Now that AI has started to reverse engineer famous software, albeit with the obvious oddities that any initial proof of concept will have, it opens a whole new door to the a world order where creativity, innovation and entrepreneurship becomes a game of who can steal the most ideas and feed them to a machine to create something that resembles a finished profitable product in a speed and volume never seen before.
It is the consequence of years of neglect, of seasoned professionals decrying how their expertise, not just their jobs, where being replaced by an imitation of their talent.
It wasn’t a battle of results, it was a full on war on productivity, on demonstrating that the value of things never lied on being able to reproduce something fast, but to come to learn why and how something should be built, purposely created to achieve a multilayered goal that could feed, nurture and produce further value exponentially while avoiding the errors of yesteryear.
To not only imply that human effort can be swapped that easily but to ask to divert all investment and resources to a glorified autocompletion machine to fuel its unrelenting advancement without seeing that the real reason was to manufacture consent to develop tools of mass surveillance and to unlock a new arms race within security and military forces, the biggest industries that exists, shows how this era is just the final step in a plan that started a few decades ago, when the fear of equality and more importantly, equity, threatened the powers that be.
I won’t be tracing the steps in detail of how they built their Trojan horse. How big tech monopolised our information. How social media removed the sources by constraining it all to a 3:2 square of just a few headlines that could only be commented on, liked or shared as quick reaction in a toxic feedback loop. How external links where treated as dangerous instead of the actual way to verify its provenance, and making sure that you knew the source, and its neutrality. How user data was harvested and sold to create user IDs, whether anonymous or not, so it could be used to target people by their biggest terrors. How those fears were used to push people to a state of constant stress and panic that would make them more gullible, easier to convince of whatever boogeyman was after their fake stability next. How the push to ultrarealistic artificially created video and audio, once the threshold of stolen text was reached (to teach the initial prototype LLMs) allowed them to create ways of poisoning the well as needed, to create a world in which any piece of content could be quantumly fake and real at the same time. How within two decades we went from a world full of possibilities for the common people to create an fair society to one in which we have given away all our freedom and become puppets of an algorithm that the higher ups are about to lose control of.
A world in which science fiction turned out to be just a premotion of reality.
I need to provide, as a web developer, my opinion of AI so it can contextualise what you are reading. After testing what stable difusion could do with a few prompts more than 4 years ago I realised that the results were only mimicking existent artwork. It didn’t take much researching to realise, with my own experience in photography, video and even drawing, that they had streamlined the work of existing human thought. To combine references, learn from them, layer them up to compose something slightly new is always been the process of creation. When it comes to the automation of parts of the process, it is always a slippery slope between how much of the previous references are left unchanged, unprocessed, or unmentioned and how unique the new piece is in comparison. It is an ongoing debate of what plagiarism is and is not. But to allow anyone to ignore the debate and jump to the outcome without the ability to trace it back, discuss it is or check its veracity was the twist that activated all the people without morals to jump into the area and become full time experts at everything. That’s were I drew the line. I am comfortable integrating a glorified search tool in my workflow to find things faster, but I won’t replace the checks and balances needed before and after I add my contribution just to pretend to be something I am not. After realising that the inherit theft could not be avoided, I have limit the use to the inevitable search results that in my opinion, should have been the only use for it. Code completion, formatting, testing and debugging have always been supported by automated tools, but to ask a machine to write something for you from scratch is another different thing.
The only reason why I was, and I am comfortable being, not staunchly anti “all types of AI” is because there is an ethical use of it, when contained to a curated and reviewed data-set that you own and can be used internally, even offline, to increase the speed of production in companies or individuals that have a history of knowledge that can be used as the source, and can be query, shaped and transform in ways that would take decades for a team to achieve on their own. But not to replace it or them, because the process of doing so has always been how we progress.
After years of hearing people rightfully predicting how their industries were being cannibalised by the precursor of Skynet, with uncanny videos with altered faces, alternative versions of any song ever released, whole books made in an instant and even consultancy firms being caught passing AI outputs as real research, we passed that point of no return.
What is the incentive of coming up with or pitching new ideas, when you know anyone can and will take them, just prompt a hollowed out version of its ethos just to resell it, launching something similar just to get the investor’s funding or people’s attention, diverting the money and racing you to a finish line within days that never cared about its purpose, its maintenance or its usefulness in the long term?
Machine learning has always needed inputs, a collection of data that had real value, in order to become the powerhouses they are today. First they hoovered the whole of the internet and came out with a few tools for free testing so the public, as usual, could help debug it. It gave them enough time to obtain almost unlimited funding, measure public opinion and counteract authors outrage, to the point where mass theft only resulted in a few settled copyright infringment fines. Hardware and software manufacturers pivot to put it at the center of their business models and their products, part of the PR machine that was needed to create a financial bubbles for them to benefit.
It was loudly sold as a mass consumption tool for anyone to increase productivity and “create”, bribing us collectively one way or another for what it was always a clone, merge and polish exercise for already made ideas.
When Napster or eMule were released, I don’t remember all professionals proudly claiming to be pirating songs in their CVs. The tool’s purpose was clear, to share all type of material, included copyrighted one, peer to peer. It was clearly a way to own content that otherwise you “wouldn’t purchase or access”. How true that sentence was in each case was the primary point of contention. Imagine a few thousand people back then, downloading Michael Jackson’s Invincible album, pitching every song a few semitones, then renaming the files with a new cover art and go around claiming they are the best producers in the world creating innovative music. We know the plagiarism would have been too obvious, but just as the thriller cover art resembled another contemporary artist work but was only publicly perceived decades later, latency or obscurity does not make it less of a theft. Just because now the software does it for you does not remove those steps from the equation. They are still there. But somehow, I now see people on LinkedIn proudly sharing how AI is helping them create at volume, shifting the focus from the usefulness of the product to the speed in which it can be “achieved”. But the value of it, it’s longevity and usefulness are no longer there, they are inferred because it looks the part.
Back when piracy was rife, we had an advert before every movie warning you of the consequences of downloading one movie, one song, literally equating them to stealing a car.
Now, deciding to stay away of using cloud-based AI to CREATE things, and instead waiting until it can be a tool to streamline tried and tested, peer reviewed pipelines within moral boundaries and proprietary data, get you ostracised to some type of utopian idealist that never understood the way the world works.
The industrial revolution gave us engines, and we forfeit the environmental impact to push us forward, and now we are seeing the consequences. But still, those engines didn’t go anywhere unless they had a human as the driving force taking that power somewhere useful. Now not even that is needed. If that is the entropy of progress, we are removing ourselves from the equation for people that seem to hate us being there.
I refuse to see that as the path forward. It feels as if we all agreed to let our mobile keyboard suggest every sentence on every conversation for us to simply click it or ask for the dice to be rolled again, instead of simply guessing the next word we already started typing. Yeah, you can tell them the topic, but next thing we know, there is no nuance, no direction, no intention, just outsourcing ourselves to a void to decide who we are and what we think. The inversion of free will, us correcting a machine trying to read our mind instead of the other way around.
What happens when there are no more knowledge or art for AI to feed from? Whenthe AI errors are inherited by more AI without human overlook or even the knowledge to spot it? When there are no more original thoughts from humans to contribute to AIs collective knowledge? And what is the effect of the lack of human progress due to the outsourcing of talent to AI?
Do we really need to ask AI to know the answer?
Ok. If you don’t believe me, believe the machine:
When AI exhausts original human content, it triggers “model collapse,” a progressive degradation of quality where systems feed on their own synthetic outputs, amplifying artifacts and losing nuance. As training data becomes increasingly homogenized, AI loses the ability to learn new patterns, effectively reaching a point where it has “nothing more to learn” and begins to produce repetitive, formulaic, and less effective results.
The erosion of human progress occurs because AI substitutes for the context-specific effort that generates general knowledge as a byproduct. When humans outsource creative and intellectual labor, they cease producing the “thin signal” of institutional insight and professional understanding that traditionally feeds the AI’s knowledge base. This creates a vicious cycle where reduced human effort leads to reduced general knowledge, making the AI less effective while human capacity to compensate atrophies.
Key consequences of this outsourcing include:
Loss of Intellectual Diversity: AI favors the familiar and efficient, sidelining alternative perspectives and reinforcing dominant narratives, leading to intellectual homogenization.
Atrophy of Creative Skills: Overreliance on AI reduces opportunities for creative risk-taking and deep thinking, causing future generations to lose the capacity to think creatively without machine assistance.
Diminished Innovation: While AI may improve individual idea quality, it reduces the diversity of group thinking, which is essential for breakthrough innovation and solving complex societal problems.
Silent Knowledge Depreciation: The decline in human-generated knowledge happens gradually as practitioners stop contributing to the renewal of the knowledge base, leading to shallower reasoning and less robust engineering in various fields.
But it is even scarier. Because some will say the solution is to wait for Artificial Super Intelligence, the one capable of creating NEW thoughts, ideas, and concepts. But who are they serving? What will be their purpose? Can a machine truly understand what the human experience is and what it needs? It is ever more common that not even humans pass that test, so I refuse to wait for that answer while that is still the case.
To expect that AI will be a long-term solution without understanding that this will only be possible if we agree to continuously feed it our stolen creativity and free labour, in order to prevent its collapse, is disingenuous at best.
To expect that using the outcomes of AI as products, or even as concepts or prototypes, without ending up in a decadent loop of regurgitation, is naive.
The people doing new things will be fueling people prompting their goals, and bypassing the work efforts of those that will give up eventually, when everything is lost.
In the short future, both sides will be angry that the tug of war between owning and getting credit for what you created vs stealing what others have done for machine learning to improve their outputs has simply destroyed everyone’s livelihoods, and the empathy or drive to bridge the gap.
Only the ones selling AI will profit, until the bubble collapsed and they can benefit from the ashes, investing in resetting the new world order with whomever agrees to their next nefarious plan, because everyone else will be destitute, in an impoverished environment that is struggling to understand what is real and what is not.
Just for them to present their newest ASI and tell us it will solve that problem too, if we sign away all our resources, assets and what is left of our free will.
Leave a Reply