I recently caught up for a morning coffee with Charles Sansbury, CEO of Cloudera. We spoke about how AI is changing how leaders view data infrastructure and the significance of organizations capable of operating at exascale. Charles shared how his conversations with Fortune 50 companies have shifted from theoretical AI applications to practical concerns about return on investment, data quality, and infrastructure. A key part of this transition is the emerging trend of ‘private AI’ -  targeted models trained on proprietary company data.

In our discussion, we explored the changing dynamics between Cloud and on-premises solutions, with Charles noting that many large companies are reconsidering on-premises options due to rising energy and compute costs. He emphasized that while Cloud adoption continues to grow broadly, the largest enterprises are taking a more nuanced approach based on cost, security, and manageability concerns.

Our discussion concluded with reflections on leadership in the AI era, the importance of team-based goals, and how organizations might structurally evolve as AI automates entry-level positions while creating new opportunities.

Key Insights

1. Data quality and infrastructure are becoming strategic priorities as companies realize that AI models require well-organized, high-quality data to deliver useful results.

2. ‘Private AI’ (models trained on proprietary company data) is emerging as the preferred architecture for large global corporations, offering better performance and security than models trained on generic public data.

3. The promised cost savings of the cloud have not always materialized for large enterprises, leading to a reevaluation of on-premises solutions for workloads that run consistently at scale.

4. Executive leadership should focus on "doing more with the same" rather than downsizing, using AI to enhance productivity through code generation, content creation, and automated processes.

5. Future-ready organizations need team members who have weathered both success and challenges, who understand the consequences of their decisions, and who actively experiment with emerging technologies like AI.

For Tom Chatfield, a British technology philosopher and author, there is nothing inevitable about technological progress. In his latest book, ‘Wise Animals: How Technology Has Made Us What We Are’ he explores how humans have co-evolved with technology throughout history. In our discussion, Tom deconstructed the common narrative of technological inevitability, using examples like the Incas' use of wheels only as toys and the complex history of Gutenberg's printing press to illustrate how technological adoption is deeply intertwined with cultural values and societal choices rather than following a predetermined path.

Tom emphasizes that while AI doesn't possess true agency, it can be seductive and influential through its design and training. AI, after all, is also a piece of cultural technology. He advocates for viewing AI as a cognitive prosthesis that should complement human strengths while compensating for our weaknesses, particularly in areas like repetitive tasks and statistical analysis.

This episode concludes with some practical insights about navigating technological change, both at organizational and personal levels. Tom believes in the importance of taking time to ask the right questions in an age of instant answers, fostering constructive disagreement to avoid groupthink, and maintaining human agency in technological systems. He also shares personal perspectives on raising children in an AI-enabled world, emphasizing the need for supervised exploration and critical engagement with technology.

Key Insights

1.  Technology adoption is not inevitable but rather a negotiation between cultural values, practical needs, and societal choices - leaders should actively shape how technology serves their organization's values and goals

2. In an era of instant answers, competitive advantage comes from asking better questions and taking time to reframe problems beyond default assumptions

3. Organizations need to build in structures for constructive disagreement and cognitive diversity to avoid groupthink and maximize the benefits of both human and artificial intelligence

4. AI should be viewed as a cognitive prosthesis that complements human strengths rather than replacing human agency - focus on automation of areas where humans are naturally weak

5. Success with AI requires strong governance frameworks that enable systematic learning and accountability, similar to aviation safety systems

On a recent trip to Toronto, I caught up with Yao Morin, the global CTO of JLL, a commercial real estate company. We discussed JLL's early adoption of generative AI technology, despite the real estate industry's traditionally conservative approach to new technologies. Yao explained how JLL has implemented AI tools like JLLGPT to process vast amounts of unstructured data in multiple languages, making it easier for employees to access and utilize information from contracts, research reports, and other documents.

For Yao, the technical aspects of implementing AI in a large organization included managing the challenges of data access control, the importance of user feedback, and the need for accuracy in specialized tools. She emphasized the significance of building trust in AI systems by providing clear references to data sources and implementing human-in-the-loop verification processes. We also discussed the potential future developments in AI, such as world models that understand causation and physical properties, which could be particularly relevant for facilities management and other real-world applications.

In our discussion, Yao shared her views on the importance of community-driven development and focusing on automating small, repetitive tasks rather than replacing entire job roles. She also explained the unique challenges of working with "wide data" in commercial real estate and the potential for AI to unlock synergies between different business lines by organizing and connecting diverse data sources. Our chat concluded with Yao's vision for the future of commercial real estate technology and her focus on solving scalable data collection and integration problems.

Key insights

1. Early adoption of generative AI can provide a competitive advantage, even in traditionally conservative industries.

2. Building trust in AI systems through transparency and human-in-the-loop processes is crucial for successful implementation.

3. Focusing on automating small, repetitive tasks can lead to significant productivity gains without threatening job security.

4. Community-driven development and user feedback are essential for driving AI adoption within an organization.

5. Solving data collection, cleansing, and integration challenges is key to unlocking the full potential of AI in data-rich industries.

Ahead of the London Frieze Art Fair, I spoke with one of the world’s most fascinating artists, Von Wolfe, who for a number of years now has explored the intersection of art, technology, and artificial intelligence. As we sat in his studio in front of an oversized oil canvas and a sophisticated technology stack, Von Wolfe shared with me his unique approach to using machine learning and generative AI to create art. Our conversation delved into the history of art, from Renaissance masters to contemporary AI-powered creations, drawing parallels between traditional artistic methods and modern technological approaches. I was introduced to Von Wolfe through his longtime curator and art dealer, Virginia Damtsa, who has also appeared on this podcast.

This episode goes beyond the use of AI as a creative tool, and touches on philosophical questions about consciousness, creativity, and the nature of intelligence. Von Wolfe shares insights on how he trains AI models on his own artwork to create emergent properties and new artistic possibilities. The conversation also explores the potential of AI to uncover hidden historical artifacts and lost artworks by processing vast amounts of data from libraries and archives.

We also discuss the concept of authorship, the significance of patterns in nature and art, and the potential for AI to continue an artist's legacy beyond their lifetime. The conversation highlights the complex relationship between human creativity and artificial intelligence, suggesting that the future of art may involve a collaborative process between artists and advanced AI systems.

Key insights

1. AI can be used as a tool to enhance creativity and uncover new artistic possibilities, rather than replacing human artists.

2. Training AI models on specific, artistic datasets can lead to emergent properties and insights that were not explicitly programmed.

3. The integration of AI in creative processes may redefine concepts of authorship and the role of the individual artist.

4. AI has the potential to process and analyze vast amounts of historical data, potentially leading to groundbreaking discoveries in various fields.

5. The future of art and creativity may involve a symbiotic relationship between human artists and AI systems, with each complementing the other's strengths.

Tim Fu, founder of Studio Tim Fu, is one of the world’s leading experts in leveraging AI in architecture and construction. Fu established his practice in 2023 after working at Zaha Hadid Architects, and now heads a diverse team of architects and technologists, creating innovative workflows and designs that embrace AI's transformative power in architectural practice. Most recently, he was chosen to be a collaborating artist with OpenAI's new Sora generative AI platform.

In a discussion recorded in his London based studio, we spoke about the intersection of artificial intelligence and architecture. Fu explained that while AI can generate a wide range of design options quickly, architects and humans in general, are still essential. In his view, the ability to make decisions, understand cultural and historical context, and balance various factors like sustainability, engineering, and social space is what sets architects apart, even when using AI tools.

In this podcast episode, we covered a wide range of topics including parametric design, the potential for AI to create new architectural aesthetics, and how AI is changing the workflow of architecture firms. Fu described how his firm uses AI to generate matrices of design options for clients, allowing for more visual conversations early in the design process. Naturally, we also talked about the provocative issue of copyrighting architectural styles in the age of AI.

Key Insights

1. AI in architecture is a tool for exploration and idea generation, not a replacement for human creativity and decision-making.

2. Incorporating AI into architectural workflows can lead to more efficient and visual client interactions early in the design process.

3. The core value of architects lies in their ability to balance multiple factors (history, culture, engineering, sustainability) in decision-making, which AI cannot replicate.

4.  Building and leveraging proprietary datasets can be a way for firms to maintain their unique design DNA while using AI tools.

5.  Leaders should consider how AI might change workflows and client interactions in their industries, while still preserving the core expertise and values of their profession.