Management is the ultimate key to AI productivity.
Professor Ethan Mollick argues that as AI evolves into autonomous agents, the core principles of management—delegation, goal-setting, and evaluation—become the most critical skills for professionals. By treating AI as a “talented but literal” subordinate, experts can leverage their subject matter knowledge to scale high-quality output at unprecedented speeds. Strategic success in this new era depends less on technical prompting and more on the ability to define what “good” looks like and managing the feedback loop effectively.
Points clés
- Professor Ethan Mollick conducted a Wharton executive MBA class where students built startup prototypes in just four days using AI.
- Tools like Claude Code, Google Antigravity, and ChatGPT allowed non-coders to create working features, shaving months off the traditional development cycle.
- The “Jagged Frontier” of AI ability remains a challenge, as users cannot always predict where AI will succeed or fail on complex tasks.
- Strategic delegation is governed by three variables: Human Baseline Time, Probability of Success, and AI Process Time.
- OpenAI’s GDPval paper shows that GPT-5.2 Thinking and Pro models tie or beat human experts 72% of the time on professional tasks.
- Experts save an average of 3 hours on a 7-hour task when utilizing a “draft-review-retry” workflow with high-performing AI models.
- Effective AI management mirrors traditional frameworks like Product Requirements Documents, shot lists, or the Marines’ Five Paragraph Orders.
- Subject matter expertise is required to evaluate AI outputs and recognize when a financial model or medical report contains subtle errors.
- Mollick posits that management skills, once labeled as “soft,” are now the “hard” skills necessary to manage an army of tireless AI agents.
- The scarcity in the AI era is no longer talent or effort, but the clarity of vision to know exactly what to ask for.
À retenir
So, it turns out those years you spent in boring management seminars weren’t a total waste of time after all—they were just accidentally training you to be the CEO of a digital legion. If you can actually explain what you want without changing your mind every five minutes, you might just survive the AI revolution. Otherwise, you’ll just be the person spending ten hours “saving time” by arguing with a chatbot that doesn’t care about your feelings or your deadlines. Good luck with that.
Sources
