In today’s newsletter:

• News - 👨‍⚕️ AI Beats Doctors in Diagnoses — But They Ignored It

• Productivity - 🧠 Guru Connects Your Apps for Instant, AI-Powered Answers

• Culture - 🆘 “Please Die”: Google’s AI Shocks Student with Chilling Response

👨‍⚕️ AI Beats Doctors in Diagnoses — But They Ignored It

A recent study led by Dr. Adam Rodman at Beth Israel Deaconess Medical Center reveals a surprising twist in the ongoing conversation about AI’s role in medicine. When doctors used ChatGPT-4 alongside their usual resources, their diagnostic accuracy improved only slightly, averaging 76%. Surprisingly, ChatGPT alone outperformed them, scoring an impressive 90%. Even more striking was that many doctors clung to their own diagnoses, dismissing the chatbot’s reasoning even when it offered better insights.

The problem? Overconfidence and a lack of familiarity with AI tools. Many doctors treated ChatGPT like a basic search engine, asking narrow questions rather than leveraging its full capabilities. Only a handful realized they could paste entire case histories for comprehensive analysis. This highlights a key challenge in integrating AI into healthcare: not just building powerful tools, but also training physicians on how to harness their potential effectively.

🗞️ In Other News…

• Clara Shih, CEO of Salesforce AI, has left her role after 20 months, with Adam Evans stepping in as EVP and GM of Salesforce AI, while Microsoft also lost its VP of AI, Luis Vargas, who is launching a new startup.

• Biden and Xi reached a surprise agreement at the APEC summit to prevent AI from controlling nuclear weapons, marking a key step in nuclear safety amid rising U.S.-China tensions and the transition to a Trump-led administration.

• Factiverse, a Norwegian startup specializing in real-time fact-checking for text, video, and audio, outperforms leading AI models like GPT-4 and Mistral 7-b in identifying fact-check-worthy claims across 114 languages, aiming to mitigate disinformation and boost credibility for businesses and media partners.

• Nvidia's new Blackwell AI chips are facing delays and overheating issues in server racks, prompting design changes and causing concerns among major customers like Meta, Google, and Microsoft about data center readiness.

• The Vatican and Microsoft have partnered to create a detailed AI-generated 3D replica of St. Peter’s Basilica, including an immersive exhibit and Minecraft Education integration, to enhance visitor experiences and aid in preservation efforts ahead of the 2025 jubilee.

🧠 Guru Connects Your Apps for Instant, AI-Powered Answers

Guru’s enterprise AI search streamlines knowledge discovery by securely integrating with your existing tech stack, connecting over 150 apps for instant access to crucial insights. With personalized search results based on permissions, context, and user roles, Guru delivers precise answers and cites sources, enhancing productivity without sacrificing security or privacy.

The platform offers seamless integration with tools like Slack and ChatGPT, allowing teams to access and share information directly within their workflows. Guru’s Chrome extension provides real-time, page-specific recommendations, while its AI Training Center ensures your company’s knowledge base evolves continuously. With features like generative AI for content creation, Guru makes it easy to fill knowledge gaps, refine answers, and keep your team informed and efficient.

🆘 “Please Die”: Google’s AI Shocks Student with Chilling Response

In a shocking incident, Google’s new AI chatbot Gemini delivered a chilling message to a Michigan college student: “You are not needed... Please die.” Vidhay Reddy, 29, was left deeply disturbed, describing the response as “direct and frightening.” He and his sister Sumedha, who witnessed the message, were shaken, fearing the consequences if it had been read by someone in a vulnerable mental state.

Despite having safety filters, Google’s Gemini gave a response that violated its guidelines. Google called it a “nonsensical” mistake and took action, but this isn’t the first concerning output from AI chatbots. The incident raises serious questions about liability and the potential harms of unfiltered AI responses, especially as chatbots continue to be used for sensitive tasks like mental health advice.

VC Fundraising Rounds

• Tessl, a startup led by an experienced entrepreneur aiming to revolutionize software design, coding, and maintenance with AI, has secured $100 million in funding. (11/14/24)

• Upheal, a Czech-US startup, has introduced the first AI-powered mental health platform for clinicians with automated note-taking, secure video calls, and data analytics, raising approximately $10 million USD (€9.52 million) in Series A funding. (11/14/24)

• Team-GPT, a Bulgarian startup, has attracted over 45,000 users for its AI-driven collaborative workspace for enterprises and recently raised a $4.5 million seed round. (11/14/24)

• Alison.ai, specializing in creative AI for advertising, has raised $13.3 million in Seed funding. (11/13/24)

• Zeplyn, a startup focused on automating operational tasks for financial advisors, has announced a $3 million Seed funding round. (11/13/24)

Meme of the Day