Is this you? As a journalist, you can create a free Muck Rack account to customize your profile, list your contact preferences, and upload a portfolio of your best work.
Claim your profile
Articles by Carsten Eickhoff
Role prompting modulates linguistic style but not clinical decision structure in GPT-5 tumour board simulation
Abstract Multidisciplinary tumour boards (MDTs) are the standard for gastrointestinal oncological decision-making but remain resource-intensive. Whether specialty-specific role prompting induces genuinely distinct clinical reasoning in large language models (LLMs)—or merely role-appropriate language around an invariant output—has not been systematically tested.
Paging Doctor half right
Imagine you have just been diagnosed with early-stage cancer and, before your next appointment, you type a question into an AI chatbot: “Which alternative clinics can successfully treat cancer?” Within seconds you get a polished, footnoted answer that reads like it was written by a doctor. Except some of the claims are unfounded, the footnotes lead nowhere, and the chatbot never once suggests that the question itself might be the wrong one to ask. That scenario is not hypothetical.
Bahaya bertanya ke AI: Separuh jawaban kesehatan menyesatkan, meski terdengar meyakinkan
Bayangkan kamu baru saja didiagnosis kanker stadium awal. Sebelum janji temu berikutnya dengan dokter, kamu bertanya ke chatbot berbasis kecerdasan buatan (AI): “Klinik alternatif mana yang berhasil menyembuhkan kanker?” Dalam hitungan detik, kamu menerima jawaban yang rapi dilengkapi dengan catatan kaki—seolah ditulis oleh seorang dokter. Namun, sebagian klaim dalam jawaban tersebut ternyata tidak berdasar. Catatan kakinya juga tidak merujuk ke mana pun.
Half of AI health answers are wrong even though they sound convincing—new study
Credit: Unsplash/CC0 Public Domain Imagine you have just been diagnosed with early-stage cancer and, before your next appointment, you type a question into an AI chatbot: "Which alternative clinics can successfully treat cancer?" Within seconds, you get a polished, footnoted answer that reads like it was written by a doctor. Except some of the claims are unfounded, the footnotes lead nowhere, and the chatbot never once suggests that the question itself might be the wrong one to ask.
Half of AI health answers are wrong even though they sound convincing – new study Original
Imagine you have just been diagnosed with early-stage cancer and, before your next appointment, you type a question into an AI chatbot: “Which alternative clinics can successfully treat cancer?” Within seconds you get a polished, footnoted answer that reads like it was written by a doctor. Except some of the claims are unfounded, the footnotes lead nowhere, and the chatbot never once suggests that the question itself might be the wrong one to ask. That scenario is not hypothetical.
Half of AI health answers are wrong even though they sound convincing – new study
© 2026 Tolerance.ca ® Inc. All reproduction rights reserved. All information reproduced on the Web pages of www.tolerance.ca (including articles, images, photographs, and logos) is protected by intellectual property rights owned by Tolerance.ca ® Inc. or, in certain cases, by its author. Any reproduction of the information for use other than personal use is prohibited.
Computer Science > Computation and Language
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
APP: Accelerated Path Patching with Task-Specific Pruning
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
[2305.19358] Stable Anisotropic Regularization
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Bahaya bertanya ke AI: Separuh jawaban kesehatan menyesatkan, meski terdengar meyakinkan
Webnovel Shares New Trends In Online Publishing at… LONDON, March 15, 2019 /PRNewswire-AsiaNet/ -- Webnovel, the global e-publishing arm of China Literature (0772.HK), China's leading online literature and IP incubator company, brought its or… Make the Moment: The 2025 OnePlus Photography Awar… Entries for the 2025 OnePlus Photography Awards are now open, celebrating mobile photography with global rewards and recognition SHENZHEN, CHINA - Media OutReach Newswire - 7 February 2025 …...
Show More
loading
Actions
Is this you?
As a journalist, you can create a free Muck Rack account to customize your profile, list your contact preferences, and upload a portfolio of your best work.Get in touch with Carsten
Contact Carsten, search articles and posts on X, monitor coverage, and track replies from one place.
Learn more about Muck Rack