Give yourself superpowers. 5 tricks for deep-thinking with AI.
How should you approach AI to work in augmentation mode? To illustrate my point, I show you how to turn a mundane performance review into a valuable self-improvement resource.
A few weeks ago, I posted an article arguing that AI should help us think more deeply, not just finish faster. The piece generated a lot of attention, both online and in meetings, and I was repeatedly asked about practical application: How should you approach AI to work in augmentation mode?
5 tricks for AI-assisted thinking
The trick here is to prompt yourself to pause the “doing” brain and engage your thinking one. Avoid directives like ”write this”, “create this”. Instead, use language and questions that add an element of unpredictability to the conversation. Approach the interaction as a conversation between two parties that can learn from each other.
We all have our own style, but if you need a starting point, try to use at least 2 of these types of questions in each AI-assisted task:
Change perspective - ask for feedback from different personas
Unpack the problem - ask “why” at least once (or even 5 times)
Explore widely - ask for 10 or 20 alternatives or arguments
Playact a conversation - how would you talk about the issue in real life?
Push boundaries - ask the AI to challenge established thinking
Example: AI and performance reviews
A few weeks ago, it was reported that JPMorgan offers team members an AI to help them with their annual performance reviews. The report described the tool as a “shortcut”, which raised concerns about responsibility. Assuming AI can only be used in automation mode, the FT report misses half the story.
Hopefully, at least some JPMorgan staffers will take the augmentation approach, using AI to develop their thinking about their performance before automating the report. How would such an enlightened team member approach the task?
Let’s see what Emma, an imaginary client services manager, asks the AI as she contemplates her annual performance:
“Take my clients’ perspective: how did I do?”
Changing perspective
“From your clients’ perspective, you likely came across as thoughtful and trustworthy. Some clients might say you could be bolder at times, but overall, they’d see you as a reliable and empathetic advisor who protects their interests.”
“Why do clients say I should be bolder?”
Unpacking the problem
“Clients likely say that because they trust your judgment and feel you sometimes prioritise safety over opportunity. To them, “being bolder” means showing confidence in your analysis and helping them act decisively when the data and timing support it.”
“Give me practical suggestions on what I could improve.”
Exploring widely
“You could improve by refining how you calibrate risk tolerance, using more data in client discussions, and practising quicker decision-making. Try diversifying communication styles for different client personalities and reinforcing advice with post-meeting summaries.”
“Act as a client who wants more data. What would you ask?”
Playacting
“I’d say, ‘Emma, I appreciate your explanations, but I’d like to see more evidence behind the recommendations. Could you show me how this portfolio has performed compared to the benchmark over the last three years?’”
“Clients do not get data. How can I make this work?”
Pushing boundaries
“When clients struggle with data, shift the conversation from numbers to stories and outcomes. Use relatable examples, such as how a balanced portfolio behaves in different market moments, or compare it to something tangible like insurance or savings habits.”
Giving your team superpowers
Now, imagine that Emma did not go through this process and started with a “Write my performance review” directive. Perhaps the AI would have enough context, and the job would be done in seconds. Yes, the process would be hyper-efficient, but the output would be a box-ticking exercise and of absolutely no value to Emma or her employer.
Instead, following the above process, Emma has gained a wealth of insights by:
Seeing her work through the eyes of customers and colleagues
Unpacking the reasons behind some of the feedback
Considering a wide range of ideas on how to improve
Trying out a new approach through improvised conversation
Pushing back on assumptions and opening up fresh ideas
In the end, the 30 minutes she spent in AI-assisted thinking have resulted in personal reflection of real value to her and her employer. The follow-up meeting with her manager will likely be more positive and productive. The performance review AI is no longer a time-saving tool she uses once a year, but a valuable resource she can return to whenever she faces a new challenge.
Automation saves a bit of time, but augmentation creates value. Like all technology, AI is only worth the trouble if it gives your team superpowers.
Thanks for writnig this, it clarifies a lot. How do you apply these? Brilliant piece!