There are hidden memes inside you – and someone out there is calling for them

How being isolated for months led me to dream up a Nobel laureate's AI twin and imagine how our personal, unusual insights could become viral wisdom.

Two months ago my ability to walk started slipping away. Since then, I’ve been stuck in my home, unable to do the things that I love: host salons, go to concerts and dance clubs, explore the city with my 4-year-old daughter, run through the woods.

My world shrank dramatically.

But this confinement has also delivered an unexpected gift:

The hunch that AI could put our deeply personal experiences to use – letting us help each other in ways we never could before.

How?

We all have hidden memes inside of us – uncommon but valuable ways of thinking and acting that we develop privately, often unaware of their viral potential. With AI they could be decoded and made accessible to people who need them most – no matter where in the world they happen to be.

Here’s one of my hidden memes:

Going up to someone you don’t know at a social event and striking up a conversation is for most people super hard. I’ve realized that there’s a very simple line I can use to overcome this:

“Is it okay if I hang out with you for a bit?”

This might sound trivial, but it has never failed me – and most people do not do this! I’ve even seen this meme take flight. One person I started talking to in this way wrote a very well received LinkedIn post about it, saying he felt he had gotten a helpful gift, something he can now use himself.

This was a simple social hack. But what about memes powerful enough to help people through life-changing challenges? Getting to know the man who couldn’t sit down gave me a clue.

Obsessing over the man who couldn’t sit down

It started with a book. Because one thing I can do in this injured state is to read books.

In “Genius Makers” Cade Metz tells the stories of the outlier weirdos behind the scientific breakthroughs that enabled the current AI revolution.

Nobel laureate Geoffrey Hinton lost his ability to sit down decades ago. As a teenager he had lifted a space heater helping his mother and injured his back. The injury eventually deteriorated to a point where he could no longer drive a car, only ride in the back seat laying down on his back. He couldn’t go on a plane, since he had to sit during take off and landing. As a professor at University of Toronto he would work standing by his desk, and talk to students while laying on it.

At Google he was tasked to go to London to vet the technology of the startup DeepMind, but refused, saying his back wouldn’t allow it. The Google engineer, pilot and skydiver Alan Eustace came up with a solution: they rented a private plane and set up straps to hold him in place, lying down on two folded seats.

Hinton said he started taking the injury seriously when he realized it might make him crippled. And he devised a mental strategy that seems successful for him – but totally counter intuitive to me:

“If you let it completely control your life, it doesn’t give you any problems.”

I would love to have a deep conversation with him, get into his mind and really understand what this means to him. And what I can learn from it.

I imagine millions of more people struggling with pain and impaired mobility would love to do the same.

He can’t talk to all of us. But an AI twin of him could, after being trained on his experiences of managing his condition, his way of thinking, his personality.

And his sense of humor. When asked in an email if he preferred to be called Geoffrey or Geoff, he could reply:

”I prefer Geoffrey.

Thanks,
Geoff”

To me it was mindboggling to realize that even with one of the most important scientists of our time, the immediate way for him to provide value to me in AI form had nothing to do with his groundbreaking research.

It was simply his unusual way of handling a very unfortunate situation he happened to be in.

The strange assignment to create an AI twin of a CEO

Does the idea of an AI twin sound like science fiction?

It’s not.

I started creating my own a year ago as a custom GPT in ChatGPT, and quickly realized that an AI model that has access to my unique experiences can see things I miss. For example that my experiments using AI to role play difficult conversations is something to bring up when doing a talk for a group of healthcare professionals, who often need to deliver tough news to patients.

I'm now in the strange situation of creating an AI twin of a public company's CEO. Our goal is to give his 1000 employees access to his unique perspectives and ways of thinking.

Along with the People Heart Business team, we've interviewed not just his colleagues, but also friends and family, gathering stories of how he acts in the office, at home, in school meetings, as a youth group leader.

What struck me was how his unusual, deeply human approach lights up people's eyes when they share these stories. Just like with Hinton, his uncommon ways of thinking and acting could inspire people far beyond his colleagues.

I noticed it in myself. I didn't just get access to new knowledge – I felt compelled to do things.

But why did these personal stories have such power?

Spending hours on my sofa reading the philosopher David Deutsch's incredible (but also incredibly dense) book "The Beginning of Infinity”, gave me some answers.

Helpful memes are everywhere – but they're trapped

Hinton's and the CEO's unusual actions are what Deutsch would call memes, or at least potential memes.

He argues human culture spreads through memes – behaviors we first observe in others and then replicate, with a bit of creative variation involved. Yes, a meme can be a type of TikTok-video that's trending right now, but also the way a parent acts in a meeting with a teacher (as in the case of the CEO).

I started seeing that many of the most valuable memes remain hidden, locked away in one person's isolated experience. Simply because they are not expressed to the people who would be receptive to them. This is a coordination problem that I believe AI can help us solve.

Having AI twins means that you don't have to be a close colleague of the CEO or Geoffrey Hinton to learn from them. By decoding the unique, unconventional ways a person moves through the world, these memes can be accessible to anyone who needs them.

And this isn't just about eccentric Nobel Laureates and CEOs. Just looking at the people around me, I see so many memes that have made a deep impression on me, and that I would love for even more people to have access to:

• My brother's way of taking his kids seriously and explaining the world to them with a level of nuance most adults don't even use with each other.

• My friend Tove’s incredible openness – she never limits herself to the boundaries of her field, the art world. Instead, she actively seeks collaborations with people whose professions seem far removed from her own: roboticists, philosophers, animal researchers to name a few.

• My colleague Klara's creative ways to care for everyone in the company. Like getting into a cab with another teammate to show up out of the blue at my door with snacks and drinks, for a Friday afternoon celebration to break my isolation. Sitting in front of my screen one minute, 100% expecting them to show up as 2D avatars in a video meeting – and then having them in the room with me the next was wild.

I've had the good fortune of experiencing these memes firsthand, learning directly from each person how to incorporate their wisdom into my life. But what about all the others who could benefit from these insights?

Could an AI twin help spread such memes intentionally? I decided to put it to the test.

A secret delivery facilitated by an AI twin

Access to a meme does not equal replication. As Deutsch describes it, “it has to compete for the recipients’ attention and acceptance with all sorts of behaviours by other people, and with the recipients own ideas.”

I saw this clearly when I spoke to the CEO's colleagues. Despite them feeling inspired and wanting to act like him sometimes, this wasn't always possible. Their context, capabilities or personality might be too different.

As one of the CEO’s colleagues described it: “I don’t know HOW he does it!”

But what if this is partly an adaptation problem? What if the meme, with some thoughtful adjustments, could be successfully replicated by his colleague? Then we'd be playing to one of AI's strengths.

Since I’ve actually already created an AI twin of my colleague Klara (a chatbot in Teams based on interviews I’ve done with her and other colleagues) – I had a chance to try this out.

Being unable to walk has taken a toll not just on me, but on my entire family. My partner Katja has had to take on extra work at home, and our daughter Ester has been stuck inside with me after preschool a lot, since I can’t take her to the park.

So I wanted to create a moment for them that sparked as much joy as Klara’s surprise visit did in me.

I asked AI-Klara what I could do for Katja and Ester, inspired by what she did for me¹. One of the suggestions was this:

A Secret Delivery. If you can’t physically surprise them by showing up, you can flip it—have something arrive for them instead.

Well, that "something" could also be "someone", right? Maybe a group of Katja's friends coming over for a surprise dinner? I figured if I ordered sushi to be delivered, I could make this happen without adding more chores for either of us.

I texted eight friends, four of them came – and Katja was over the moon! She described how it felt almost surreal, having her expectations of an unremarkable evening turned upside down when she came home to find a dinner party and sparkling wine waiting.

Your hidden memes can be part of your legacy

A month ago I made an exception to my isolation and ventured far beyond my home. I took my crutch and went to Fotografiska in Stockholm to see the AI-powered reflection machine I helped create for their exhibition "SPACE – A Visual Journey" – and to show it to my family. I was happy to learn it had been used 70,000 times since the opening.

Looking back, I realize many of its questions were actually hunting for hidden memes:

• “What experience from your own life do you think could be useful for a young astronaut moving to a space station?”

• “What experience from your life could be valuable if you were helping to build a new society on Mars?”

• “What wisdom or insight from your life do you think would be valuable for future generations exploring space?”

Discussing art and life with my daughter, prompted by the machine, and seeing other visitors do the same showed me that AI could enhance rather than replace human connection – even in something as intimately human as experiencing art together.

It was a meaningful moment for me, but unfortunately it came with a price: my foot deteriorated even more. I've since received my fourth diagnosis. Only in the last few days have I finally received clear signals that I'm recovering. Thankfully I’ve been able to drop the crutches at home.

But being in this state of pain, gradually declining mobility and uncertainty for months has made me reflect a lot on negative outcomes.

In the last few years, as the world has become increasingly unstable, I’ve found myself repeating a mantra as a way to appreciate all that’s good in my life:

“Everything I have can be taken from me at any moment.”

Losing my ability walk was a stark reminder of the truth in this.

But what if I lost my life? What would I want to leave behind? For my daughter, my partner, my friends, and the people I have or have not met.

I can’t think of anything more valuable to me than my ability to positively affect the world in my unique way. I believe decoding my hidden memes and allowing access to them via an AI twin could be a way to do just that.

Find your hidden memes – and let them fly across the world

How do you find your hidden memes? You can start simple. Ask yourself:

• What is something I'm proud of doing that most people do not?

• What unconventional insight do I have that could be useful for others?

• What do others find both novel and inspiring in the things that I do?

But identifying your hidden memes is just the beginning. The real magic happens when they start connecting with the people who need them most.

And I suspect the value will not only be in the pure transmission of the meme. I have a somewhat bananas hypothesis I want to test:

What if interactions between a person and someone else's AI twin could create meaningful bridges between humans? Think about it – my conversation with AI-Klara led to a real-world dinner party that wouldn't have happened otherwise. This made me feel grateful towards the real Klara.

And that would have been the case even if I hadn't known her!

I would like to try out an experiment: creating a network of trust where we can access each other's hidden memes. Not only to learn and be inspired, but also to explore the mechanisms of meme transmission via AI.

Are you one of the outliers who feels this idea is compelling? Send me an email (john@ai-salongen.se) and I'll let you know what I have in mind.

Cultural evolution at warp speed

Could this really have an impact beyond a few individual stories and simple life hacks?

I think it might. Consider how much the evolution of human culture has been held back by slow transmission of helpful memes.

Take meditation: no one knows how long it took to go from a fringe behavior by some oddballs to widespread practice in India. But even after that, it took another few thousand years before it was a mainstream phenomenon in the west.

It required both coordination (reaching the right people, e.g. the counter culture movement) and then adaptation (e.g. stress researcher Jon Kabat–Zin introducing a secular program in a medical setting) for it to really take off.

And in my case, it only clicked when I stumbled upon it in the context of a psychology master's program.

But what if AI could help us radically speed up the cultural evolution?

What if our most valuable, unique ways of moving through the world were given a fighting chance to go viral?

That's what I want to explore.

/John

1. In this case the meme wasn't part of AI-Klara's knowledge base yet – I added it in my question. But the point I'm trying to make here is that AI is highly capable of adapting memes to fit a person's specific context.