Unexpected Psychological Impact: How AI Challenges Human Cognition

The world is captivated by AI. Nearly one billion people now use OpenAI products, a milestone reached in just two years. It's the classic Silicon Valley formula: Build something amazing, make it affordable, get us hooked, and then figure out how to turn a profit.

We're drawn to AI because it offers cognitive shortcuts on an unprecedented scale. However, this path may not lead to a positive outcome for many. We'll start by offloading a few tasks to AI, only to soon discover it's handling everything. In the process, we risk losing our intellectual edge, our jobs, and future opportunities.

But this isn't an inevitable future. Here’s how to recognize the road ahead—and choose a different route.

The approaching turning point

I first tried ChatGPT in March 2023. Today, I use ChatGPT or Claude daily. AI has accelerated my thinking and boosted my productivity, but I've also noticed a growing mental complacency.

I used to meticulously review every draft AI produced. Now, it delivers a solid first attempt about 90% of the time, and I'm losing the drive to double-check its work.

A year ago, I believed the workforce would split into AI users and non-users. I now see that prediction was flawed. Within five years, AI will be ubiquitous. The true division will be between those who actively direct their AI tools and those who passively surrender their reasoning to them.

The cognitive cost of offloading thinking

Humans have long delegated mental labor. Before the written word, bards memorized the entirety of Homer's Iliad. Today, technology serves as an extension of our minds, allowing us to hand over tasks like calculations, navigation, and note-taking.
AI is a different beast. It can tackle nearly any intellectual challenge, and it feels highly efficient. So, we begin outsourcing innocently enough. You ask AI to draft an email. It does a good job and saves you ten minutes. Next, you request a presentation outline. It delivers.

You gradually assign it more complex duties, like formulating strategy. As you come to depend on AI to perform the work, your own skills slowly diminish. 

A study from Microsoft and Carnegie Mellon revealed that generative AI can impair our critical thinking. When knowledge workers trust the AI's output, they are less inclined to engage their own cognitive faculties.

Those of us who trust AI, myself included, rely on our own judgment to verify its results. However, this approach has two major flaws: 1) We overestimate our ability to catch AI's errors, and 2) The urge to skip verification grows stronger over time. 

AI drivers vs. AI passengers

Over the next decade, the knowledge economy will separate into two camps: the AI drivers and the AI passengers.

AI passengers will cheerfully delegate their cognitive responsibilities to AI. They'll paste a prompt into ChatGPT, copy the output, and present it as their own work.

In the short term, they will be praised for their speed. But as AI operates with minimal human input, passengers will be deemed redundant for contributing no unique value to the AI's production. 

AI drivers will insist on remaining in control. They will treat AI's work as a preliminary draft and scrutinize it thoroughly. They will also periodically switch it off, making time for independent thought.

Long term, the economic gap between these groups will widen significantly. AI drivers will capture a disproportionate share of wealth, while passengers become easily replaceable.

Becoming an AI driver

Establish yourself as the boss of AI by following these principles: 

1. Leverage your expertise. Apply AI in domains where you already have knowledge, and maintain a critical stance toward its suggestions.
2. Engage in dialogue, don't demand answers. Instead of asking, "What should we do with our marketing budget?" provide AI with constraints, data, and alternatives. Debate its proposals.
3. Stay hyper-vigilant. Be an active participant. Never assume the output is sufficient. Challenge yourself to ask, "Is this truly a sound recommendation?"
4. Practice active skepticism. Continuously question AI from your own perspective. For example, "Doesn't this overlook the potential risks of the venture?"
5. Resist the urge to outsource every first draft. A blank page can be intimidating, but it's essential for stimulating your own cognitive processes.
6. Make the final decision and take ownership. Let AI inform your medium-to-high stakes decisions, but the ultimate call is yours. Take responsibility for your choices as a human.

Your mind is a terrible thing to waste

AI now provides a thought partner available around the clock, with "expertise" on virtually any subject. 

Yet you stand at a crossroads. You will witness many colleagues opting out of "active thinking" and delegating their decision-making to AI. Many won't even realize their cognitive abilities have weakened until it's too late, and reversing the trend will be difficult.

Don't let this be you. Use AI to test and fortify your reasoning, not to supplant it. 

The critical question is no longer, "Will you use AI?" It is, "What kind of AI user will you choose to be: a driver or a passenger?"

Greg Shove is the CEO of Section.