As generative AI becomes embedded in everyday knowledge work, employers are facing a new workforce question: not simply whether employees can use AI, but whether they can take responsibility for what it produces. Learning scientist Rich Carr is tackling that question with “Stake Skills: Owning the Work AI Does for You,” a new handbook that argues AI accountability should become a distinct professional capability alongside hard and soft skills.
Published by Brain-centric, LLC, the book introduces a practical framework for documenting AI-assisted work, verifying machine-generated output and identifying the decisions that remain human. The approach arrives as companies increasingly use AI in hiring, productivity and knowledge work while simultaneously trying to distinguish genuine employee capability from machine-generated output.
Generative AI has changed the economics of producing knowledge work. A report, presentation, analysis or draft that once required hours can now begin with a prompt and appear in seconds.
That speed creates a less obvious problem for employers: who is accountable when the machine-generated work is wrong?
Rich Carr’s new book, Stake Skills: Owning the Work AI Does for You, approaches that problem from the workforce-skills side rather than from the perspective of AI detection software. Its premise is that organizations need a way to establish what an employee asked an AI system to do, what the machine produced, what the employee verified or rejected, and which decisions ultimately belonged to the human.
Carr describes this capability as a third category of professional skill. Hard skills demonstrate technical capability, while soft skills cover interpersonal effectiveness. “Stake Skills,” by contrast, are intended to establish accountability for AI-assisted work.
The distinction could become increasingly relevant as employers rethink how they assess talent.
Gartner predicts that 75% of hiring processes will include certifications and testing for workplace AI proficiency by 2027. Gartner also expects organizations to increasingly assess candidates without AI assistance as concerns about critical-thinking and independent problem-solving grow.
That points toward a more complicated definition of AI literacy. Knowing how to prompt a large language model is one capability. Knowing when its answer is unreliable, checking the underlying evidence and deciding whether the output should be used is another.
Carr’s framework centers on a relatively simple mechanism: a recurring record of AI-assisted work. His proposed weekly log captures four elements — what was produced, what AI generated, what the worker verified or discarded, and which decisions were made by the human.
The concept is particularly relevant to HR and learning teams because it moves AI skills away from abstract training and toward observable workplace behavior.
The broader workforce data suggests there is a genuine need for that distinction. A global KPMG and University of Melbourne study of more than 48,000 people across 47 countries found that 57% of employees said they hide their AI use and present AI-generated work as their own. The same research found that 66% of employees rely on AI output without evaluating its accuracy, while 56% reported making mistakes at work because of AI.
For HR leaders, those findings create a difficult balance. Organizations want employees to adopt AI because it can improve productivity, research and content creation. At the same time, undisclosed or poorly supervised AI use can introduce factual errors, compliance problems, intellectual-property concerns and sensitive-data exposure.
That makes documentation potentially useful as a complement to AI-use policies.
The approach also differs from AI-content detection. Detection tools attempt to determine whether text or other material was generated by AI. Carr’s model instead focuses on the worker’s process and accountability. In an enterprise environment, that could mean asking not only “Was AI used?” but also “What did the employee do with the output?”
That distinction matters because AI-generated work is increasingly difficult to separate cleanly from human work. Employees may use AI for brainstorming, research, restructuring, coding, editing or drafting while substantially changing the resulting material themselves.
Carr has applied the methodology to the book itself. According to the project’s publicly available production record, the book was drafted with AI and then subjected to human verification and decision-making. The author has published a running record showing what the machine produced, what he checked, what he changed and what he ultimately accepted.
That self-documentation is arguably the most interesting part of the project. Rather than treating AI involvement as something to conceal, the framework makes the production process part of the artifact.
For enterprise HR teams, the bigger question is whether ideas like this evolve into formal workforce practices.
Organizations already use learning-management systems, skills taxonomies, competency frameworks and performance-management platforms to document employee capabilities. AI could require those systems to capture a new layer: how employees exercise judgment when machines participate in the work.
That could eventually influence recruitment assessments, professional development and performance reviews. An employee who can effectively use Microsoft Copilot, Google Gemini, Salesforce AI or other enterprise AI tools may still need to demonstrate domain knowledge, verification skills and independent judgment.
Gartner’s latest workforce research reinforces that challenge. In July 2026, the firm reported that 95% of organizations had implemented AI in some capacity during the previous year, while only one in five had achieved significant or transformational value. Gartner also found that 22% of CHROs said at least one business leader had stopped hiring for entry-level roles because of AI automation.
Those numbers suggest that AI adoption is becoming less about simply deploying tools and more about redesigning how organizations develop and evaluate people.
Stake Skills does not provide an enterprise HR platform or automated assessment system. Instead, it offers a behavioral framework for a workforce increasingly operating alongside generative AI.
Whether “stake skills” becomes an established HR competency remains to be seen. But the underlying question is unlikely to disappear: as machines take on more of the production work, organizations will need better ways to identify the human judgment that remains responsible for the result.
Market Landscape
The emerging AI workforce market is shifting from AI literacy toward AI readiness and accountability.
Traditional corporate AI training has largely focused on tool familiarity, prompting and productivity. The next stage is likely to place greater emphasis on verification, responsible use, domain expertise and independent judgment.
Gartner’s research reflects that transition. Its hiring forecasts point toward formal AI proficiency testing, while its workforce research warns that rapidly changing skill requirements can create confusion rather than readiness.
For HR technology vendors, this creates opportunities around AI skills assessments, learning platforms, workforce analytics, competency management and performance systems. Companies such as Microsoft, Salesforce, Workday and other enterprise-software providers are increasingly positioning AI inside existing employee workflows, making the question of human accountability relevant across the HR technology stack.
The opportunity is not necessarily to track every prompt an employee enters. A more practical model could be lightweight documentation around consequential work — particularly decisions involving customers, finances, compliance, hiring or regulated information.
In that context, Carr’s weekly log represents a deliberately simple counterpoint to increasingly complex AI governance systems: establish a record, verify the work and make the human decision visible.
Top Insights
- Rich Carr’s Stake Skills proposes AI accountability as a distinct workforce capability built around verification, documentation and human decision-making.
- Gartner expects AI proficiency testing to enter 75% of hiring processes by 2027, increasing pressure on candidates to demonstrate practical AI readiness.
- KPMG and University of Melbourne research found 57% of employees hide AI use, highlighting a growing gap between adoption and workplace transparency.
- The framework shifts attention from detecting AI-generated content toward documenting how employees evaluate, modify and take responsibility for AI-assisted work.
- HR teams may increasingly need systems that measure not only AI usage but also human judgment, verification skills and independent capability.
Join thousands of HR leaders who rely on HRTechEdge for the latest in workforce technology, AI-driven HR solutions, and strategic insights
