Insights from Human+Tech Week , the future of strategy, economics of AI, self-improving models, and more

📖In this issue

• Insights: AI & The Future of Strategy

• Insights: Economics and Division of Labor

• How humans use LLMs determines diagnostic accuracy

• Amazon sees net corporate jobs losses from AI

• Self-improving AI models adjust their own weights

• Podcast: Amir Barsoum on AI transforming services, pricing innovation, improving healthcare workflows, and accelerating prosperity

💡Insights: AI & The Future of Strategy 

On Thursday as part of Human+Tech Week I ran a short AI & The Future of Strategy event where Silicon Valley legends John Hagel, Charlene Li and I, together with event participants, dived deep into this highly under-addressed topic. Read the full event summary, or see below for very brief selected highlights.

🔄 Strategy in Flux: A Shifting Landscape
AI and democratized creation are transforming strategy from a top-down planning exercise into a dynamic, organization-wide learning and imagination process.

🌊 The Pressure is Real — But Fear Blocks Imagination
Fear doesn’t shorten timelines—it prevents leaders from thinking long-term, making emotional readiness the foundation for effective strategy.

🧠 AI + Humans: Complementary Intelligence in Strategy
AI enhances cognition, but humans remain essential for judgment, imagination, ethical framing, and asking the right strategic questions.

🔁 Zoom Out, Zoom In: Building Momentum Across Time Horizons
Combining long-term vision with short-term focused action allows organizations to build momentum while navigating uncertainty with purpose.

🔎 Knowledge + Trust: A Flywheel for Differentiation
In an AI-saturated world, competitive edge comes from trust-based relationships that unlock proprietary data and contextual insights.

⚖️ Strategic Polarities: Balancing Focus and Flexibility
Effective strategy manages the tension between long-term coherence and short-term agility, using vision and feedback to continuously adapt.

🏗️ Can Large Organizations Transform in the AI Era?
Yes—but only if they overcome fear, scale transformation from the edge, and are led by courageous, visionary leaders.

💡Insights: Economics and Division of Labor 

The invite-only Human-Agents Coevolution Symposium the day before the main Human+Tech Week brought together leaders to discuss a wide range of vital topics that will shape how humans and AI interact and create value.

I participated in the Economics and Division of Labor discussions. See the article for the full summary of what we covered. A few very brief selected highlights below.

🔁 Co-Evolution of Human and AI Capabilities
AI and humans are not in a static split of roles, but in a continuous process of mutual adaptation that reshapes work and capability.

🧠 Designing AI Workflows to Enhance Human Growth
Well-structured AI systems can challenge and support humans to grow, rather than diminishing their role through over-automation.

⚖️ The Strategic Divide: Augmentation or Replacement
Organizations face a critical choice between using AI to empower all workers or replace many with a few elite users.

🛠️ Preserving Joy, Craft, and Meaning in Human Work
AI can enable new forms of creativity, but risks hollowing out craftsmanship unless we deliberately elevate quality and care.

🧩 Elevating Critical Thinking and Metacognition with AI
AI tools that prompt reflection and deeper inquiry can strengthen human judgment, rather than replace it.

💰 Rethinking Economic Incentives and the Role of AI
AI presents an opportunity to reshape incentives and value flows beyond short-term profit toward long-term societal benefit.

🧭 Centering Human Agency in AI System Design
AI should be designed to amplify human autonomy, ethical reasoning, and sense-making, not replace them.

🧠🤖Humans + AI

How humans use LLMs determines diagnostic accuracy

“A paper by researchers at the University of Oxford found that while LLMs could correctly identify relevant conditions 94.9% of the time when directly presented with test scenarios, human participants using LLMs to diagnose the same scenarios identified the correct conditions less than 34.5% of the time.”

Just add humans: Oxford medical study underscores the missing link in chatbot testing
Paper: Clinical knowledge in LLMs does not translate to human interactions 

Amazon sees net corporate jobs losses from AI

Amazon is using AI in "virtually every corner of the company", with anticipated net corporate job losses from its current employee count of 1.5 million. CEO Andy Jassy said staff who embraced such changes would be "well-positioned" at the company.

Amazon boss says AI will replace jobs at tech giant

Self-improving AI models adjust their own weights

Another step towards self-improving AI. "Self-Adapting Language Models" (SEAL) update their own model weights to improve performance, guiding their own adaptation process.

Self-Adapting Language Models
Summary of key points

🎙️This week’s podcast episode

Why you should listen

As a venture builder and investor Amir brings a range of valuable insights into the realities of building AI-driven startups, including service innovation, pricing structures, healthcare and more.