How to Pick the Right AI Strategy: Cyborg vs Centaur

A recent study from the Boston Consulting Group (BCG), MIT, Harvard, Wharton, and Warwick Business Schools revealed that consultants employing ChatGPT completed tasks 25% faster and achieved 40% higher quality results compared to their peers who did not use AI.

This has profound implications considering consultants from the Big 3 (BCG, Bain, McKinsey) are widely seen as the pinnacle of knowledge workers, staffed by Ivy League brainiacs and physics PhD’s (I exaggerate, but not much as my former Google colleagues mostly hailed from their ranks).

More interestingly, AI was found to be a skills leveler. When comparing the top 50% of consultants with the bottom 50%, in terms of initial quality performance, those in the lower half saw a massive improvement of 43% after utilizing AI compared to their counterparts at a modest 17%.

These are incredible results. Yet relying too much on AI can also decrease performance due to complacency and producing generic work. This begs the question:

How do we properly use AI?

Today, I’ll provide some answers by expanding two distinct AI strategies coined by Ethan Mollick*, a Wharton professor and one of the study’s authors, into actionable tactics that you can deploy in your own work.

Centaur to Enhance Speed: AI makes you a Manager

When I advise my clients on Artificial Intelligence, I challenge them to either see themselves as AI’s manager (Centaur) or AI’s collaborator (Cyborg).

The Centaur strategy, like the mythical creature of its name sake, clearly delineates where human input begins and machine output ends. This approach views AI as a managerial tool, suitable for tasks requiring high execution but low creativity.

Here, AI is deployed under human guidance to do specific tasks – like writing website copy, synthesizing meeting notes and eveb designing travel itineraries. The key is to give the AI a good amount of contextual information to work with.

Here’s an example of how I would work with ChatGPT to come up with podcast taglines and website copy

.Another one of my personal favorite use cases is turning voice memos into linear essays in my voice using an app like Audiopen. While its output isn’t anywhere close to my standards for publishing, it does solve the “blank page” syndrome that plagues so many writers. My best ideas come from conversations, so having AI capture those thoughts and arrange them into linear narratives is like having a personal scribe following me around.

All these tasks are within well-defined parameters. Most importantly, as the AI’s manager, I know what good output looks like. This is also the most common way of using AI and showcases how it can enhance the efficiency (i.e. execution speed) of knowledge workers, as shown in the above study.

Finally, I should add that there’s a relationship between efficiency and quality in real life scenarios: a knowledge worker who’s freed up from execution can better focus on strategy and quality.

Cyborg to Enhance Quality: AI as an Advisor

Speaking of quality, the Cyborg strategy sees AI as an advisor and collaborator, blending human and machine capabilities. This approach shines in scenarios requiring high creativity and planning, such as developing user personas or launching new products where you lack domain expertise.

Say you wanted to start a new podcast, but have never interviewed anyone. You can begin by asking ChatGPT:

Once you answer these planning questions, you can then ask ChatGPT to create a launch plan:

By providing AI with enough context of your objectives, it can generate creative outputs that push the boundaries of your existing knowledge and capabilities. This is one of the ways that AI can be used as a skills leveler that enhances the creativity and quality of individuals who may initially be lower performers.

Match the Strategy to the Problem

Selecting the right AI strategy hinges on the nature of the task and the project's stage. The Centaur approach is well-suited for tasks with clear goals and a focus on execution, whereas the Cyborg strategy is ideal for exploratory phases where uncertainties are high or creativity is needed in order to make informed decisions.

My rule of thumbnail is to gauge the chaos level of your problem. The higher the chaos, the more Cyborg you should go. The more certain and known the output is, the safer it is to deploy Centaur tactics.

I often show the following diagram in my workshops to illustrate the fluctuation of chaos over the course of a project. Mind maps are great for bringing us through high chaos phases, which goes hand in hand with Cyborg strategies.

For instance, I told ChatGPT to convert the above podcast launch plan into a mind map (OPML format) file and opened it up in MindMeister, allowing me to dive into the details without losing sight of the bigger picture. You know…typical strategic planning stuff perfect for Cyborgs ;)

When it comes to executing on any of these child node tasks (i.e. designing logos, writing podcast descriptions, coming up with interview questions), Centaur strategies will crank out results 10x faster than if you did it manually - but only if you provide enough guidance like a proper manager.

By intentionally aligning our strategies and tactics with the challenge at hand, we can harness AI to produce excellent original work while escaping the trap of over-reliance and mediocrity. The above use cases are unique to me as an entrepreneur and consultant, but the principles apply to just about any field.

How do you apply generative AI to your work and business strategy? Let me know in the comments below! 👇

*I recommend giving Prof. Mollick’s article a read. He and his colleagues introduce a brilliant concept called the “Jagged Frontier”. It's a good mental model to think about evolving AI capabilities.