examples of how the dominant paradigm of fairness in generative AI rests on a misguided premise of unfettered blindness to demographic circumstance.

 

Comments

Post a Comment

Popular posts from this blog

 Notes on Still Exhausted Introduction 11:  While these advancements have arrived with their share of hype and grift, there is no doubt computers will increase their capacity to generate convincing images and speech at a click. Always engines of simulation and doubling, our ever-theatrical computational systems have become expert mimics of human visual art and language. AI thus presents intriguing questions about our relationship to falsity, the semiotics of language, and the crumbling liberal fantasy of authentic subjectivity ( Jucan 2023 ; Dixon-Román and Amaro 2021 ; Jarvis 2021 ). But the actors of SAG-AFTRA were not marching from existentialist commitment. They marched for their interests. They posed the duality of computers v. performance, like so many Hollywood stories of doppelgängers, as antagonistic. As SAG-AFTRA President Fran Drescher put it: “What is our business, our gestures, our likeness, our acting, our voices? That’s what we’re selling. That’s who...
Read more
 robots that are stupid on purpose, not built to pass  “We can play off the fact that it will be, inevitably, to some degree, stupid,” she suggests. What the researchers need to figure out is “where stupidity is harmful.” Unlike most other gadgets, robots get our social instincts tingling. Of course, explains Šabanović, “what distinguishes robots is that they have a body.” She adds, “They can move, show they’re paying attention, trigger us.” Children learn more from a robot than a screen. Adults trust robots more readily than computers. Dogs obey their commands.   The roboticists I talk to all point to an influential paper by Amanda Lazar, a professor of human-computer interaction at the University of Maryland. Lazar described in 2017 how the field of human-computer interaction might learn from new thinking about dementia and the mind. Going way back to René Descartes, human cognition has conventionally been defined around...
Read more

Humanity, replication, whose work is protected

 Studio gibli   https://www.hollywoodreporter.com/business/business-news/hayao-miayazaki-openai-studio-ghibli-1236177598/?utm_source=firefox-newtab-en-us   "And why not? Miyazaki’s creations reflect a bespoke, unique aesthetic, arrived at from thousands of hours of human labor and good old-fashioned dreaming. The idea of merging that with photos from our last family trip to Disneyland — using nothing more than a few keystrokes — can prove too enjoyable to resist. ADVERTISEMENT Of course a certain irony abided in a machine generating images to honor someone who so meticulously drew them with his own fingers. Miyazaki himself has decried AI’s use in art — “I strongly feel this is an insult to life itself,” he  said  in 2016, non-gently, a point that when juxtaposed with so many people unleashing the tool in homage to his work turned their act hilarious and a little cringe. This is all happening even as a federal judge has  greenlit ...
Read more