📣 We need better metaphors for LLMs; they are sociotechnical systems, not human
💭 To challenge the anthropomorphic metaphor, we propose a spectrum of alternate metaphors that reveal LLMs' underlying system to encourage critical engagement: https://t.co/Ywvhwo90yJ #chi2026
there is a whole genre of content about people preparing to bully humanoid robots when they inevitably are forced into our lives… huge disconnect between what tech bros are building and what people want
every day i feel so grateful to be in grad school, surrounded by caring researchers. this fall i'm studying the history of love in southeast asia, woven sculptures, cognitive neurology, and bikolano linguistics 🍭
many user issues with LLMs are rooted in this obsession with anthropomorphism. if design is meant to make a system clearer, anthropomorphism is often doing the opposite of that: convincing people to believe in empathy and intelligence that does not exist. we need better metaphors
we need to practice restraint with hello kitty collabs. that girl is on too many objects now. and that’s coming from someone who is typing this with a golden hello kitty around her neck
Grad students are grad students because we care about the world and want to improve it. We take pay cuts to pursue our passions, to live in service to publicly funded research that is meant to benefit everyone. This is all heartbreaking to live through and watch.
I’m a 2nd year PhD student in computer science at Harvard & I just passed my qualifying exam! It should be a celebratory milestone, but I feel so much dissonance being in academia rn. My spine has been withering away at my desk while the walls of the institution crumble around us
Most recently, the attacks on international students have been shocking and shameful. My PhD cohort is 55% international, and each person makes this place better with their presence and research.
i think (platonic, romantic) love is bending time and space to let people shape you: wanting every moment to last longer, spending time apart revisiting memories together, eternalizing them somehow
What is the best way to learn behaviors from videos? By modeling point trajectories, our method, ATM, helps robots learn even from videos of humans and other robots 🤝
Grateful to have worked with amazing collaborators; check it out below!
Would you come out to ChatGPT? Or should you? Check out our paper (@nicksu28, Yiyang Mei, Yinru Long, Krzysztof Gajos) where we studied how LGBTQ+ people used LLM-based chatbots for mental health support. https://t.co/pp919OZAev #CHI2024#LGBTQ#mentalhealth#LLM#ChatGPTPlus