Saltar al contenido
Radio Logo

AI Daily

Amy Iverson
Último episodio

737 episodios

  • AI Daily

    AI Moves Into Chips, Finance, and Global Power Politics

    19/11/2026 | 29 min
    AI Daily Podcast explores the latest breakthroughs and tensions shaping the future of artificial intelligence — from AI-designed chips to the geopolitics of cloud-based model access.

    In this episode, we examine reports that OpenAI and AMD have achieved major chip-design gains using AI. According to OpenAI president Greg Brockman, AI models helped AMD compress nearly a year of design progress into less than five months, producing reported improvements of up to 14% in power, performance, and area across already highly optimized GPU designs.

    The discussion highlights how AI is entering real semiconductor production workflows. OpenAI’s models generated circuit-level changes, while Synopsys electronic design automation tools verified that the designs functioned correctly. Some of these AI-assisted improvements were reportedly included in a recently taped-out AMD chip, signaling a move from experimentation into manufacturing.

    We also explore the emerging feedback loop between AI and hardware: better AI helps design better chips, better chips power more advanced AI systems, and those systems can further improve infrastructure, deployment, and data center efficiency. OpenAI’s broader partnership with AMD — including plans involving six gigawatts of AMD processors — could represent a deeper co-evolution of AI models and AI hardware.

    The episode also covers OpenAI’s custom inference chip project, Jalapeño, developed with Broadcom, where AI reportedly helped reduce circuit area by 10% to 13% in certain components. As inference becomes one of the largest cost drivers in AI, these kinds of efficiency gains could have major industry-wide implications.

    Next, we turn to AI governance in finance. At the Singapore FinTech Festival, Ant Digital Technologies introduced its Agent Factory platform and emphasized the concept of “Know Your Agent”, or KYA. As AI agents begin operating inside financial systems, institutions will need ways to identify, authorize, monitor, and audit them — much like they already do with customers and businesses.

    Finally, we analyze the controversy around Microsoft’s reported $500 million arrangement to provide ByteDance with access to OpenAI models through Azure. While OpenAI restricts direct access for many Chinese companies, Microsoft’s role as a cloud and commercialization partner raises difficult questions about AI access, export controls, and national security.

    Lawmakers have warned that powerful AI tools could strengthen ByteDance’s recommendation systems, advertising, moderation, translation, creator tools, and internal engineering capabilities — while also raising concerns about dual-use applications. The story reflects a broader shift in AI policy: frontier models are no longer distributed only as software products, but through cloud platforms, APIs, and managed enterprise services.

    Together, these stories show that AI innovation is moving far beyond chatbots and benchmarks. It is becoming embedded in semiconductor design, financial operations, cloud infrastructure, and global technology competition. The central question is no longer just how powerful AI can become — but where it is deployed, who controls access, and how much trust society is willing to place in intelligent systems operating inside critical workflows.
    Links:
    ‘Unprecedented Results’: AMD Packs a Year of Chip Progress Into 5 Months Using OpenAI, Greg Brockman Says
    Ant Digital Technologies Brings Agent Factory to Singapore FinTech Festival, Accelerating Agentar's Global Rollout
    John Moolenaar Calls Out Microsoft for ‘Playing Both Sides’ of US-China AI Race
  • AI Daily

    AI Moves From Demos to Real-World Infrastructure

    18/11/2026 | 31 min
    AI Daily Podcast explores the latest innovations in artificial intelligence technology, focusing on how AI is moving beyond demos, chatbots, and pilots into the operational core of banks, telecom networks, airlines, insurers, manufacturers, governments, and critical infrastructure.

    In this episode, we examine a major shift in enterprise AI: the future is no longer just about model capability, but about secure deployment, governance, data readiness, auditability, and real-world integration. As AI systems gain access to code, customer data, APIs, automation tools, and internal workflows, organizations must rethink cybersecurity from the ground up.

    The discussion highlights Commonwealth Bank’s new cybersecurity tool, CodeCheck, and its broader AI playbook for Australian businesses. While CodeCheck appears to be a free software security tool, it reflects a much larger trend: major institutions are treating secure AI adoption as critical infrastructure. With AI lowering the cost and speed of cyberattacks, defenders increasingly need AI-powered tools to counter AI-driven threats.

    We also look at examples from Commonwealth Bank, Telstra, Qantas, and CrowdStrike, showing how cyberattacks are scaling rapidly and how some intrusions can unfold in seconds. The episode explains why AI agents can become insider risks if they are over-permissioned, poorly monitored, or manipulated, making access controls, observability, audit trails, and independent trust layers essential.

    The episode then turns to sovereign and industry-specific AI, including HCLTech and Cohere’s partnership to develop AI systems for Canada and Germany. For regulated sectors such as insurance, financial services, manufacturing, and government, data residency, compliance, and governance are not optional—they are often the deciding factors in whether AI can be adopted at all.

    We also explore a key reality check in enterprise AI adoption: companies are moving fast, but many are not yet ready. New Dun & Bradstreet figures suggest that while surveyed Indian businesses have AI projects underway, only a small percentage say their enterprise data is fully ready for AI at scale. This highlights a growing gap between AI ambition and production readiness.

    In Indian manufacturing, AI adoption is being studied across hundreds of MSME factories in sectors such as textiles, pharmaceuticals, and electronics. The conversation focuses on practical deployment challenges, including legacy machinery, inconsistent data, energy costs, limited capital, and workforce training.

    Siemens’ Eigen Engineering Agent is featured as an example of AI evolving from general-purpose copilots into specialized industrial agents. Connected to Siemens’ TIA Portal automation software, the agent can support PLC code generation, test logic, and automation engineering—showing how AI is becoming more deeply connected to industrial workflows.

    The episode also covers AI-powered Earth observation, where machine learning is helping analyze massive streams of satellite imagery. From crop monitoring and flood mapping to deforestation tracking, wildfire risk assessment, and climate research, AI is turning raw satellite data into actionable intelligence for governments, emergency responders, farmers, and planners.

    Across cybersecurity, manufacturing, sovereign AI, and Earth observation, one theme stands out: AI innovation is shifting from flashy demonstrations to operational control. The winners in the next phase of AI may not be the companies with the most impressive demos, but those that can build trusted systems that work safely inside real organizations, real infrastructure, and real-world decision-making environments.
    Links:
    'Experimental science': firms warned on AI safety risk
    HCLTech and Cohere Plan Sovereign AI Projects in Canada and Germany
    India Survey Finds Only 4% Have Data Fully Ready for AI at Scale
    5 Ways Artificial Intelligence is Making Satellite Data More Valuable
  • AI Daily

    AI Daily Podcast: AI Moves From Hype to Real-World Accountability

    17/11/2026 | 27 min
    AI Daily Podcast explores the latest news and innovations shaping the future of artificial intelligence technology, from safety systems and platform governance to AI pricing, enterprise deployment, and persuasive voice automation.

    In this episode, we examine how AI is moving beyond benchmark races and into the real world, where questions of trust, accountability, access, and oversight are becoming just as important as model performance. The discussion begins with OpenAI’s new image classifier designed to help detect potential child sexual abuse material, including AI-generated content. Developed with the National Center for Missing and Exploited Children, the tool represents a shift from traditional hash-matching toward AI-based risk assessment for unknown and synthetic images.

    We also look at OpenAI’s broader safety strategy for ChatGPT, including stronger interventions when users seek content sexualizing children. Rather than simply refusing harmful prompts, the system may now issue warnings, remove conversations, suspend access, and report detected cases to NCMEC—highlighting a wider industry shift from passive refusal to active platform enforcement.

    The episode then turns to Apple’s reported plan to introduce daily usage limits for advanced Apple Intelligence features. While basic Siri functions may remain unlimited, more compute-intensive AI tools could become part of a metered or subscription-based model. This raises a major question for consumers and companies alike: as AI becomes more powerful, who pays for the infrastructure behind it?

    Google’s Gemini Notebook also takes the spotlight with its new ability to transform uploaded notes, documents, and research into interactive websites. This signals a move beyond chatbots toward AI-generated interfaces such as study tools, dashboards, quizzes, timelines, and productivity apps—bringing new possibilities for education and work, while also raising concerns about accuracy, data access, and responsibility.

    We also explore Tone, a voice AI company associated with Jordan Belfort, and its focus on optimizing not just what AI says, but how it says it. By adjusting pace, pitch, emphasis, and volume, voice AI can become more persuasive in sales, customer service, narration, and other commercial settings. But this innovation also raises ethical concerns around disclosure, consent, manipulation, and use in sensitive industries such as finance, healthcare, insurance, and politics.

    The episode then shifts to institutional AI adoption, focusing on Hong Kong’s financial sector. A new report from the Hong Kong Institute for Monetary and Financial Research says generative AI adoption among financial institutions has reached 90 percent, showing that banks, insurers, and asset managers increasingly view AI as core infrastructure rather than an experimental tool.

    A major theme is the rise of agentic AI—systems that can plan, take action, use tools, monitor results, and operate within defined workflows. In finance, these systems could help with compliance by identifying regulatory changes, alerting affected teams, opening workflow tickets, and tracking remediation. But they also introduce new risks involving hallucinations, auditability, data protection, and accountability.

    Finally, we connect enterprise AI adoption to the broader infrastructure market, including uncertainty surrounding Nvidia after reports about Jensen Huang’s resignation as CEO. While Nvidia has been central to the AI boom, demand for compute is now expanding across finance, healthcare, government, retail, and other sectors as organizations deploy AI for inference, monitoring, retrieval, fine-tuning, and automated workflows.

    This episode argues that the AI industry is entering a new phase: from spectacle to implementation, from rapid expansion to disciplined execution, and from AI that simply generates content to AI that can act safely, transparently, and accountably inside critical systems. The central question is no longer just what AI can do—it is how AI should be governed, priced, limited, trusted, and deployed.
    Links:
    OpenAI Rolls Out Classifier, Aims to Detect AI-Generated CSAM
    Apple has introduced daily usage limits for Siri AI in the latest iOS, macOS, and iPadOS Betas
    Gemini Notebook can now turn your notes into interactive websites
    Jordan Belfort Launches Tone, an AI Voice Startup to Revolutionise Sales
    Hong Kong Financial Sector Sees 90% A.I. Adoption, Report Finds
    Cramer Says Possible Nvidia Commitment Cuts Could Pressure Debt-Heavy AI Companies
  • AI Daily

    AI Moves Into Critical Infrastructure

    16/11/2026 | 20 min
    AI Daily Podcast explores the latest news in artificial intelligence innovation, focusing on the infrastructure, industries, and public systems that are shaping the next phase of AI.

    In this episode, we look beyond consumer chatbots and headline-grabbing model launches to examine how AI is moving into aviation, semiconductors, data centers, biotech, and national industrial strategy.

    First, we discuss the FAA’s AI-powered SMART platform, a system designed to help predict aviation risks and operational problems. Although officials say SMART was not involved in a recent airline ground stop, the story raises important questions about trust, transparency, auditing, accountability, and human oversight when AI is used in safety-critical environments.

    Next, we cover SK Hynix’s plan to invest about $1.05 billion over five years to support South Korean semiconductor suppliers. As advanced AI depends on high-bandwidth memory, specialized chips, manufacturing tools, and resilient supply chains, this investment highlights how AI progress increasingly relies on physical infrastructure and industrial ecosystems.

    We also examine reports that Samsung SDS is in early talks to host OpenAI and Anthropic in data centers. While no deal has been confirmed, the discussions show how major technology and industrial companies are positioning themselves as full-stack AI infrastructure partners, combining data-center capacity, power systems, IT services, construction, and memory supply.

    Finally, we look at Flagship Pioneering’s plan to invest an initial £100 million in a London-based unit focused on building “AI-native” platform companies. Known for creating Moderna, Flagship is betting that AI-driven discovery will transform biotech, healthcare, planetary science, and research by generating and testing ideas at scale.

    Together, these stories reveal a major shift: AI innovation is no longer only about smarter software. It is about the systems beneath it — aviation safeguards, semiconductor supply chains, data centers, power capacity, cooling, scientific platforms, procurement, and government-backed industrial strategy.

    AI Daily Podcast breaks down why the future of artificial intelligence will depend not just on better models, but on the trusted physical and institutional infrastructure needed to deploy them safely, responsibly, and at scale.
    Links:
    Questions swirl around FAA's new SMART AI system, though officials say no link to ground stop
    SK Hynix Plans $1.05B Push for South Korean Chip Suppliers
    Samsung SDS in Early AI Data Centre Talks With OpenAI and Anthropic, FT Says
    AP Business SummaryBrief at 2:31 a.m. EST
    Flagship Pioneering Commits Initial £100 Million to London Company-Creation Lab
  • AI Daily

    AI Enters Its Execution Era

    13/11/2026 | 28 min
    AI Daily Podcast explores the latest innovations reshaping artificial intelligence technology, from the infrastructure powering frontier models to the scientific breakthroughs AI may soon help unlock.

    In this episode, we examine how AI is entering a new industrial execution phase. The conversation begins with Nvidia, whose future leadership and long-term strategy raise important questions about the company’s role at the center of the AI ecosystem. Beyond GPUs, Nvidia’s dominance depends on a full-stack platform spanning CUDA software, networking, cloud partnerships, server systems, memory coordination, and global supply chains.

    We also look at Google’s Gemini 4 Hydrogen, a fast and practical model designed for agentic workflows. Rather than focusing only on maximum theoretical power, Hydrogen points toward operational AI systems that can use tools, write code, run tests, search information, and complete complex multi-step tasks. Its adjustable reasoning levels suggest a future where AI dynamically balances speed, cost, and quality.

    The episode then turns to a potentially historic development: reports that OpenAI may have produced a possible path toward solving the Navier–Stokes problem, one of mathematics’ seven Millennium Prize Problems. If verified, this would mark a major moment for AI-assisted discovery, showing how advanced models might contribute not just to automation, but to new knowledge.

    At the same time, we discuss why caution is essential. Mathematical breakthroughs require rigorous verification, human expertise, ethical data practices, and careful attribution. AI may become a powerful collaborator in science and education, but it does not replace deep understanding, critical judgment, or the human ability to decide what counts as meaningful truth.

    Key takeaway: the next era of AI will be defined not only by bigger models or faster chips, but by dependable, integrated systems that combine hardware, cloud infrastructure, safety, governance, developer tools, and human expertise.
    Links:
    Nvidia without Jensen Huang. What lies ahead for Asian chipmakers?
    Google unveils Gemini 4 Hydrogen, its new AI model built for speed
    Despite AI breakthroughs, math majors say human thinking is important
Más podcasts de Tecnología
Acerca de AI Daily
Everything that's happening in the rapidly changing world of Artificial Intelligence, OpenAI, Bard, Bing, Midjourney, and more.
Sitio web del podcast

Escucha AI Daily, Inteligencia Artificial y muchos más podcasts de todo el mundo con la aplicación de radio.es

Descarga la app gratuita: radio.es

  • Añadir radios y podcasts a favoritos
  • Transmisión por Wi-Fi y Bluetooth
  • Carplay & Android Auto compatible
  • Muchas otras funciones de la app
AI Daily: Podcasts del grupo