The Useful Unknown

Berk Bayri
  • 0.0 (0)
  • Management
  • Updated Daily

A podcast about strategy, AI, innovation, and the messy space between knowing and doing. I explore how better questions, sharper thinking, and useful uncertainty can lead to better decisions, products, and work.

Episodes

  1. Sep 23

    Build for the wave you can't name

    The web, social, mobile, UX, product, data and AI all arrived with different vocabulary and surprisingly familiar organizational behavior. In the final episode, Berk Bayri argues that prediction is overrated and adaptation is the more durable capability. The previous eight episodes — problem selection, use cases, pilots, handoffs, ownership, teams, judgment and memory — resolve into one operating idea: the ability to learn on purpose. The goal is not “full AI adoption,” but an organization that can form a reliable opinion about something new, test it cheaply, stop intelligently, scale deliberately and begin the next decision from what it already learned. Featuring a Satya Nadella excerpt on seeing the web shift while missing where the value layer would emerge. References Stanford HAI, 2026 AI Index Report — Economy. Organizational AI adoption reached 88% of surveyed organizations in 2025; generative AI was used in at least one business function by 70%; agent deployment remained comparatively early. https://hai.stanford.edu/ai-index/2026-ai-index-report/economy Deloitte AI Institute, State of AI in the Enterprise 2026. Survey of 3,235 business and IT leaders across 24 countries. The report describes a continuing gap between broad access/experimentation and deeper process redesign, transformation and production scaling. https://www.deloitte.com/us/en/about/press-room/state-of-ai-report-2026.html Thabit Atobishi and Saeed Nosratabadi, Dynamic Capabilities for AI-Enabled Exploration: Antecedents, Mechanisms, and Innovation Outcomes (2026 preprint). Survey-based study of 245 senior executives; proposes a mediated relationship in which organizational readiness and technology compatibility support sensing capability and AI-enabled exploration, which in turn relate to innovation performance. Treat as emerging academic evidence, not settled universal causality. https://arxiv.org/abs/2607.02645 Alexander Ettinger, Enterprise Architecture as a Dynamic Capability for Scalable and Sustainable Generative AI Adoption (2025). Literature review plus 16 expert interviews; highlights strategic alignment, governance, organizational agility and data-governance maturity in scaling GenAI. https://arxiv.org/abs/2505.06326

  2. Sep 23

    The company that forgets

    Organizations store enormous amounts of information and still struggle to answer a basic question: why did we decide this? Episode 8 separates search from memory and documents from knowledge. Berk Bayri argues that some of the most valuable organizational context is reasoning — the thesis, evidence, dissent, confidence and constraints behind a decision — and that AI becomes far more useful when it can inspect how a company has thought, not merely retrieve what it has stored. Featuring a Ray Dalio excerpt on converting historical patterns into indicators and explicit decision rules. References Anne S. R. Marx, Ricardo M. Avelino, Torbjørn Netland and Mennatallah El-Assady, Do We Have the Knowledge We Need? Rethinking Human-AI Decision-Making in Corporations (2026 position paper). Describes organizational knowledge as fragmented across software systems, tacit expertise and manual documents, and explores how knowledge availability should affect human/AI agency. https://arxiv.org/abs/2606.15575 Bhada Yun et al., Generative AI in Knowledge Work: Design Implications for Data Navigation and Decision-Making (2025). Study involving knowledge workers and product managers; highlights adaptable user control, transparent collaboration and integration of background knowledge with external information, while noting risks including overreliance and missing context. https://arxiv.org/abs/2503.18419 NIST, AI Risk Management Framework. Background for provenance, traceability, accountability, lifecycle risk and the need to distinguish authoritative information from model-generated output. https://www.nist.gov/itl/ai-risk-management-framework

  3. Sep 23

    Judgment is the scarce resource

    One serious study finds AI-assisted customer-service workers faster; another finds experienced developers slower. Episode 7 uses that tension to challenge the idea that productivity is a property of the model. Berk Bayri looks at expertise, tacit context, meaningful human oversight, creative abundance and the danger of polished output arriving before the thinking is mature. The episode reframes “human in the loop” as an actual design problem: which human, with what authority, information and consequence? Featuring a Brian Eno excerpt on “premature sheen” and why finish can arrive before an idea is ready. ──────────────────────────────────────── References Erik Brynjolfsson, Danielle Li and Lindsey R. Raymond, Generative AI at Work, Quarterly Journal of Economics 140(2), 2025. AI assistance increased customer-service productivity by 15% in the studied firm; gains were heterogeneous and stronger for less-experienced workers. https://academic.oup.com/qje/article/140/2/889/7990658 METR, Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity, July 2025. RCT with 16 experienced developers and 246 real tasks; AI access increased task completion time by 19% in this specific setting. https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/ METR, We are Changing our Developer Productivity Experiment Design, February 2026. METR says later tools likely provide larger productivity benefits, but selection effects prevent a reliable estimate from its follow-up experiment. https://metr.org/blog/2026-02-24-uplift-update/ Harvard Business School Institute for Business in Global Society, AI Won’t Make the Call: Why Human Judgment Still Drives Innovation, September 2025. Summarizes research on the continuing role of experience and judgment in evaluating ideas and strategy. https://www.hbs.edu/bigs/artificial-intelligence-human-jugment-drives-innovation Anne S. R. Marx et al., Do We Have the Knowledge We Need? Rethinking Human-AI Decision-Making in Corporations (2026 position paper). Discusses fragmented organizational knowledge and proposes allocating human/AI agency based on task attributes and knowledge availability. https://arxiv.org/abs/2606.15575

  4. Sep 21

    Build the team around the problem

    “Cross-functional” often means fourteen people on a recurring meeting, each representing a different function and returning to a different priority queue. Episode 6 offers a different model: a small core that protects continuity around the problem, with specialists moving in and out as their knowledge becomes relevant. Berk Bayri draws on creative-technology and multidisciplinary work to explain shared vocabulary, context evaporation, temporary teams and the leadership value of high-quality access with low-frequency interference. Featuring an Ed Catmull excerpt on building environments where strong people can collaborate into something greater. References Yichun Peng et al., Structural Diversity Drives Disruptive Scientific Innovation (2026 preprint). The study analyzes a century-scale dataset of approximately 260 million scientific publications and reports that structural diversity — bridging distinct prior collaboration/knowledge communities — is a strong predictor of disruptive scientific innovation. This is scientific-team research, not direct evidence that one enterprise team structure is universally optimal. https://arxiv.org/abs/2605.12514 University of Minnesota College of Continuing and Professional Studies, How Cross-Functional Collaboration Drives Success (2025). Used as background on cross-functional collaboration and the challenge of bridging departmental expertise. https://ccaps.umn.edu/story/how-cross-functional-collaboration-drives-success Deloitte / NIST material from earlier episodes remains relevant background for cross-functional governance of AI and the need to connect business, technical and risk functions. https://www.deloitte.com/us/en/insights/topics/emerging-technologies/ai-agents-scaling-faster.html https://www.nist.gov/itl/ai-risk-management-framework

  5. Sep 14

    Who owns the new?

    Where should AI live — Technology, Product, Strategy, Marketing, a dedicated lab? Episode 5 argues that the reporting line is often the wrong first question. The more useful design problem is decision rights: who validates the problem, funds exploration, sets boundaries, stops duplication, decides when evidence is sufficient and owns the consequence at scale? Drawing on multidisciplinary creative-technology work, Berk Bayri describes a model where a small central capability convenes and learns while long-term ownership moves toward the part of the business that can operate the result. Featuring a short Steve Jobs excerpt on collaboration and clear responsibility at Apple. References Deloitte Insights, Agentic AI Is Scaling Faster Than Guardrails, April 2026. Based on Deloitte’s survey of 3,235 business and IT leaders across 24 countries; only 21% reported mature governance for agentic AI. Deloitte highlights cross-functional governance involving IT, legal, compliance and business leaders.https://www.deloitte.com/us/en/insights/topics/emerging-technologies/ai-agents-scaling-faster.html PwC, AI Agent Survey, May 2025. Survey of 308 U.S. executives. Among companies using agents, 45% reported fundamentally rethinking operating models and 42% redesigning processes around agents. Trust was lower for higher-stakes autonomous tasks.https://www.pwc.com/us/en/tech-effect/ai-analytics/ai-agent-survey.html NIST, AI Risk Management Framework. Background for the claim that governance and accountability should span design, development, use and evaluation rather than sit in one technical function.https://www.nist.gov/itl/ai-risk-management-framework The concept of organizational ambidexterity is established management theory describing the tension between exploitation of current capabilities and exploration of new ones. This episode uses it as an interpretive lens rather than claiming one universally correct org design.

  6. Sep 6

    The handoff nobody designed

    “The pilot was successful” is not the end of the story. Episode 4 looks at the difficult transition from protected experiment to ordinary operations: budgets, permissions, support, monitoring, training, governance, workflow redesign and the context that has to travel with the artifact. Berk Bayri argues for co-ownership before transfer and asks a practical scaling question: what must become true for this system to remain valuable when the pilot team stops protecting it? Featuring a short excerpt from Satya Nadella on why AI adoption at enterprise scale is fundamentally a change-management and process-change problem. References Deloitte AI Institute, State of AI in the Enterprise 2026. In a survey of 3,235 business and IT leaders across 24 countries, 25% said they had moved 40% or more of AI pilots into production; 54% expected to reach that level in the following three to six months. Deloitte also reports that only 30% were redesigning key processes around AI.https://www.deloitte.com/us/en/about/press-room/state-of-ai-report-2026.html Cloudera, “The Great AI Re-Architecture” press release, August 11, 2026. Based on a survey of 1,500 enterprise architects, cloud infrastructure leads and data architects. 95% reported delaying or canceling AI initiatives in the prior year because of data governance, compliance or regulatory challenges; 72% said their data architecture required significant overhaul. This is vendor-sponsored research and should be described as such if quoted on air.https://www.cloudera.com/about/news-and-blogs/press-releases/2026-08-11-ninety-five-percent-of-enterprises-have-delayed-ai-projects-as-infrastructure-limitations-spark-the-great-ai-re-architecture.html NIST, AI Risk Management Framework and Generative AI Profile. Background for lifecycle governance, monitoring, evaluation and risk management.https://www.nist.gov/itl/ai-risk-management-frameworkhttps://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.600-1.pdf

  7. Aug 31

    A pilot is a question, not a small product

    The best pilot is not the smallest product you can build. It is the smallest experiment that still exposes the uncertainty capable of killing the idea. Pilots become dangerous when they acquire a second job: prove that the project deserves to scale. Episode 3 reframes the pilot as a decision instrument rather than a miniature product. Berk Bayri looks at polished demos, dirty data, whole-workflow measurement, stop criteria, governance and the politics of sponsorship. The central idea is simple: a successful pilot is one that leaves the organization knowing something important it did not know before — even when that knowledge is “do not continue.” Featuring a short excerpt from Jeff Bezos on the difference between experimental failure and operational failure. References UK Government / Behavioural Insights Team, AI-Assisted vs Human-Only Evidence Review: Results from a Comparative Study (published 2025). The AI-assisted review was completed 23% faster overall, with large time savings in analysis and synthesis but more time required for revisions. The authors explicitly caution that this is a case study and not broadly generalisable.https://www.gov.uk/government/publications/ai-assisted-vs-human-only-evidence-review/ai-assisted-vs-human-only-evidence-review-results-from-a-comparative-study NIST, AI Risk Management Framework. The voluntary framework is designed to incorporate trustworthiness considerations into the design, development, use and evaluation of AI systems. NIST’s current site notes that AI RMF 1.0 is under revision and that a Critical Infrastructure Profile concept note was released in April 2026.https://www.nist.gov/itl/ai-risk-management-framework NIST, Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile (NIST-AI-600-1). Used as background for lifecycle risk, evaluation and governance considerations specific to generative AI.https://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.600-1.pdf

  8. Aug 31

    Stop looking for AI use cases

    The use-case workshop feels productive because it produces visible output. But a spreadsheet of AI ideas can become a substitute for understanding where the organization is actually paying for friction, uncertainty and bad decisions. Most companies do not have an idea shortage. They have a problem-selection problem. In Episode 2, Berk Bayri takes apart the familiar “AI use-case” workshop and replaces it with a more demanding question: where is the organization repeatedly paying for friction, uncertainty or poor decisions? Through sales, content and productivity examples, the episode shows why the same AI tool can create very different outcomes depending on task, expertise and workflow. It also explains why specificity matters more as software gains more autonomy. Featuring a short excerpt from Andrew Ng on the advantage of concrete, falsifiable ideas. References Stanford HAI, 2026 AI Index Report — Economy. 88% organizational AI adoption; 70% of organizations using generative AI in at least one function; agent deployment remained in the single digits across nearly all business functions.https://hai.stanford.edu/ai-index/2026-ai-index-report/economy Deloitte AI Institute, State of AI in the Enterprise 2026. Survey of 3,235 business and IT leaders across 24 countries. 30% said they were redesigning key processes around AI; 37% reported relatively surface-level usage with little or no change to underlying processes; only 21% reported mature governance for agentic AI.https://www.deloitte.com/us/en/about/press-room/state-of-ai-report-2026.html Erik Brynjolfsson, Danielle Li and Lindsey R. Raymond, Generative AI at Work, Quarterly Journal of Economics 140(2), 2025. AI assistance increased customer-service agent productivity by 15% in the studied firm, with heterogeneous effects across workers.https://academic.oup.com/qje/article/140/2/889/7990658 METR, Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity. Randomized controlled trial: in this specific setting, AI access increased task completion time by 19%.https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/ METR, We are Changing our Developer Productivity Experiment Design, February 2026. METR says late-2025 tools likely provide greater speedups, but selection effects make the newer estimate unreliable.https://metr.org/blog/2026-02-24-uplift-update/

  9. Aug 31

    Solving the wrong problem brilliantly

    The most dangerous transformation failure is not always a bad solution. Sometimes it is a very good solution aimed at a problem that never deserved the effort. This episode is about how leaders can separate technological momentum from actual strategic value. AI can make teams faster, more productive and more convincing before anyone has proved that the work matters. In this episode, we look at the organizational reflex behind “we need a website,” “we need an app,” and now “we need AI” — and why the better starting point is evidence, consequence and the quality of the problem itself. The episode moves from support automation and UX research to executive pressure, innovation fatigue and the hidden cost of spending organizational belief on the wrong bet. Featuring a short external excerpt from Tim Brown on the importance of starting with the right question. References Stanford Institute for Human-Centered Artificial Intelligence, 2026 AI Index Report — Economy. Organizational AI adoption reached 88% of surveyed organizations in 2025; generative AI was used in at least one business function at 70%; agent deployment remained in the single digits across nearly all business functions.https://hai.stanford.edu/ai-index/2026-ai-index-report/economy Deloitte AI Institute, The State of AI in the Enterprise: The Untapped Edge (2026). Survey of 3,235 business and IT leaders across 24 countries. Deloitte reports that 34% of companies described AI as being used to deeply transform their business; 37% reported surface-level use with little or no change to underlying processes.https://www.deloitte.com/us/en/about/press-room/state-of-ai-report-2026.html NIST, AI Risk Management Framework. Used as background for the emphasis on context, measurement, governance and lifecycle thinking rather than technology-first deployment.https://www.nist.gov/itl/ai-risk-management-framework

  • 9 Episodes

About

A podcast about strategy, AI, innovation, and the messy space between knowing and doing. I explore how better questions, sharper thinking, and useful uncertainty can lead to better decisions, products, and work.

Information

  • Creator
    Berk Bayri
  • Years Active
    2026
  • Episodes
    9
  • Rating
    Clean
  • Copyright
    © Berk Bayri
  • Show Website
    The Useful Unknown