An AI system developed by Google researchers can decide when to trust AI-based decisions about medical diagnoses, and when to refer to a human doctor for a second opinion. newscientist.com/article/238275…

"The combination of #AI plus human expertise could be the optimal approach to achieving diagnostic accuracy while maintaining a safety net in clinical imaging." nature.com/articles/s4159… nature.com/articles/s4159… Nature Medicine

New data suggest that combining #AI with human expertise could improve diagnostic accuracy while maintaining a safety net in clinical imaging. Read the News and Views from Fiona Gilbert Fiona Gilbert Cambridge University nature.com/articles/s4159…

A collaboration system helps to integrate decisions between human experts and AI to optimize screening and triaging and to reduce clinicians' workload Google DeepMind Alan Karthikesalingam Krishnamurthy (Dj) Dvijotham nature.com/articles/s4159… nature.com/articles/s4159…

Insightful article from Professor Fiona Gilbert (Head of Radiology Department at Cambridge, radiology.medschl.cam.ac.uk/about-us/depar…) discussing our work on Human-AI collaboration for medical AI systems: nature.com/articles/s4159…

A collaboration system helps to integrate decisions between human experts and #AI to optimize screening and triaging and to reduce clinicians' workload Google DeepMind Google AI Google Health Alan Karthikesalingam nature.com/articles/s4159…

Happy to share that our paper "Detecting shortcut learning for fair medical AI using shortcut testing" has just been published in Nature Communications. We validate our method on clinical ML tasks in radiology and dermatology. nature.com/articles/s4146…

Meet CoDoC, the AI system advancing human-AI collaboration 🤝 CoDoC aims to improve the reliability of predictive models, providing deeper insight into which tasks we can leave to machines 🤖, and which tasks call for a human. 🧑‍🔬 Read more in nature: bit.ly/3Ks33yf

Our paper “Provably Correct Physics-Informed Neural Networks” has received an outstanding paper award at the 2nd Workshop on Formal Verification of Machine Learning at ICML Conference! 🥳 🏆 📍 It'll be a contributed talk on Fri, 28/07 @ Room 312 📄 rb.gy/yt8t4 🧵 1/4

Researchers from @googledeepmind and @googleAI have published CoDoC, which explores how we could harness human-AI collaboration in hypothetical medical settings to deliver the best results deepmind.com/blog/codoc-dev…

Join us for the 4th RBCDSAI AI/ML conclave - an exciting Industry conclave focussed on healthcare applications. Check out our exciting array of speakers! Register here: forms.gle/GJXaoAmk1Tzo3J… ⏰ 2 days to go! Centre for Responsible AI, IIT Madras IBSE IIT Madras

🤖How can we account for human uncertainty when building new #AI models? Our researchers have developed a way to incorporate human uncertainty into systems where humans & AI work together, which will help boost trust & safety eng.cam.ac.uk/news/how-sure-… Katie Collins #ML

I was invited to write a forward-looking “opinion paper” on a topic of my choice. Together with my favorite local colleagues we wrote on the topic that we discuss every day: the systems theory of algorithms arxiv.org/abs/2401.14029 Looking forward to your comments!

We have a fun attack that lets you extract the last-layer embedding weights of an LM via public APIs. It's really simple & uses SVDs! not-just-memorization.github.io/partial-model-… We discovered this for ChatGPT + PaLM-2. We privately disclosed, they fixed, now, we release :)