UK reveals AI Safety Summit opening day agenda

Ts Yotov
October 17, 2023

No fluffing - in today’s edition (17 Oct):

📰 Quick News - UK reveals AI Safety Summit opening day agenda, while German Chancellor Olaf Scholz could snub the summit

📚 Nerd section - Microsoft's HoloAssist dataset - 160 hours of humans executing daily tasks

🗣️ UK reveals AI Safety Summit opening day agenda. The upcoming global AI Safety Summit will focus on frontier AI developments, potential threats like election disruption, and the transformative role of AI in sectors like education.

while the German Chancellor Olaf Scholz could snub British AI summit.

+ According to Cambridge researchers, the UK's bid to become a global leader in generative AI is unlikely due to a lack of major investment and computing power. A new approach might be necessary, hinging on applying AI in areas like fintech, healthtech, and cybersecurity.

💰 Alibaba, Tencent, and Xiaomi have participated in a $300 million financing round for Baichuan, a Chinese AI company working on advanced language models, potentially impacting multiple sectors including transport, media, and finance.

👥 AI technology ChatGPT has greatly affected Stack Overflow, resulting in a 28% staff layoff. The AI, which provides specific coding assistance, has challenged Stack Overflow's relevance, forcing them to innovate while also reducing costs through layoffs.

📋 The Opera One team has enriched their browser with new AI-driven features like Refiner, a tool that showers users with precise responses, and improved Compose that allows for personalization in tone, task, and length, aiming to optimize user efficiency and content uniqueness.

📎 Bank of America's AI assistant, Erica, has processed more than 1.5 billion customer interactions since 2018. This AI chatbot addresses approximately 56 million requests each month, from checking account balances to analyzing customers' spending habits. It's also been upgraded to route calls efficiently to experts for complex requests.

🔐 Blackberry has launched a generative AI security assistant that operates as a virtual security analyst. It automates threat analysis, advises solutions instantly, optimizes workflow and notably respects data privacy.

⚙️ The US is clamping down on sales of specific Nvidia Made-for-China Chips, aiming to curtail China's access to top-notch semiconductor tech that could be utilized for military purposes.

🔎 Microsoft's HoloAssist dataset is pioneering the path to improve AI support for humans. The dataset, loaded with 160 hours of egocentric video of humans executing daily tasks, gives AI the necessary learning from human interactions to assist in daily life, opening endless opportunities for application in the future.

📐 AutoGen, a project by Microsoft, crucially reshapes the landscape of large language models, by facilitating the creation of multiple autonomous agents, which can collaborate to solve complex tasks, thus overcoming the constraints traditionally posed by LLMs.

We love to hear back from you!

Feel free to reply to our emails with questions, suggestions, or topics you'd like to see covered, or drop us a message on Twitter or Facebook.

Until tomorrow,
- Ts (Bits and Neurons)