What stood out to me is the idea that AI can remove not only work, but also the gradient through which people used to learn how to hold a problem.
That seems especially important when redesigning work around AI. “What should this person stop doing?” is a useful question, but perhaps it needs a companion: “What will this person no longer learn if we remove it?”
Some tasks are pure repetition and should disappear. Others may be carrying hidden developmental value — not because the output matters, but because the sequence of doing the work teaches judgment.
That feels like an important distinction for deciding what to delegate.
Great article, Sergio. I'm reminded of workflow mapping to identify the human vs. AI skills/capabilities needed in every step of the process to achieve (x). Also reminded of the more macro shift to creating outcomes vs. outputs. Both are spurred on by AI, both require a reverse-engineering of capabilities required to create the outcome. Everything unnecessary stripped away, no vague "development" or "culture." Everything for a purpose.
On a related note, I've pushed the pause button on strategy work because I've realized that it's far more important to nurture the human capability of judgment and decision making, so I'm bringing back my "human intelligence toolbox." I echo your worry: humans love to take the easy road, which means the important, slow work usually gets put on the back burner until the consequences of cognitive offloading to AI show up. At which time it will be too little too late.
Totally aligned. I’ve also refreshed my posts on the website exactly on decision-making and judgement. Interestingly AI tools are making even more evident shortcomings in human capabilities. Even in my individual experiences with Agents of my Orchestrator, what I invest more time into are: retrospective analysis, feedback on agents inputs, decisions on alternatives, judgement on what to stop doing… and it is “a lot” of time!
What stood out to me is the idea that AI can remove not only work, but also the gradient through which people used to learn how to hold a problem.
That seems especially important when redesigning work around AI. “What should this person stop doing?” is a useful question, but perhaps it needs a companion: “What will this person no longer learn if we remove it?”
Some tasks are pure repetition and should disappear. Others may be carrying hidden developmental value — not because the output matters, but because the sequence of doing the work teaches judgment.
That feels like an important distinction for deciding what to delegate.
Totally agree. And requires a real study of how work is done beyond tasks
Great article, Sergio. I'm reminded of workflow mapping to identify the human vs. AI skills/capabilities needed in every step of the process to achieve (x). Also reminded of the more macro shift to creating outcomes vs. outputs. Both are spurred on by AI, both require a reverse-engineering of capabilities required to create the outcome. Everything unnecessary stripped away, no vague "development" or "culture." Everything for a purpose.
On a related note, I've pushed the pause button on strategy work because I've realized that it's far more important to nurture the human capability of judgment and decision making, so I'm bringing back my "human intelligence toolbox." I echo your worry: humans love to take the easy road, which means the important, slow work usually gets put on the back burner until the consequences of cognitive offloading to AI show up. At which time it will be too little too late.
Totally aligned. I’ve also refreshed my posts on the website exactly on decision-making and judgement. Interestingly AI tools are making even more evident shortcomings in human capabilities. Even in my individual experiences with Agents of my Orchestrator, what I invest more time into are: retrospective analysis, feedback on agents inputs, decisions on alternatives, judgement on what to stop doing… and it is “a lot” of time!