top of page


Our esteemed founder Nicolas Roope wrote this here thing (bad grammer and speling = human being) about the averaging effect of Ai and why that might not be a terrible thing for creativity and innovation. He penned it in readiness for a panel at the rather excellent Fearless event last week and was published on New Digital Age . But here it is below in this email, because we at GGGGGGGGG know how hard clicking on links can be.


MEAN MACHINES When a bunch of ambitious, artful friends decided to set up an agency over 20 years ago, we had one idea in mind: to demonstrate that if you pushed things toward the exceptional as an agency, you’d bear fruit. Despite having been in and out of a vast speculation bubble, digital was still nascent and still a deeply misunderstood and undervalued medium. After the crash, what was left of the young digital industry was churning out mediocrity by the bucket load. We believed that the potential of the medium needed to be discovered or invented because high standards hadn’t yet been set, and without these aspirational waypoints, they would stay where they were. Throughout my career as a digital pioneer, average has always been the enemy. Its gravity is immense because, numerically speaking, average is the natural target. It’s where the volume lurks – in the domain of the illusive everyman. The average provides shelter for the faint-hearted because the risk, in average, appears low. There is safety in these large numbers. In contrast, fast forward a few years, and we’re in the age of extremes, where cultural divides are exacerbated by social media, pushing users out to the edges in pursuit of an ever greater slice of their time and attention. The average has been out of fashion for some time. We’ve been living with the cultural and social consequences of this extremism. It’s written a couple of crazy chapters in recent history. Chapters many would like to forget. So it’s interesting to observe the averaging engine gain traction faster than any technology preceding it. In the pursuit of pattern matching in massive aggregates of content, LLMs must use the mean to centre and blend its view into something “accurate”. Sampling everything and moulding that into something coherent necessarily results in shaving off corners as every word, phrase and concept converges on a single answer from a mash of million samples. In my agency days, the battle with average was grounded in the belief that the comfortable, familiar and bland doesn’t move people, excite them, or inspire them to imagine something different and therefore fails to inspire the change you want to see. So now we have an incredible technology that does average better than anyone or anything. The sum total of everything that has been said up until this moment. The ultimate soup of assimilation. And while these models may make a pretty good guess about what will happen next, they don’t know for certain. We can always choose to take a sharp left turn, buck the inevitable, and do something outside of the predicted pattern. What I’ve described is essentially creativity. Lateral not inevitable. Conjuring, synthesising and inventing, not just following a fixed method, emulating, plagiarising or painting by numbers. This isn’t just a tension between creativity and AI but an age-old one between creativity and boring predictability; only these new technologies have brought it into sharp relief. The camera’s invention both rendered painting pointless and yet also liberated it by removing its former primary task of accurate representation. The result was an explosion of creativity that exploited this new freedom, that in turn created the space for culture to flourish. A new way to look at and experience the world spawned a new world. LLMs may have just nuked the tedious, average creative tasks we never wanted to do in the first place. But maybe these mean machines will also result in an explosion of creativity for our times? And finally, place a higher status and value on true creativity as the boring, bland middle ground has finally been commodified irreversibly.


If you're interested in transcending the average / expected to fend off the computational competition, please get in touch.


Recent Posts

See All


bottom of page