Working (& Learning) with Machines
How AI Team Mates Are Transforming Knowledge Work—and What It Likely Means for L&D
This week saw the publication of a groundbreaking study, "The Cybernetic Teammate: A Field Experiment on Generative AI Reshaping Teamwork and Expertise." The study explores how artificial intelligence is transforming workplace performance, expertise sharing, and social engagement.
By demonstrating that individual knowledge workers + AI can match the output of conventional teams, break down silos, boost emotional engagement, enhance wellbeing and empower non-experts to perform at the level of seasoned professionals, this and research like it signals profound implications for L&D.
As AI evolves from a tool used to speed up low-value tasks into a more dynamic “cybernetic teammate” with the potential to transform an individual’s performance, L&D leaders are compelled to reimagine the very foundations of their profession — from how we define expertise and how we structure progression frameworks, to how we define the very concept of “learning and development”.
This week’s blog post unpacks these insights and provides a suggested roadmap for building the next generation of AI-ready learning environments.
Let’s go!
Research Summary for L&D Teams
When reading the paper, four key things stood out for me:
1. AI as a Performance Enhancer
The study demonstrated that individuals using GPT-4 outperformed solo workers and matched the performance of two-person teams. Specifically, individuals with AI showed a 0.37 standard deviation improvement in solution quality, compared to a 0.24 SD improvement for traditional teams without AI.
AI also improved efficiency: individuals using AI spent 16% less time on the task than those without AI, and AI teams saved 13%. Yet, they produced longer, more detailed solutions, indicating deeper and broader exploration and analysis.
Implication for L&D: AI fluency is a productivity driver. Basic familiarity with AI isn’t enough—L&D must intentionally develop skills to enable employees on both the individual and team level to work as productively as possible with AI across the end to end work flow.
2. Breaking Down Functional Silos
The study also showed that AI fundamentally shifted how participants generated ideas across their expertise areas. Without AI, Commercial and R&D workers overwhelmingly stuck to their disciplinary boundaries—Commercials offered business-centric ideas; R&Ds offered technical ones etc. With AI, these distinctions disappeared: all participants using AI generated balanced solutions across domains, regardless of background.
This might suggest that AI has the potential to replicate the integrative and business benefits of cross-functional collaboration, by surfacing diverse perspectives, enabling divergent thinking and in the process filling knowledge gaps.
Implication for L&D: The long-imagined concept of learning in the flow of work might, thanks to AI, become a reality. L&D teams must being to imagine a world where, instead of stopping work to learn, employees are constantly learning by doing — using AI to explore new ideas, solve problems they haven’t faced before, and level-up their capabilities as they work.
In this new model, “training” isn’t something you pause work to do — learning is the work.
3. Emotional Engagement & AI
Contrary to expectations, working with AI led to significantly more positive and fewer negative emotional responses than working solo. Specifically:
Individuals using AI saw a measurable increase in positive emotions (e.g., excitement, energy) related to their work.
Teams working with with AI reported an even higher boost in positive emotions.
At the same time, negative emotions like anxiety and frustration decreased by across all AI conditions.
Importantly, the positive emotional states generated by AI rivalled those of human-human collaboration, suggesting that AI might substitute some of the motivational and emotional benefits of teamwork.
Implication for L&D: AI Companions which deliver not just “always on” performance management but also motivational support, encouragement and coaching are a viable and likely future.
4. Democratising Expertise
The study split participants into “core-job” (experts in product development) and “non-core-job” (non-experts). Without AI, non-experts underperformed significantly. But with AI, non-experts working alone performed as well as experts working in teams without AI.
AI also flattened dominance dynamics in teams. Without AI, team members’ backgrounds led to skewed outputs. With AI, teams produced more balanced solutions, suggesting AI helped integrate voices more equitably.
This shows AI’s ability to close the expertise gap, enabling those without domain depth to contribute meaningfully.
Implication for L&D: In the AI era, we are required to reimagine who we train, how we train them and when. Those L&D teams which embrace AI in the flow of work will focus less on developing deep specialisation and more on developing a generic “signature” set of skills and knowledge necessary to optimise AI-collaboration. This will enable a broader set of employees to contribute meaningfully across a broader range of disciplines and tasks.
Five Power Moves for L&D Teams
So what might all of this mean for L&D teams in knowledge-based industries in the near term?
Here are five power moves that early innovators will likely make to get prepared for the potential rise of the AI team mate:
1. Rethink Skills Taxonomies
Update competency frameworks to reflect the skills needed for effective human-AI collaboration. This will likely include things like:
Prompt engineering (framing questions, refining inputs, validating outputs)
AI output evaluation (critical thinking, bias detection, relevance checking)
Judgment in AI-augmented decision-making (knowing when to trust AI vs human insight)
Collaboration protocols for AI-augmented teams (division of labor between human and machine)
The ability to work with AI isn’t just another technical skill set — it’s likely to become the foundational capability for modern knowledge work.
2. Rethink Training
Move beyond static courses and toward dynamic, embedded learning. In order to leverage the full potential of the AI co-worker, L&D will likely evolve to enable:
Just-in-time AI-enabled learning within the flow of work
Scaffolded skill-building, where AI supports novices while challenging experts
Reflection and feedback loops, supported by GenAI tools acting as coaches
In this new model, “training” isn’t something you pause work to do — learning is the work.
Instead of stopping work to learn, employees are constantly learning by doing — using AI to explore new ideas, solve problems they haven’t faced before, and level-up their capabilities as they work.
The ability to work with AI isn’t just another technical skill set — it’s likely to become the foundational capability for modern knowledge work.
3. Rethink Hiring
Partner with talent and recruitment teams to reshape hiring criteria:
Prioritise AI fluency, adaptability, learning agility, and collaboration mindset
Introduce AI simulations or prompt-based tasks into recruitment assessments
Redesign onboarding to rapidly up-skill new hires via AI co-pilots, helping them deliver value from day one
In a world where AI reduces the gap between novice and expert, potential becomes more important than pedigree.
4. Rethink Progression
Progression frameworks must evolve beyond linear, tenure-based models:
Create AI-enabled performance ladders, where growth is measured by the ability to deliver outcomes with increasing autonomy and AI leverage
Define new roles like AI-enhanced generalists and domain-specific “super-users”
Introduce AI mentorship models, where experienced employees coach others in effective AI collaboration
Career growth in the AI era swill likely centre around the worker’s ability to harness AI effectively to produce high quality outputs at pace and scale.
5. Rethink Structures
AI challenges the value of traditional departments organised by deep, narrow expertise. L&D must help organisations:
Shift toward cross-functional, AI-augmented teams with shared language and learning pathways
Enable on-demand teaming, where talent is mobilised across projects based on need, not reporting lines
Reduce reliance on static role-based learning, and build modular, problem-based development experiences
As AI breaks down functional silos, L&D must break down learning silos too — equipping employees to move fluidly across contexts and collaborate through shared AI skills, fluency and tooling.
Conclusion: Does the Future of L&D Start Now?
The findings from The Cybernetic Teammate study are not just interesting — they’re likely to be catalytic. AI is no longer a passive tool at the edges of work. It is emerging as an active partner in how we think, learn, and perform. For L&D, this means rewriting the playbook.
If we pick up the potential of AI and run with it, we have the opportunity to enter an era where learning is no longer confined to classrooms, credentials, or content libraries. It’s embedded, adaptive, and collaborative — with AI as both catalyst and co-pilot. From levelling up non-experts to supporting expert performance, from emotional support to structural transformation, AI is redefining what’s possible for human development at work.
But seizing this opportunity requires bold moves which are typically not seen in large corporates in general or in L&D teams specifically. As ever, the impact that AI will have on L&D will depend ultimately on the vision & actions of the humans around the table.
Happy innovating!
Phil 👋
PS: If you want to dive deep into the cutting edge of AI-powered L&D, apply for a place on my bootcamp.