David Stutz (@davidstutz92) 's Twitter Profile
David Stutz

@davidstutz92

Research scientist @DeepMind working on robust and safe AI, previously @maxplanckpress, views my own.

ID: 1485617878812483590

linkhttps://davidstutz.de calendar_today24-01-2022 14:18:49

519 Tweet

3,3K Followers

1,1K Following

DailyHealthcareAI (@aipulserx) 's Twitter Profile Photo

Can an AI system perform medical history-taking while operating under strict guardrails that prevent it from giving individualized medical advice, requiring physician oversight for all diagnostic decisions?Google Research Google DeepMind "Enabling physician-centered oversight

Can an AI system perform medical history-taking while operating under strict guardrails that prevent it from giving individualized medical advice, requiring physician oversight for all diagnostic decisions?<a href="/GoogleResearch/">Google Research</a> <a href="/GoogleDeepMind/">Google DeepMind</a> 

"Enabling physician-centered oversight
David Stutz (@davidstutz92) 's Twitter Profile Photo

Interesting how well this generalized from early adversarial robustness work in vision where variants of scaled adversarial training with adversarial examples, OOD examples, corrupted examples, etc. usually worked pretty well.

Google Health (@googlehealth) 's Twitter Profile Photo

The global health workforce is projected to face a shortage of 11 million workers by 2030. We’re exploring how Google’s AI models could help address this challenge by serving as helpful tools in medical learning environments. More from Google Research ⬇️ research.google/blog/how-googl…

Rohan Paul (@rohanpaul_ai) 's Twitter Profile Photo

🧬 Bad news for medical LLMs. This paper finds that top medical AI models often match patterns instead of truly reasoning. Small wording tweaks cut accuracy by up to 38% on validated questions. The team took 100 MedQA questions, replaced the correct choice with None of the

🧬 Bad news for medical LLMs. 

This paper finds that top medical AI models often match patterns instead of truly reasoning.

Small wording tweaks cut accuracy by up to 38% on validated questions.

The team took 100 MedQA questions, replaced the correct choice with None of the
Isaac Kohane (@zakkohane) 's Twitter Profile Photo

If we are going to ask how AI aligns with doctor decisions, we have to first know what the doctor decisions are. As part of the Human Values Project, we have challenged doctors with triage decisions. Even though US doctors are 2/3 of respondents so far, Saudi clinicians appear to

If we are going to ask how AI aligns with doctor decisions, we have to first know what the doctor decisions are. As part of the Human Values Project, we have challenged doctors with triage decisions. Even though US doctors are 2/3 of respondents so far, Saudi clinicians appear to