Can AI enhance our critical thinking skills?

I was talking to a friend who’s a teacher over the weekend. She mentioned that her school has moved from written exams to conducting interviews, as an effort to prevent ‘cheating’ using generative AI. This made me feel sad because learning to write well = learning to think well, not the other way around. It’s critical that we continue to develop our minds, which means embracing AI, not fighting it. And writing with LLMs rather than not writing at all.

This got me thinking about how generative AI will shift what we value from humans in the workplace, and what we test for in the education system in a future where outcomes are co-created between individuals and technology.?

Those of us old enough to remember panicked headlines about Wikipedia ruining the education system, will also remember that it was obvious when classmates had used Wikipedia to write their essays. (I attended a university with a tuition system of peer critique and live discussion where we had the benefit of comparing answers). But this ‘cheating’ didn’t cheat the system, it cheated students out of good grades. Higher education doesn’t value the retention of information, it values original critical thinking using a combination of obvious and found sources so the internet actually enriched original thinking rather than diminished it when used well.

I majored in Art History and when I was writing my thesis, it was the journey of discovery that yielded the most interesting original thought, sifting through source material. Reading modern critical thinking and applying it to dusty source material, taken down from the top shelves of libraries and sometimes untouched for a decade (you could see when books were last taken out in our system). When researching the first woodblock prints, I noticed that Art Historians kept referring to the craftspeople as illiterate. This conclusion was drawn because the illustrations often didn’t match the text.?

I visited a chapel later in the week, sat in a pew admiring the stained glass windows, when I had an epiphany - no, not a religious one - an insight for my thesis. It dawned on me that these people’s main experience of imagery would be inside the walls of a church. And during those experiences they would hear a sermon that wouldn’t be the exact same illustrated story in a stained glass window. I looked into this further and discovered that people at the time aspired to divine experiences, achieved by becoming ‘spiritually overwhelmed’ by multiple senses at the time. My thesis was born - that these woodblock cutters were in fact literate, deliberately placing different images next to written stories - trying to scale and replicate spiritual overwhelm.

I want to be clear, this isn’t a Luddite rant.?

The internet made more information, more available, and the process of discovery more accessible. It also brought together communities to discuss and record their lived experience and perspectives openly, which gives an even better living record of how different groups feel. LLMs will let us ingest more widely drawn sources faster than ever before, but the main critical thinking skill they will help us to usefully develop is the formation of excellent questions. An HBR Study of LLMs used in brainstorming showed that a group who brainstormed for at least 30 minutes before using an LLM got to stronger ideas than a group who used the LLM to begin because it led them down more generic paths from the beginning. Whether you’re in education or business, or just trying to be a better writer, deciding what the most compelling, lateral and interesting questions are, will become your differentiating skill to hone.

Questions like the one I asked myself for my thesis that weren’t about the information, but sought new ways of looking at the human experience:

Where in life do the images and text not match?

How could deliberately mis-matched images and text enhance the human experience??

What biases inform the assumption that early woodblock printers were illiterate?

Returning full circle to the conversation with my teacher friend, I wondered whether in the future, we would evaluate the questions asked of LLMs alongside final written essays. In math and science, it’s common to show your workings out. How you approach the formula can be as important as the answer itself.?

What are your tips for feeding LLMs rich and varied source data and asking great questions?