ai
Eye on A.I. is a biweekly podcast, hosted by longtime New York Times correspondent Craig S. Smith. In each episode, Craig will talk to people making a difference in artificial intelligence. The podcast aims to put incremental advances into a broader context and consider the global implications of the developing technology. AI is about to change your world, so pay attention.
A new study by Virtana reveals that 60% of enterprises cannot automatically identify the root cause when their AI workloads fail. This lack of observability makes it difficult to implement effective remediation, leading to a high percentage of AI workloads failing.
Paul Appleby, CEO of Virtana, predicts a shift in how AI return on investment (ROI) will be measured. He believes the focus will move from pure model performance to operational efficiency and cost-effectiveness, as the initial 'gold rush' mentality for AI investment subsides.
Paul Appleby, CEO of Virtana, explains that true observability for AI factories is challenging due to the complexity of the entire system, from data pipelines to orchestration and infrastructure. He highlights that many current solutions only offer partial telemetry, making it hard to pinpoint issues.
Paul Appleby of Virtana points out that inefficient AI operations lead to idle GPUs and significant energy waste. He emphasizes the need for better utilization and throughput to maximize ROI from AI investments, and reduce the environmental impact of data centers.
Paul Appleby, CEO of Virtana, states that the rapid investment in AI infrastructure, often termed 'AI factories,' is outpacing the implementation of necessary governance and controls. This oversight increases operational risk for businesses.
Paul Appleby of Virtana explains that their observability platform is designed to provide end-to-end visibility for AI factories, a critical need as AI adoption accelerates. The platform aims to help companies manage complex systems and identify issues that hinder performance and ROI.
Braden Rogers, Chief Customer Officer at Island, discussed the company's new AI-focused security features designed to manage the risks associated with autonomous agents and AI tools integrated into enterprise environments. These capabilities aim to provide visibility and control over AI usage, ensuring compliance and data protection.
The conversation highlighted how the traditional browser, initially designed for information consumption, has evolved into an application delivery platform. Island aims to transform the browser into a more robust application delivery platform by building in local mechanics and policies for user experience, data protection, and productivity.
Braden Rogers described the challenge enterprises face with the 'chaos' of AI tools, emphasizing the need to empower users while maintaining safety and regulatory compliance. He advocates for an 'AI first' mindset, integrating sanctioned AI into workflows rather than blocking everything, to foster user productivity and innovation.
Island's AI Protect feature allows for safe usage of any AI application, whether consumer or enterprise-grade. Braden Rogers explained that AI Protect can live anywhere the user does, whether in an enterprise browser, a consumer browser with an Island extension, or via Island Desktop, asserting policies to govern AI resource usage.
The discussion touched on the risks of autonomous agents and AI tools designed for consumers being integrated into enterprise environments. Braden Rogers highlighted that organizations struggle with the rapid influx of these tools, needing to balance user empowerment with fundamental security and regulatory requirements.
Kriti Sharma, CEO of IFS's Nexus Black unit, stated that their AI solutions are designed to deliver value to clients within approximately three weeks. This is achieved by focusing on real-world problems and being present on-site with clients.
Nexus Black tailors AI solutions for field technicians by considering their practical constraints, such as the inability to remove safety gloves to type. This approach leads to the use of voice transcription and other adaptable form factors for their AI tools.
Kriti Sharma highlighted the challenges of deploying AI in industrial environments, particularly in offshore settings where Wi-Fi is often unavailable. Her team prioritizes building AI solutions that operate effectively on the edge, with compressed knowledge deployed on devices.
Nexus Black worked with William Grant, the maker of Glenfiddich and Hendrick's Gin, to address production loss. They found that the distillery was spending 38% of its time on emergency and corrective fixes for machinery, impacting batch production.
Nexus Black leverages AI models like Claude from Anthropic to interpret complex industrial documents, such as piping and instrumentation diagrams. This allows them to help technicians understand how issues with one component might affect other parts of a plant.
Kriti Sharma described Nexus Black as having a team of 'elite engineers' who are experts in building production-grade AI products. This team combines their technical prowess with deep subject matter expertise to solve challenging problems in industrial settings.
Rubric has launched "Rubric Agent Cloud," a new offering that combines their AI platform with data and identity security. The goal is to manage the risks associated with AI agents operating within organizations. This move follows Rubric's acquisition of Preda base, an AI infrastructure company founded by Doug, who previously worked at Google.
Rubric positions its Agent Cloud as a security layer above existing agent orchestration tools. The company focuses on providing consistent guardrails and runtime controls for AI agents, regardless of where they are built or deployed. This is seen as crucial because AI agents are becoming more complex and integrated into business operations.
Doug, co-founder and CEO of Preda base, shared his background in AI, including his work at Google on the Google Assistant and Google Cloud's AI platform (later Vertex AI). His early focus was on resilient machine learning algorithms for privacy-preserving use cases. He also mentioned his experience at Uber building machine learning frameworks.
Rubric's core business began with data and cyber resilience, focusing on business continuity against downtime and cyber threats. Their philosophy evolved to 'assume breach,' emphasizing resilience rather than just prevention. With the rise of AI agents, Rubric has expanded its focus to address the unique security challenges posed by these new systems.
The increasing adoption of AI agents presents a significant challenge for organizations due to a lack of robust security frameworks. Unlike traditional software, AI agents' operations are harder to monitor and control, making them vulnerable to risks such as unauthorized data exfiltration. Rubric aims to solve this with their Agent Cloud offering.
Philip Goff argues that Galileo's decision to separate consciousness from the domain of science, by deeming it purely mathematical and quantitative, created the mind-body problem. Goff suggests that this division, influenced by religious context and Galileo's own philosophical innovations, led to science focusing on the physical world while leaving subjective experiences like qualia outside its scope. He proposes pansychism as a way to reintegrate consciousness into a scientific framework.
Philip Goff notes a historical trend in attributing consciousness to a wider range of beings, starting with only humans, then extending to animals, and recently to insects and reptiles. This progression suggests a broader understanding of consciousness, which aligns with the pansychist view that consciousness is a fundamental property present even in the simplest building blocks of the universe.
Philip Goff proposes pansychism as a potential solution to the mind-body problem, suggesting it can help reintegrate consciousness into the scientific framework from which Galileo arguably excluded it. He argues that Galileo's division of reality into a quantitative physical world and a non-quantitative mental world, while enabling scientific progress, left subjective experience unaddressed.
Philip Goff challenges the common assumption that Galileo's separation of consciousness from the physical world was solely for scientific expediency. Goff posits that this move was actually quite 'rebellious' against the prevailing Aristotelian and Church views, which attributed qualities like colors and sounds to the physical world itself, not just the soul. Galileo's division, by placing subjective qualities outside the purview of mathematical science, set the stage for the mind-body problem.