Curiosity Killed The Cat, But Could AI Kill Curiosity? (Video)

Welcome to The Hornet. I'm Rob Hornby, Co-CEO of global consulting firm AlixPartners, and this is a collection of my perspectives on AI. Episode 3. Curiosity killed the cat, but could AI kill curiosity? Curiosity is famous for putting cats at risk. but it also gave us art, science, the humanities, medicine, and the eventual dominance of our species.

I'm a well-known felinophile, but this is a fair trade. I also consider myself curious. I'm always engaged in audio or physical books, online courses, writing projects, and a never-ending doctorate, alongside a very full-time job. The out-of-work aspects are crammed into weekends, flights, jet lag hours, and morning routines. Along the way, I've encountered some notoriously difficult texts.

Most recently the writings of Hans-George Gadamer. Sometimes it's taken weeks for the ideas, structures and language to become clear to me, but when I finally see something for the first time, all the effort seems worth it. This is deep curiosity at work. However, when trying to understand the incredibly challenging French non-philosopher François Laruel,

I asked ChatGPT to explain his thinking in simpler terms and immediately regretted it. The dumbed-down summary made me feel like a cheat and I knew I'd forfeited any delayed gratification from grappling with his work for myself. Nonetheless, using AI for what's known as cognitive offload has some positive features.

Microsoft reported that 68% of participants in a large-scale experiment completed writing and planning tasks with AI assistants in less than half the usual time. Another study found that cognitive offload significantly reduced mental effort. This sounds like the basis for the productivity gains that everyone is searching for.

However, the benefits come from altering the way our brains work in complex tasks. MIT labs found that AI cognitive assistance reduced psychological engagement in the activity and compromised memory recall afterwards. In the Microsoft study already mentioned, 62% of participants stopped searching elsewhere once AI had provided an answer, even when they were not fully convinced that it was the right one. Based on these findings, curiosity appears to be fundamentally compromised by AI's support. However, this is only half true. It turns out that humans engage in two types of curiosity, deep and shallow or diversive. Both types are normal in humans, but whereas deep curiosity depends on attentional stability and working memory to engage in sustained cognitive tasks, shallow curiosity is characterized by continuous shifts in focus. AI is causing us to over-index on shallow approaches and we are adapting to dopamine hits from small bursts of curiosity rather than the delayed gratification associated with deeper thinking.

There is also growing evidence that AI, along with other digital platforms such as social media, is eroding our capacity for critical thinking and making us less tolerant of ambiguity. These issues are very worrying to those of us who resist the idea of human replacement in favour of AI augmentation.

These studies clearly show that pairing people with AI is not guaranteed to enhance human performance, even if it improves outcomes. We need AI that's actually good for our brains. Happily, there is some evidence that the right use of AI can bring out the best in us. Framing generative AI as a co-thinker rather than an answer generator supports deeper curiosity if users are encouraged to ask better and more layered questions. In that vein, a 2024 study found that when students used AI to extend their questioning rather than provide an answer, epistemic curiosity increased significantly. As business leaders, I propose there are four things we should do. Insist that augmented users routinely leave the AI ecosystem.

This should involve following up on primary sources cited by AI, triangulating automated outputs with independent data, and consulting with other humans about the task. Design AI augmentation with human cognitive function as a priority. It's not enough to give staff access to AI and just leave them to figure out what to use it for.

Skilled human-centric design should ensure that deep and shallow curiosity are balanced. Ensure some things are done the hard way on purpose. Occasionally, efficiency can be sacrificed in favour of deep learning disciplines and delayed gratification. Value deep curiosity. Sustained cognitive engagement is beneficial to business and should be rewarded. Initiatives requiring this skill should be compensated at a premium.

So what can we conclude? There is strong evidence to suggest that AI enables humans to deliver outcomes more quickly, effectively and efficiently than they can unaided. However, there's a cognitive pitfall to avoid that can lead humans into a long-term decline in critical thinking by using the technology in the wrong way. As such, augmentation should be measured not only by outcomes, but also by the health of the human brain. Deep curiosity has taken humanity a long way, and we should steward it safely for future generations during the AI transition. Unlike our feline friends, we only have one life, not nine, in which to get this right.

To view the full article please click here.

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.