From a mind-controlled bionic arm in high school to ML research at SEAS, the Kempner Institute, NASA and Johns Hopkins, Benjamin Choi shows how applied math and CS can turn noisy brain signals into real-world impact. https://t.co/86rxAGIlEM
3/ We also find (in line with @1e0sun) that steering along geometric directions causally shifts emotional tone. A parallel analysis on human brainwave data reveals similar structure — suggesting shared principles across artificial & biological systems.
https://t.co/gBOJau3WSl
1/ LLMs were never designed to "feel" — but they develop internal representations of emotion whose geometry parallels human affective processing. Excited to share our take alongside recent concurrent work in this space from @AnthropicAI and @1e0sun!
w/ @mweber_PU
2/ Using manifold learning on hidden state activations, we find that LLMs organize emotions along valence–arousal dimensions, forming a parabolic geometry aligned with human psychology. We show this geometry can be exploited for uncertainty quantification in affective processing.
Benjamin Choi's robotic arm operates using an algorithm driven by artificial intelligence that interprets a user’s brain waves. https://t.co/SD4yT7S9Zy
Excited for my first ever research presentation @HRI_2021! Many thanks to my research mentor @ChungHyukPark and the SERC program @PotomacSchool. Formal Verification for Human-Robot Interaction in Medical Environments in LBR Booth 31 at 13:30 MT #HRI2021 https://t.co/LsViMyfscN