The Artificial Intelligence Podcast

The CIA is leveraging generative artificial intelligence to enhance operations and stay ahead of adversaries. The agency's first chief technology officer, Nand Mulchandani, is leading the effort to develop AI applications that augment human capabilities, rather than replace them. The CIA is creating its own generative AI platform, similar to Chat GPT, to help analysts process vast amounts of data and spark creative thinking. However, the agency is cautious about the technology's limitations and potential biases, recognizing the need for human judgment to weigh the importance of incoming information.

Google has unveiled its sixth-generation Tensor Processing Unit, called Trillium, which boasts significant improvements in performance and energy efficiency compared to its predecessors. The new chip offers 4.7 times more peak compute performance, doubled high-bandwidth memory, and enhanced inter-chip communication, making it capable of handling AI tasks faster and more efficiently. It's also 67% more energy-efficient than the previous TPU v5e, allowing it to run for longer periods without excessive power consumption. The Trillium chip will be available in Google Cloud, although pricing and availability details have not been disclosed.

This week, South Korea is hosting a significant gathering focused on the risks and regulation of artificial intelligence, building on the work started at last year's inaugural AI safety conference in Britain. The Seoul summit aims to mitigate the threats posed by cutting-edge artificial intelligence systems, including algorithmic bias and existential risks to humanity. Delegates from over two dozen countries will attend, including the US and China, to discuss how to address the risks of AI and set shared standards to manage it. The meeting will feature updates from leading AI companies and digital ministers sharing best practices to protect society from AI's negative impacts.

A new federal assessment warns that artificial intelligence programs pose a significant threat to the democratic system of elections, particularly with the 2024 election cycle approaching. Next-generation technologies can be exploited to influence and sow discord in elections by creating or sharing altered or deepfaked pictures, videos, or audio clips. The Department of Homeland Security analysis outlines how generative AI tools can be used to confuse or overwhelm voters and election staff. Experts caution that authorities must be prepared to defend against AI-disseminated fake news and educate the public to counteract misinformation.

JPMorgan Chase is taking a leading role in AI adoption in the banking sector with a new initiative to provide prompt engineering training to all new employees, preparing them for the future of artificial intelligence. This move is expected to have a significant impact on the bank's operations, with AI set to greatly affect its 60,000 developers and 80,000 operations and call-center employees. The bank has been at the forefront of AI adoption, ranking number one in Evident's AI Index reports and employing nearly six times more AI staff than the average bank.

Open AI has partnered with Reddit, gaining real-time access to the social media platform's vast user-generated content to improve its Chat GPT technology. In return, Reddit will use Open AI's advanced language models to develop innovative features. The deal also makes Open AI an advertising partner on Reddit, expanding its reach. While some users have raised concerns, the partnership is seen as a significant step forward for AI technology, enabling Open AI to develop more sophisticated models and enhancing the online experience for users.

AI-generated news anchors are being used to spread disinformation and propaganda, with China at the forefront of this trend. A recent example in Taiwan featured a fake news anchor attacking the outgoing president, Tsai Ing-wen. Experts warn that as the technology becomes more accessible, AI-generated deepfake news anchors will continue to spread, potentially influencing elections and sowing confusion among voters.

Scientists have developed a novel machine-learning framework that can automatically map out phase diagrams for novel physical systems without requiring large, labeled training datasets. This physics-informed approach leverages generative artificial intelligence models to estimate the probability distribution of measurement statistics in a physical system, creating a classifier that can determine the phase of a system given certain parameters. This method is computationally efficient, works automatically without extensive training, and can help scientists discover unknown phases of matter autonomously, with significant implications for materials science, quantum computing, and thermodynamics.

Open AI, a leading artificial intelligence research organization, has dissolved its Superalignment team, which aimed to mitigate long-term risks associated with advanced AI systems. The team was established just a year ago and was supposed to dedicate 20% of the company's computing power over four years. The departure of two high-profile leaders, Ilya Sutskever and Jan Leike, who were instrumental in shaping the team's mission, has raised concerns about the company's priorities on safety and societal impact. Leike believes more resources should be allocated towards security and safety.