While some of us are just starting to use AI, there’s another trend emerging on the other side of the AI use spectrum: superusers who have gone off the AI slop cliff.
I credit Thomas Jiralerspong for his cheeky take on the AI Productivity Mountain. His graphic illustrates how those of us striving for optimal AI use seek the mythical Karpathy Peak. That’s a place where you’re getting as much useful work out of AI as possible. It’s named after Andrej Karpathy, who coined the term “vibe coding.”
While we seek the peak, Jiralerspong warns we can quickly fall off the slop cliff, the point where your actual productivity with AI drops off while your perceived productivity keeps climbing. If you aren’t careful, you’ll then descend to a deeper place of psychosis where you start to believe AI is magic and you’re producing masterpieces with the LLM wind at your back.
To stay grounded in reality with AI, it’s helpful to reflect on a recent MIT study. Researchers there looked at AI use among consultants, working to determine the best balance between human and machine. They divided their subjects into three categories:
- Cyborgs engaged in what the researchers call “fused co-creation.” Throughout every stage of the project, they collaborated closely with the AI tool—probing its suggestions, allowing it to lead the way, and taking its advice on some occasions while pushing back against it on others.
- Centaurs practiced a more constrained form of engagement that the researchers call “directed co-creation.” Centaurs knew which questions they wanted answers to and asked more specific questions to get them. Unlike cyborgs, who engaged in conversational back-and-forth, centaurs maintained structured and controlled interactions with AI, harnessing it as a tool for targeted efficiency.
- Self-automators demonstrated what the researchers describe as “abdicated co-creation,” offloading the task almost entirely to the AI. This approach, which delegated analytical and evaluative thinking to the AI, produced quick results that were polished but lacked depth.
Centaurs produced the most accurate recommendations, and on persuasiveness they matched cyborgs, while self-automators lagged on both. How consultants used AI also shaped what they learned. Centaurs leaned on their own expertise and deepened it through targeted questions. Cyborgs gained little subject knowledge but became skilled at solving problems with AI, a valuable skill that MIT professor Katherine Kellogg says leaders should build alongside expertise. Self-automators saw no skill gains and produced weaker work. Let this be a warning that when we allow AI to do it all, especially on work that demands thoughtfulness and domain expertise, we edge closer to the slop cliff.
Here are practical tips to ensure your AI use stays productive:
- Decide your role before you open the tool. Name what’s yours (judgment, relationships, the final call) and what AI can draft. That’s centaur thinking.
- Use the “explain it without the AI” test. Before sharing work, ask whether you can explain where the facts came from and why the conclusion holds.
- Never ask the AI to grade its own work. Instead, verify with a source, a colleague, or your own expertise. AI has a habit of valuing what it produces.
- Draft your thinking first, then bring in AI. Write your own view in three bullets before prompting. This keeps your unique perspective from being replaced by the average of an LLM.
- Do some work the old way on purpose. Regularly complete a task without AI to keep your skills sharp. Accept that good work takes time.
AI amplifies what we bring to it. Bring judgment, experience, and healthy skepticism, and it makes us better. Bring autopilot, and it pushes us over the slop cliff.
Founded in 2004, Lead Star is the company behind the best-selling books SPARK, Leading from the Front, and Bet on You. Lead Star helps professionals reach new levels of success through its innovative leadership development programs.
