The Artificial Intelligence Podcast

The Pentagon has ramped up deployment of its artificial intelligence (AI) technology in a bid to match with China's pace. Its ambitious AI project, known as Replicator, aims to deploy thousands of inexpensive AI-enabled autonomous vehicles by 2026, which would force hard decisions about the AI tech's maturity and trustworthiness. While many countries are developing AI-powered weapons, concerns about ethical use and lack of regulatory oversight persist. The Pentagon's commitment to maintain human control over AI systems is being challenged by advances in AI technology that could reduce humans to merely supervisory roles.

Social media behemoth Meta is developing two AI-powered generative models, Emu Video and Emu Edit, for enhanced content editing and generation on Facebook and Instagram. Emu Video uses text and image inputs to generate video clips, while Emu Edit focuses on precise image manipulation. Despite being in the research phase, the models show potential for creators, artists, and animators. However, regulators are keeping a close watch on Meta's deployment of these AI-based tools.

In a recent announcement, YouTube is introducing an experimental AI feature, Dream Track, which enables select users to create music using voice clones of well-known artists including Demi Lovato, John Legend, and Charl XCX, among others. Using AI technology, the feature will generate music based on the user’s description of desired song qualities. Moreover, the AI-created content will carry a watermark for transparency. The functionality is similar to Google’s MusicLM, but includes pop-star vocals. The pilot feature highlights the potential of AI to revolutionize music creation and streamline the song-creation process.

In this special episode of the Artificial Intelligence Podcast, we sit down with Neil Mandt, a 5-time Emmy Award-winning producer and a visionary in the entertainment industry. Neil shares insights from his 30+ year career, discussing the highs and lows of Hollywood, his pioneering work in augmented reality, and his unique experiences in sports media production. We dive into the changing landscape of Hollywood with the advent of AR/VR and AI, explore Neil's role in innovating true crime storytelling through VR/AR, and talk about Neil's early start to his career. This episode is a treasure trove of wisdom and inspiration for anyone passionate about media, technology, and the art of storytelling.

Microsoft has announced its first ever artificial intelligence (AI) chip, Maia 100, which could compete with Nvidia's AI graphics processing units. The tech giant also introduced an Arm-based chip, Cobalt 100, for general computing tasks, targeting competition with Intel. Both chips will be available via Microsoft's Azure cloud. However, timelines for releasing the Maia 100 weren't disclosed. The move is part of larger industry trends, with tech companies like Alibaba, Amazon and Google diversifying cloud infrastructure options for their clients.

Artificial intelligence may soon facilitate human exploration and colonization of Mars. Scientists have developed an AI chemist capable of producing oxygen from Martian meteorites, a critical resource currently lacking in Mars' atmosphere and crucial for both breathing and propulsion fuels. This system functions within the Martian environmental conditions unaided and without human intervention, making it a significant stride towards cost-effective and simplified manned missions to the planet.

YouTube is set to introduce labels for videos containing realistic, AI-generated content to combat misinformation. The move aims to help users discern between synthetic and real content, especially given the rise in generative AI tools that can easily create persuasive text, images, video, and audio. This follows similar steps taken by TikTok and Meta, with the labels expected to start appearing by early next year. Creators who fail to comply might face penalties including content removal or suspension from YouTube's Partner Program.

In recently conducted MLPerf training tests, NVIDIA emerged as the frontrunner, with Intel closely following and Google trailing significantly. The tests measured the ability of computer systems to train machine learning neural networks, with a particular focus on generative AI. Notably, NVIDIA’s 10,000 GPU supercomputer posted impressive results, while Intel managed to make substantial progress using its Gaudi 2 accelerator chip. These tests saw notable participation from tech giants, underscoring their growing interest and investments in AI and machine learning technologies.

FBI warns that criminals are exploiting AI technology to transform innocent images into explicit content, leading to victim harassment and extortion. Responses to this growing threat include engaging in open conversation about online safety with children and practicing due discretion while posting online. The federal agency is collaborating with other law enforcement entities to better identify AI-created images.

Nvidia has unveiled the H200, a state-of-the-art GPU designed to train and deploy elaborate AI models. The chip boasts next-gen HBM3 memory and will deliver output at nearly double the speed of its predecessor, H100. Nvidia's stocks have soared thanks to the popular demand for AI GPUs, with the firm expecting $16 billion in revenue this quarter. The high-end H200 is expected to hit the market in Q2 of 2024.