from oral traditions to AI reasoning
in pre medieval times, irish monks spent their lives copying christian texts decorated with their own inspiration of the famous interlace designs, mix of calligraphy and color illustration, blended with rigourous text reproduction for transmission to future generations. That was a notable step from oral traditions to text based transmission in a country where the mythic druid culture, based on storytelling oral transmission, was vanishing.
From paper to electronic storage
Until the 90ies of the twentieth century, there was no internet. Text transmission had been rolling via paper to brain and vice versa for centuries. At this time, I myself used paper magazines to copy, letter by letter, code snippets onto ms dos basic programs. then rapidly , text processing software like Microsoft Word popularized copy and paste as visual computing was born, with the introduction of windows based interfaces. That was in the late 80es, under the strong influence, of course, of the newly introduced Apple Macintosh (1984), the real inventor of the clipboard concept that goes with copy, cut and paste. Today we all use copy and paste, and windows users have the privilege of native clipboard history, even though most of you haven’t ever heard about it (mac users can install free clippy) !!
I recently (december 2025) read an article about a translator complaining that his last job before losing everything to AI was to feed AI : “most jobs involved training the machine that would replace us” (David Creuze, translator, interviewed by french newspaper Liberation, 01/12/2025). Well I’m sorry to say that this is not new : we’ve been replacing ourselves with computers since they existed. Copying text letter by letter was the first step in the global transfer of knowledge from brain and paper to silicon memory.
Typing code with the keyboard, letter by letter
My initial focus in the 80es was on to play with algorithms : young teenage wanted to develop games. That was the time when writers started to use computers as notebooks, while artists started working with Paint, then Photoshop, and now with Procreate or Illustrator. We all worked on more or less linear process until HTML, HyperText, came into action around the new millenium. That’s when we started reading non linear. That’s when we started having multi branch stars in our head, or may be another common analogy would be… the now famous spider web. The WEB 1.0 was in a way still very linear though, as interactions from user to server were limited, the advent of forms and server side processing with HTML preprocessor languages like ASP or PHP went out live, and in less than 10 years we switched from the free internet encyclopedia to the nowadays data collection fair where the reader is now a product, source of endless analytics and business ideas. The smartphone, which I first met in 2007 in the hand of a local business leader queuing for bread in my favourite village bakery, extended this perfect human data gathering sensor.
Playing music with software in the 21st century
music is about mathematics, but my own practise as a traditional performer on modern replicas of old wooden instruments involves body dancing and breath management on modern transcription of old repetitive tunes schemas that took centuries to migrate from the 15th century european courts to rural areas where I live. We are one of the many performers of oral transmission of rhythm, melodies, and more recently harmonies. Of course we’ve using some kind of paper transmission too, I never really learnt how to read music but could decruypt a single irisgh Jig or Reel from time to time when friends lack the time for reciprocal and inimate pedagogy. And 15 years ago a colleague of mine mentionned ableton and this morning 15 years later I found out how to create subtle groove just buy moving squares around a virtual board. Nice. That does not lead to a smooth transition with 2025’s AI explosion though, as I decide not to ask anything about music to those guys… But still, using Ableton is a strong step in music production that summarize 50 years of music automation that produced very intereseting human creation like Daft Punk, Vangelis (for the french).
into the future : 2014, Leopold Aschenbrenner writes the Situational awareness
in a prospective essay written in june 2014 about the decade to come, the author anticipates the forthcoming massive data centers and super intelligence that is supposed to overcome human intelligence. And in less than a year, most of the current situation he describes are more than outdated. The short term memory problem of chatbots and generative AI is now long term solved with the integration , for example, of code editors with most AI models. Specific chatbots companies relying on mail machines come out every day. The data bottleneck problem, postulating that the existing dataset running learning session is not growing fast enough to sustain growth, shouldn’t be a worry : since the writing of this document (june 2024), chat bots have been fed with no less than 2 billion user natural language queries per day (source : demandsage). In August 2025, Anthropic updated its policy to allow user conversations to be used for Claude training, requiring users to opt in or out by September 28, 2025 (techcrunch). OpenAI uses ChatGPT conversations for training by default (users must opt out), though this doesn’t apply in Europe or to enterprise customer (Cnbc, Claude)
Photoshop 2024 vs 2025 exemple, and the brain capacity loss and gain
surprise, i accidentally used Photoshop 2025 instead of 2024 and could not find my usual very efficient selection tools, like the magic wand : took me 5 minutes to search for my favourite tool and finally reverted to a previous version I luckily left on my laptop. It seams we have crossed a peak there where new AI augmented software is LESS efficient for an experimented user than previous versions, for me anyway at this time this was case. Yes sometimes more is less. Of course I have no doubt that 5 years of AI progress might make my former tools irrelevant, but in the meanwhile I am happy to rely on my own experience, a wonderful alchemy of networked reciprocal symbiotic neuronal connection from eye to brain to hand, to achieve the simple task of isolating a scanned drawing from the raw paper where the original idea was layed out . I often think of our human successors using only GPS to drive, cycle or just wander about, and there is little doubt that, relying on those , within a few centuries , humans might completely loose the ability to orientate using brain spatial memory, finding this capacity as absurd as today with consider fire starting with branches like our ancestors did for thousand of years. In the meanwhile, of course, new brain features will appear, like, as of january 2026, the new skill of prompting AI vibe coding session.