The Danger of AI Overuse

Bill Gates once said, if you want a problem solved, give it to the lazy person. What he meant was lazy people like to cut corners and do things the most efficient way possible, because they don’t like to put in effort. This laziness isn’t always bad. It mirrors what calculators did for math or GPS did for navigation: we offloaded the tedious parts. Why spend a longer time doing something when an easier way has already been invented? But the risk is when offloading becomes over-reliance. If we stop exercising the muscles of critical thinking, creativity, or memory, those muscles weaken.

It’s easy to go with the path of least resistance. The danger is when we are outsourcing so much of work to AI where we outsource our thinking to it too.

This is where we’re getting AI slop, AI slop everywhere. We see it on Linkedin. We see it on Instagram. We see it on Reddit. Did you know that 50% of Linkedin content today is AI generated?  

For the AI providers, AI laziness is also profitable. The less effort people put in, the more they depend on the tools, and the more entrenched the technology becomes in everyday workflows, but this is the beginning of a dangerous crisis that affect a whole generation of knowledge workers.

There have been multiple studies released recently that illustrate how incessant use of AI causes cognitive decline. One of the most compelling recent findings comes from MIT, where researchers conducted a four-month study using EEGs on 54 participants. They were divided into three groups: one used ChatGPT for essay writing, another used Google, and the last relied solely on their own effort. The ChatGPT group showed notably weaker brain engagement, poorer memory retention, and less sense of ownership over their essays. Over time, they became more passive, often defaulting to copy-paste prompts. The brain-only group, by contrast, displayed stronger neural connectivity, greater engagement, and higher satisfaction with their work. Even after switching tasks (from AI-assisted to solo writing), former AI users continued to show reduced brain activity.

These results point to long-term cognitive costs, including what researchers describe as “skill atrophy,” lower creativity, reduced critical inquiry, and greater susceptibility to manipulation.

A study published in Sociocybernetics examined 666 participants across various ages and education levels. It found that frequent AI tool use was significantly associated with lower critical thinking skills, largely mediated by cognitive offloading (i.e., relying on AI to do thinking for us). Younger participants, in particular, had higher dependence on AI and lower critical thinking scores.

Some researchers raise the specter of cognitive atrophy and reduced brain plasticity from heavy AI reliance. A neuroscience-oriented column flagged the risk of diminished “cognitive load” and long-term erosion of thought patterns. More anecdotal and opinion pieces, like those in The Wall Street Journal and Vox, argue that while AI can boost productivity, overdependence may undermine creativity, memory, and attentional capacities.

Find the Balance

This isn’t an article telling you to stop using AI altogether. Instead, it’s an attempt to figure out what should you outsource and what should you keep? How do we use AI to make ourselves more powerful? How can we guard from cognitive atrophy, use AI to our advantage but at the same time, make our brains smarter? We have to draw a line. But how do we draw this line? That’s what I plan on investigating.

In my keynotes, I’ve often prompted my audience to take their one role and divide it up into a series of tasks. There are tasks that are ripe to outsource to an AI system.

There are also tasks that we know is hard for an AI to take on. Those often deal with strategy, human dynamics, and planning. These are tasks that we know we need to get better at because AI will never be able to replace that part of our work.

“Spend human hours on human-necessary things”

 

So what should you use AI for and what should you keep human?

It’s a good idea to learn two terms that I teach in my AI workshops: Narrow and General AI. Narrow intelligence belongs to an AI system that is dedicated to knowing or doing a specific sets of tasks. General intelligence is much closer to actual human intelligence. We are much better oriented in doing a huge variety of tasks. AI researchers have also found it difficult to achieve AGI, so much so, that in recent months, the CEO of OpenAI has stopped wanted to measure the level of AI advancement with that term. Something like a GPT 5 can do a variety of things, but still it is unable to completely replace a human.

Use AI For:

Narrow, repetitive, data-heavy tasks that don't require novel thinking:

·       Brainstorming and first drafts

·       Summaries, notes, and highlights

·       Copy variations, titles, prompts

·       Research scaffolds and source lists

·       Data cleanup, analysis, formulas, scripts

·       Translations and tone rewrites

Rule of thumb: If you could demo it to a junior employee on Loom and they could replicate it, automate it.

Keep Human:

Novel problems, human dynamics, and strategic decisions that require judgment:

·       Anything involving empathy, negotiation, hiring, feedback

·       Strategy, pricing, policy, tradeoffs

·       Legal, compliance, safety approvals

·       Final signoff on public or high-stakes work

·       Original thinking that breaks from patterns

Why? AI finds patterns in existing solutions. Novel problems have no pattern to match. That's where your human brain excels.

Finding the balance

Do you know that Serenity prayer? “Grant me the serenity to accept the things I cannot change, The courage to change the things I can, And the wisdom to know the difference”. In the age of AI, I think it’s “Let AI take what’s rote, let me keep what’s human, and give me the wisdom to know the difference.”

The Workflow That Protects Your Thinking

To maintain cognitive strength while using AI, don't hand over the thinking. Use this sequence:

1.     You define the goal, audience, constraints, and non-negotiables

2.     You outline your approach and rough solution

3.     AI produces drafts or options

4.     You critique, add context, mark what's wrong

5.     AI revises with your feedback

6.     You finalize, fact-check, and own the decision

This keeps your brain active in the critical thinking phase while letting AI handle execution.

Your Competitive Edge in the Age of Sameness

Here's what worries me: your competitors have the same AI tools you do. The same access to information, the same frameworks, the same GPT-4.

Pure knowledge is no longer an advantage.

Your edge is:

·       Discernment - separating signal from noise

·       Context - applying years of tacit knowledge AI can't access

·       Originality - combining ideas in ways patterns can't predict

·       Judgment - making calls in ambiguous, high-stakes situations

The Question You Should Ask Yourself

Is AI making you faster at being mediocre, or is it amplifying your unique capabilities?

If Shakespeare had ChatGPT, would he have written more predictable plays, or used it to prototype faster and push his imagination further?

The technology isn't the bottleneck. Your imagination is.

AI will replace the version of you that does the same thing repeatedly. It will amplify the version of you that thinks strategically, creates boldly, and exercises judgment.

Use AI to save your brain for what only your brain can do.

How do you discern what to outsource and what to keep? I would love to hear about your model!

--

Hello! I’m Sharon Gai, a keynote speaker on AI and its effects on workers and society.