Chapter 39 Further Reading

On AI Productivity Research

"Quantifying the Impact of AI on Developer Productivity" (GitHub, 2023) The landmark GitHub Copilot randomized controlled trial that found developers using AI coding assistants completed tasks 55% faster. Essential reading for anyone making the productivity case for AI tools. The full research paper is available through GitHub's research publication archive and has been widely covered in academic and trade press.

"Generative AI at Work" (Brynjolfsson, Li, and Raymond, 2023 NBER Working Paper) A rigorous study of AI assistance in customer service contexts that found 14% productivity improvement overall, with the largest effects for less experienced workers. The finding that AI assistance is particularly beneficial for newer employees has important implications for organizational AI adoption strategy.

"GPTs are GPTs: An Early Look at the Labor Market Impact Potential of Large Language Models" (OpenAI, Eloundou et al., 2023) While focused on labor market implications rather than individual productivity, this paper's task-level analysis of what AI affects provides a useful framework for practitioners thinking about which of their tasks are most AI-susceptible — and therefore most worth measuring.

"Experimental Evidence on the Productivity Effects of Generative AI" (Dell'Acqua et al., Harvard Business School, 2023) Research examining AI use among consultants at Boston Consulting Group. The study found significant productivity and quality benefits for tasks within AI's "frontier" of capability, but quality decrements for tasks outside it. The "frontier" concept — understanding where AI's capability actually extends — is directly applicable to the measurement framework in this chapter.

On Productivity Measurement and Knowledge Work

"Measure What Matters" by John Doerr The definitive guide to OKR (Objectives and Key Results) frameworks. While not AI-specific, the discipline of distinguishing outputs from outcomes, and of defining metrics that actually correlate with what you care about, is directly applicable to AI effectiveness measurement.

"The Effective Executive" by Peter Drucker Drucker's classic work on knowledge worker productivity — including his argument that managing one's time is the foundation of effectiveness — remains essential reading. His framework for identifying which activities actually create value is the foundation for the "stop doing" analysis described in this chapter.

"Thinking in Bets: Making Smarter Decisions When You Don't Have All the Facts" by Annie Duke Duke's work on probabilistic thinking and decision quality is relevant to AI measurement: many practitioners judge AI use by outcomes rather than process quality. Duke's framework for distinguishing good decisions from good outcomes helps practitioners design measurement systems that track what they can control (prompt quality, verification rigor, use case selection) rather than just outcomes.

"Work Rules!" by Laszlo Bock Former Google SVP of People Operations Bock's account of Google's data-driven approach to human performance. Particularly relevant: the chapters on performance measurement and the surprising (and counter-intuitive) findings that emerged when Google actually measured what drove performance outcomes.

On ROI Analysis and Value Demonstration

"How to Measure Anything: Finding the Value of Intangibles in Business" by Douglas Hubbard The essential guide for practitioners who need to quantify things that "can't be measured." Hubbard's techniques for establishing measurement ranges, using calibration, and applying probabilistic reasoning to uncertain quantities are directly applicable to AI ROI analysis.

"Value Proposition Design" by Osterwalder, Pigneur, Bernarda, and Smith While focused on business model design, the framework for mapping customer gains and pains applies directly to making the case for AI investment — whether to your own leadership or to clients. Particularly useful for Elena-type practitioners who need to articulate the value of AI-assisted consulting work.

On Quality Measurement

"Continuous Delivery: Reliable Software Releases through Build, Test, and Deployment Automation" by Humble and Farley While software-focused, the measurement discipline described in this book — particularly the emphasis on tracking defect rates, cycle times, and change failure rates — provides a model for the kind of rigorous quality measurement that knowledge workers often lack. The framework for distinguishing "leading indicators" (measurement points before work is delivered) from "lagging indicators" (measurement points after delivery) is broadly applicable.

"The Lean Startup" by Eric Ries The build-measure-learn loop and the concept of validated learning are directly applicable to AI effectiveness measurement. Ries's argument that measurement without a hypothesis generates noise rather than signal is important: know what you're testing before you collect data.

Measurement Tools and Frameworks

"Getting Things Done" by David Allen Allen's productivity system is relevant for practitioners who want to integrate AI effectiveness tracking into their existing work practices without creating additional overhead. His principle of capturing everything in trusted external systems applies directly to effectiveness journal maintenance.

Toggl Track (toggl.com) and similar time tracking tools For practitioners who want more precise time measurement than estimation allows, time tracking tools that integrate with common work environments can automate part of the data collection burden. The precision comes at the cost of friction in implementation.

Airtable (airtable.com) A flexible database-spreadsheet hybrid that many practitioners use for AI effectiveness tracking when they want more structure than a spreadsheet but less complexity than a full analytics platform. Its views and filtering capabilities make task-category analysis easier than raw spreadsheets.