Skip to main content

AI is not to blame for idea recycling (74)

Why fuss over originality?

The most recent iteration of AI technology, generative AI, has been controversial too say the least. Some praise it for the efficiency it brings; others blame it for killing individual creativity. Part of that criticism has merit: generative models do not invent out of intuition or by their "gut".

But what does “original thought” even mean? Is there such a thing as pure originality? And why does our civilization’s latest technology work by remixing and recombining what came before? That’s what I want to explore in this post.

A tale from the R&D multiverse

Let’s start with a story. I work in an R&D department of a multinational company; a big part of my job is feasibility testing for new product ideas. My team and I have been developing one particular pitch for about five years. At the project's start, management handed us pitch slides with polished graphs and a bold statement: this new technology could disrupt up to 30% of the relevant market.

Over those five years we did the work. We modelled the technology, ran safety and cybersecurity analyses, and based on our new insights re-estimated the potential disruption at roughly 38%.

That outcome felt satisfying: the number seemed to confirm and even exceed the original ambition. We adopted the 30% figure as a reasonable threshold in our reports, papers and external messaging, even reusing the original graphs in modified form.

Here’s where it gets twisted. Recently, while assigned to strategic work, my team read consultancy reports submitted to the company over the last years. To my astonishment, one report dated less than a year before our project began had warned that a new technology could disrupt about 30% of the market — and included graphs very similar to the ones we had been using. We were looking at the seeds of our project.

The ouroboros of knowledge

This story about our little R&D project got me thinking. This probably started by one senior leader just wanting to explore the potential threat that the consultants warned him about. Some years of research later, the worry was a bit more substantiated and a bunch of more publications and papers were made public. This "new" knowledge probably will be processed by more analysts down the line. Eventually, at someone's desk a very similar "potential 30% disruption" report might appear. And we might not even have been the first link in the chain.

I think I understand why this happened. Its all about institutional incentives. Consultants, managers, and analysts operate under reputational pressure. A neat, memetic number like “30%” is catchy: it travels. Slides get copied, summaries are made, and an idea acquires momentum independent of its original evidence. The process is social as much as intellectual.

Crucially, this chain of regurgitation happened before accessible generative AI became common. For example, music sampling and the recent obsession over nostalgic reboots and franchises, show remixing predates generative AI. The image that comes to mind is the ouroboros — a snake eating its tail. Is that what our culture has (d)evolved into?

Break the loop

True creativity however does not necessarily need to come from "thin air". In fact, most discoveries have historically being all about mistakes or applying known tools to a different domain. Having access to the same pool of data, but looking at them from a different angle *can* still create new information. True originality is all about interpretation, not production.

Perhaps that’s why generative AI behaves the way it does: it connects patterns from the data we collectively produce. We have realised there is power in our shared information, but our risk aversion sometimes prevents us from spotting the patterns ourselves. So we build machines to connect the dots.

How do we use that to our advantage? Break the loop: take the insights the machines give you and own them. Ship the product; see what happens. It might succeed, it might fail — or it might do something surprising. Either way, making is how we learn. Generative AI is not the origin of idea recycling; it’s an accelerator and a mirror. If we want less echo and more novelty, change the incentives: experiment, tolerate failure, and value interpretation over repetition. Let’s find out!


 

Comments

Post a Comment

Popular posts from this blog

Easter feast and task pipelines (48)

A Greek Easter feast As this post goes live, your host and author is most likely indulging in the joys of Easter Sunday. In Greece, where I come from, Easter is the biggest holiday of the year. And, as you might expect from a culture famous for its love of food, the Sunday feast is a central part of the celebration. The meal varies by region and family tradition, but one thing remains constant: roasted lamb. It symbolizes the sacrifice of Jesus, and it's always the star of the table. If you're reading this on Easter Sunday, there’s a good chance I’m savoring some leftover lamb while you scroll through these lines. Now, let me tell you—preparing this feast is no small feat. Roasting the lamb and getting the rest of the meal ready is a coordinated effort, usually involving many helping hands. To make sure everything is done in time for the classic Greek lunch hour (around 14:00), the tasks must be broken down and distributed efficiently. This is where the real planning begins....
Post a Comment
Read more

I think, therefore I am (35)

I think, therefore I am Consciousness—or self-awareness—feels like one of the most distinctly human traits. Sure, some animals are clever, but only a handful, like certain primates, seem to pass the “mirror test” and recognize themselves in their reflection. So, what exactly is consciousness? The truth is, it’s hard to pin down. Interestingly, an accessible definition of consciousness doesn’t come from philosophy or biology, but from medicine—specifically anesthesiology. In this field, consciousness is defined as the awareness of pain and our surroundings during surgery. It's a practical, measurable approach: you’re either aware or you’re not. This definition not only makes consciousness easier to grasp but also highlights its connection to our sensory perception. This view brings us to the classic "dual theory" of consciousness. In this framework, the body and mind are seen as separate entities, with the brain acting as the “control room.” The mind issues commands, the b...
2 comments
Read more

A chess game played at a two hundred kilometers per hour (66)

A chess game played at a two hundred kilometers per hour That is how the sport of modern fencing is often portrayed by its practitioners, or, frankly, anyone with an inclination to dramatize. Often, in student fencing circles this phrase is used ironically to juxtapose how simple, pragmatic and brutal this sport can sometimes be. However, I am inclined to approach this phrase by visualizing its contents; fencing is indeed an ancient game, requiring wit and is supplemented with technology unimagined by previous generations. It really is a marriage cold calculation and passion. A passion which is shared by the previous blog exploring modern fencing technology, and which drives one to explore developments of fencing technology even deeper. How Horses Became Formula 1 A common passerby would not think of modern slender, antenna-like blades as swords. And they would be completely right, none of the three fencing weapons (saber, foil and epee) are swords. Assuming the sole purpose of a sword...
Post a Comment
Read more