AI Safety Innovations, Election Misinformation, and Controversial Investments

In this episode, we explore groundbreaking advancements in AI safety with Meta's new backtracking technique that reduces unsafe outputs. We also discuss the alarming rise of AI-generated election disinformation and its implications for democracy. Additionally, we cover the recent ban on AI use in child protection cases in Victoria, and the announcement of a £10 billion AI project in Northumberland backed by a pro-Trump billionaire. Join us as we navigate the complex landscape of AI technology and its impact on society.

Sources:
https://www.marktechpost.com/2024/09/26/meta-ai-researchers-propose-backtracking-an-ai-technique-that-allows-language-models-to-recover-from-unsafe-generations-by-discarding-the-unsafe-response-and-generating-anew/
https://apnews.com/article/artificial-intelligence-elections-disinformation-chatgpt-bc283e7426402f0b4baa7df280a4c3fd
https://www.theguardian.com/australia-news/2024/sep/26/victoria-child-protection-chat-gpt-ban-ovic-report-ntwnfb
https://www.theguardian.com/technology/2024/sep/25/labour-to-announce-10bn-ai-project-in-northumberland-backed-by-pro-trump-billionaire

Outline:
(00:00:00) Introduction
(00:00:44) AI-created election disinformation is deceiving the world
(00:04:30) AI ban ordered after child protection worker used ChatGPT in Victorian court case
(00:07:54) Labour to announce £10bn AI project in Northumberland backed by pro-Trump billionaire