IBM’s AI strategy, Apple AI leadership shakeup, GenAI’s economic impact, Runway’s video leap...

In a wide-ranging interview, IBM CEO Arvind Krishna discussed the company’s AI evolution, the risks and opportunities of the current generative AI boom, and IBM's focus on enterprise solutions and quantum computing. Krishna candidly admitted that IBM’s early Watson push—especially into healthcare—was premature, but emphasized that foundational research and infrastructure are now paying dividends as IBM pivots to modular, scalable AI offerings and hybrid cloud deployments. He remains skeptical of the likelihood that today’s large language models will achieve AGI, suggesting that academic breakthroughs—rather than corporate R&D—are more likely to drive the next leap in AI capabilities. Krishna also cast doubt on the sustainability of current, massive AI-related capital expenditures, especially in the consumer space, and predicted significant but not catastrophic job displacement, with new roles emerging as AI augments rather than simply replaces human workers. IBM’s bet: productivity gains from AI are real if used to empower employees, with quantum computing positioned as a future “add” rather than a total replacement for CPUs/GPUs. The Verge

Apple has replaced its head of artificial intelligence, John Giannandrea, with Amar Subramanya, a former Microsoft and Google DeepMind executive, in a high-profile move as the company races to catch up with rivals in generative AI. Subramanya, described as a “renowned AI researcher,” will oversee Apple’s Foundation Models, ML research, and AI safety—reporting to Craig Federighi. This leadership shakeup comes as Apple faces mounting criticism for lagging behind competitors, particularly with Siri’s stalled evolution and delayed AI-centric features. Sources note that CEO Tim Cook had lost confidence in Giannandrea's ability to execute on product development, and that Subramanya’s hiring is seen as a crucial step for Apple to regain footing in the AI arms race. CNBC Engadget Financial Times The Guardian

Runway has launched its Gen-4.5 text-to-video AI model, which claims to outperform Google and OpenAI on key benchmarks for high-definition, photorealistic video generation. According to Runway and independent coverage, Gen-4.5 delivers “cinematic and highly realistic outputs” with detailed visual accuracy and improved adherence to user prompts. While limitations remain—such as occasional issues with object permanence and causal reasoning—the model’s ability to produce lifelike detail and nuanced motion (including fluid dynamics) has raised the bar for creative AI tools, intensifying competition with OpenAI’s Sora and similar models. CNBC The Verge

IEEE Spectrum highlights a pivotal shift in artificial intelligence: after a decade of progress driven by ever-larger datasets and models, leading labs are now investing billions to develop interactive “classrooms” or reinforcement learning (RL) environments. These RL environments allow machines to experiment, fail, and learn through action—mirroring human problem-solving in messy, unpredictable digital and real-world settings. The article argues that future leaps in AI capability will stem not just from more data, but from immersive, interactive environments that teach agents to reason and adapt, particularly for complex, multi-step workflows and high-stakes scenarios like disaster relief. IEEE Spectrum

Nvidia has deepened its partnership with Synopsys by taking a $2 billion stake, aiming to combine Nvidia’s AI platform with Synopsys’ expertise in computer-modeled design and engineering. This deal underscores Nvidia’s centrality in the global AI infrastructure buildout, which is now creating widespread shortages of advanced chips—driving up costs for both data centers and consumer electronics. The resulting “chip crunch” threatens to increase prices for a broad range of gadgets, as demand for AI-capable hardware outpaces supply. CNBC CNBC CNBC

Hewlett Packard Enterprise (HPE) is upgrading its Private Cloud AI stack with new Nvidia technology and launching an AI Factory Lab in France to accelerate enterprise AI adoption. Meanwhile, HSBC has partnered with European AI startup Mistral AI to implement generative AI across its banking operations, following on the heels of Bank of America’s $4 billion tech investment. Both moves reflect the financial sector’s urgency to leverage AI for operational efficiency and competitive advantage. The Register The Register

Multiple sources confirm that Apple’s AI platform—particularly Siri—has struggled to keep pace with advances from competitors like Google and OpenAI. Insiders attribute much of the stagnation to organizational and leadership challenges under Giannandrea, whose departure is linked to the company’s inability to deliver a more personalized, AI-driven Siri despite years of effort. The new AI leadership is expected to prioritize the development of foundational models, safety protocols, and the integration of advanced generative AI into core Apple products. Gizmodo TechCrunch 9To5 Mac The Guardian

The US FDA has announced the rollout of “agentic AI” tools—AI systems capable of autonomously completing multi-step tasks—for staff engaged in premarket reviews, inspections, compliance, and administrative functions. While the technology is described as exploratory and does not replace human judgment, all AI outputs are subject to human validation to ensure reliability. This marks a significant step in the digital transformation of regulatory agencies, as they seek to balance efficiency gains with the need for oversight and accountability. STAT News

Samsung has formed a dedicated acquisition unit to compete in the race to supply advanced memory chips for AI applications. As the AI hardware arms race heats up, the South Korean technology giant is looking to mergers and acquisitions to secure crucial capabilities and maintain its market share against rivals like Nvidia and TSMC. Financial Times

A substantive debate between Financial Times and MIT Technology Review columnists examines why AI’s impact on productivity is not yet apparent in broad economic data. While some executives and studies cite modest early returns—only 5% of generative AI projects have produced measurable business value—others argue that a lag is typical for transformative technologies. Optimists predict a coming “J curve” of productivity, while skeptics warn that current AI tools are too narrow to revolutionize major sectors. The consensus: meaningful gains will require more than cost-cutting and automation; AI must be integrated to augment workers and create new roles. MIT Technology Review Financial Times

A new study from Anthropic (covered by ZDNet) analyzed 100,000 user interactions with the Claude chatbot and found that AI helped users complete tasks 80% faster on average. Projecting forward, the study estimates that current AI models could boost annual labor productivity growth in the US by 1.8%—potentially doubling the rate over the next decade. However, the report cautions that the diversity of use cases and uneven ROI across organizations makes it difficult to precisely quantify AI’s macroeconomic impact at this stage.

--- Zdnet