I saw a post recently that I haven’t been able to get it out of my head. I apologize in advance to the original poster because I can’t remember who it was (or even where I saw it) but they basically said, if you are using AI for brainstorming, you are starting from ideas that already exist.

This made me chuckle. It also made me a’ha. It’s so true even though it’s a use-case frequently cited by people for LLMs in their day-to-day.

AI is built from information that exists, so if you are using it as your starting point for idea generation (whether that be for social content or product development), you are, absolutely, tapping into ideas that are already out there.

I find this insight interesting given how common the problem of the blank page is and how easy it is to lean on Claude or Chat GPT to get some spaghetti on the wall. It’s a jumping off point, right? This approach can even be lauded as an ethical use of the technology because, well, it’s not being used for the real creative work, right?

But is it?

If your starting point is from a list of ideas already predicted by an algorithm, how original can the thinking be?

I’ve honestly been guilty of this myself. I struggle alongside a lot of people with how to ethically use the technology in ways that are additive to productivity while also not sucking the humanity out of the work to ensure my environmental footprint is cautious and careful. The blank page seems like an easy lift. But is it a limiter? Has it been? I feel like the only answer to that question is, YES!

My mind has been chewing on this thought. It also unlocked memories of facilitated brainstorming sessions that we had at the beauty brand where I worked for over a decade ago. The team would be taken offsite. We were broken into teams comprised of people we didn’t work with on a daily basis. We often had toys and markers at our tables to play and doodle with during the sessions and we were led through a variety of brainstorm techniques to push into idea territory outside the everyday.

I remember specifically one of the exercises called Brain Writing. Brain Writing is a group method where six participants each write three ideas in five minutes on a sheet of paper, then pass the piece of paper to the person next to them. That person then writes 3 more ideas in the next 5-minute time slot and it keeps going from there until all six participants have added their ideas to each sheet. It’s called the 6-3-5 method which ultimately yields 108 ideas in 30 minutes.

This exercise can feel very difficult in the moment after rounds 1 and 2 when the easy ideas are already listed, forcing you to dig deeper for more. But the cool thing about this exercise is that the best ideas were often the ones that came later. Truly creative ideas emerge when the obvious ones are exhausted. Everyone reaches first for what is low-effort because that is what is already out there in some shape or form—which is also basically what AI does, right?

There is an old quote from Linus Pauling in a 1960s article in Fortune Magazine where he said, “The best way to have a good idea is to have a lot of ideas.” And, to me, that’s where the human element comes in.

It can be uncomfortable. No doubt it is hard. Sitting in front of a blank canvas leaves you alone with your brain. But if you set that timer and just keep digging, it’s amazing what can come out. I can guarantee it’ll be more innovative than what all these bots can produce.

In addition to that exercise, another one that stands out for me is a silly exercise the facilitator would use before jumping into a brainstorming session called Zip/Zap/Zop. This exercise is a ball game where the group stands in a circle. The first person with the ball makes eye contact with a different person, says “Zip” and throws the ball to that person to catch. When (and if) the person catches the ball, they say “Zap” and then repeats the process by finding someone else to make eye contact with, say “Zop” and throw the ball. This continues for numerous rounds: Zip, Zap, Zop and repeat. Inevitably, people miss the ball or say the wrong Z word. No biggie. No harm done.

The facilitators framed this exercise as helping us shift from our reptile-brain level of thinking (i.e. survival mode) to a higher level of critical thinking. However, when I researched the exercise, that science of what is happening in the brain is actually slightly different. What a game like Zip/Zap/Zop does is create psychological safety within the group through eye-contact and experiences of failure. That way, when shifting into brainstorming, the session becomes a safe place to throw out ideas that might be considered bad or dumb. The fear of judgement from the group is minimized.

I would often roll out eyes at these games, but I have to admit they worked. They created comradery and the result was productive. The group was in synchronicity, felt safe and were able to generate way more ideas in that environment. AI brainstorming just cannot compare to that. AI outputs are, at the heart of it, reductive. But a group of people, coalesced with a goal and bonded by these types of activities, is expansive. Ideas are literally limitless.

In addition to that, this topic also has me considering the power of human creativity and the brain. Beyond the brainstorming (or maybe as a part of it), most of the idea-generating we do is focused on solving a problem. The output may be innovative, but it also serves a purpose. There is a need that has to be fulfilled. And one of the most beautiful aspects of the brain and how it works is the ability for it to keep working on something even after a person has physically transitioned to another task.

Who hasn’t walked away from a problem, gone for a walk or slept on it only to find, hours and activity later, a solution or new approach appear?

That a-ha moment or “Big Magic,” as author Elizabeth Gilbert calls it, feels like divine inspiration. In reality, it’s just an output of how the brain operates. Scientifically there is a Default Mode Network (DMN) that floods our brains with options, then the Salience Network determines relevance further shaping the idea. After that, the Executive Control Network (ECN) finishes the process by testing and refining.

Said another way, your brain works on a problem obsessively and gets stuck so you take a break, go for a walk or wash the dishes. Then, all of a sudden, there it is… the answer. Eureka!

What happened here, in actuality, is that the effort of the initial work session loaded the brain with the problem into its working memory. The brain begins chewing on it. Then, when the concentrated attention moves away from the problem and onto another task, an unconscious shift takes place in the DMN with the brain running a more expansive query than the concentrated attention could manage. (**If you want to read more on how this works, see this great Neurosity article. This is where I lifted my knowledge of brain mechanics.)

Back in my 30s and 40s, I did a lot of endurance running. I finished six marathons and spent many of hours pounding the pavement while training for those races. During this time, I was also the most prolific in my blogging. That expanse of time spent on the trail allowed me to disconnect and gave my brain time to chew on topics and experiences which ultimately became the output of my writing.

This is the “Big Magic” that Gilbert writes about. And, again, something I would wager that AI cannot do in the beautiful way that we humans can.

Now, on where ideas germinate from? That’s a debate and I’m not qualified or sourced enough to suss that out here. One theory is that all human thinking is ultimately combinatorial—which, essentially, is all that LLMs do. It’s the other side of the coin from the theory of divine inspiration, or Gilbert’s “Big Magic,” or the concept of genius that dates back to the Romans and Greeks who believed a guardian spirit was the source of inspiration. You could go deep in the weeds here, but the quandary I’m trying to pontificate within this post is the question: are human ideas better than robot ones?

And my intuition tells me, Yes! For all of the reasons stated above and then some. I believe our ideas are better because there is a human-element contained in the material output.

Obviously, that POV is up for debate. AI does really good B+ work. But who wants to settle for that? It’s reductive.

I discovered what I consider a hilarious study conducted by Wharton in 2025 published in Nature Human Behavior titled, “ChatGPT decreases idea diversity in brainstorming.”

In the task, participants were given two items that were unrelated to each other (a brick and a fan) and asked to make a toy from them. In the group that used ChatGPT, 94% of the ideas developed shared overlapping concepts. And 9 people called the toy the same thing—they used the exact same name: Build-a-Breeze Castle. In the group with the human-generated ideas, with no AI assistance, all the ideas were unique with no overlaps.

The team ran this experiment several times and the lack of group-level diversity was statistically significant. Furthermore, users using ChatGPT felt a lower personal responsibility for their own ideas. The narrower pool of ideas produced a weaker sense of the idea being their own.

So when considering this (and the post that prompted this essay), my question remains, “Why would anyone want to limit their ideas to what AI has already been trained on?”

I’m going to throw one more study at you. Recently, Stanford ran a large-scale, blind study focused on comparing AI-generated ideas to human-expert-generated research ideas.

And spoiler alert: the AI-generated ideas scored higher than the human-expert ones. These results were significant and consistent across three separate tests.

But the twist? There was a follow-up study on the ideas that were actually executed—both the AI and human-expert ones. Once executed, the scores for the LLM-generated ideas dropped significantly against the human-written ideas across every metric: novelty, excitement and effectiveness. The human ideas closed the gap in execution and some of them ended up scoring higher than AI ones.

Don’t get me wrong. AI can be very good at producing what sounds novel. It can quickly put surprising combinations and unexpected concepts on the page before you even take a sip of coffee. But ultimately, will those ideas hold up? Will they resonate? Can they compete with the spark of the human mind?

What I’ve determined through this exercise is that the common elements of originality are a combination of diversity of ideas and unique perspectives. That’s what makes our things and ideas truly human. It’s the type of work that, I still believe, cannot be outsourced.

I’m not here to tell you not to use AI. I use it myself for a variety of tasks including some of my own editing and proofing (and I know there is a whole cohort of people who argue against that).

We have to determine what our own boundaries are navigating this technology in terms of how we will use it, what our ethics are, what are the absolutely-nots. Those are personal determinations.

But what I do know is that our brains are special. Our thoughts can be magical. Limiting oneself to the ideas generated by an LLM as a starting point is not the most successful set-up.

As I said at the beginning of this essay, I’m as guilty as anyone of using Claude to fill the blank page because it’s easy. It’s low effort. But it’s also borrowed content.

This is, if anything, a challenge to myself and a good reminder for the next time that I open a new file. Instead of jumping to my virtual assistant for help, a better option may be to do a quick exercise first to settle my thoughts and get into the right frame of mind. This could be a breathing exercise, a quick meditation, 20 jumping jacks or just talking the problem out loud. Once the pump is primed, I then proceed to brainstorm and put thoughts to page: the good, the bad, the ugly. Silence the inner critic. Spend time in my head and brain and work the problem.

After that initial session, I can remind myself it’s okay to move on. I can clean out my email, take a walk, do something different and let the project cook in the background.

I’m pretty certain that when I return, I will have something very build-able developed from brain sparks and a-has that have transformational power.

AI often feels like taking a short-cut. But the time invested on refining ideas from a robot also should be considered. That is also time spent—with way less ownership over the end product. Birthing our own ideas sprinkles more of who we are out into the universe with our unique fingerprint and, hopefully, a little proof that we were here.

# # #

Photo by Mark Fletcher-Brown on Unsplash