For some time now, apocalyptic predictions have been appearing in the press, even in much of the IT-themed media about mass layoffs of workers "fired because of AI", "automation that will make human labour useless", and so on and so forth.

In this post, I would like to clarify a few points, distinguishing between the narrative and reality. Will there be many people fired "because of AI"? Some, but not the tsunami that is being heralded. For example, far fewer translators are already needed than in the past because, to translate texts where 10 humans were previously required, today 1 human and an LLM do the same volume of work in the same amount of time, with the human acting as a proofreader for the LLM. In low-criticality areas, identity checks, typically for travel documents, will be carried out where regulations require it using a ticket + biometrics, with a single human hopping between several gates where there are problems, instead of one human per gate.

Already here we reach the first divide between narrative and reality: is biometrics actually necessary? No. For instance if we verify identity with a digital signature, simply a smart-card with a PIN, this verification is far superior to biometrics. If we want to open a bank account remotely, it is much safer to receive a contract as PDF PAdES-signed by the bank, to which we add our own signature if we agree; in one go we certify the identity, the time of signing, and freeze the contract conditions so that no one can modify the PDF without invalidating signatures. It costs much less than viewing an ID and the person's face to "recognise" them, and it is immensely more secure.

This example already shows that a large part of what is being automated via ML solutions should actually be done without ML at a much lower cost and with much better innovation; it isn't done like that because someone's cost is another one profit.

Someone will say, "Eh, but there have been a lot of layoffs lately!" yes: bank staff are being laid off because after decades people are finally starting to use internet banking on a large scale, and so having numerous branches no longer makes any sense, hardly anyone goes there anymore, it's a pointless cost. Many IT professionals are being laid off because IT companies are no longer innovating; they have crashed against their centralised development model that doesn't scale, and today, while continuing with this because the last thing they want is to let go of the golden goose that made them super-rich, they can no longer carry on as before. While their services crumble and users, fed up with problems, start to consider alternatives, which imply what neither the banks, nor the IT giants, nor the real estate giants want, they try their best to trap the majority. How? Well, with something that runs on their systems and which for now doesn't work great at a self-hosting level, and with public subsidies to keep making a profit.

With what justification?

With those mentioned above: LLMs are great PR; under the guise of surveillance and persuading the masses sold out to governments, coffers are being filled. Desktop hardware is being made hyper-expensive so that most people move from desktops where they can be sovereign with free software, to mobile or rather almost entirely to the cloud, starting with students who give up on (FLOSS) Desktops just when they could learn them best.

Today, most people are starting to use desktops a bit. Students who are starting to loathe spreadsheets, preferring Python instead; people choosing FLOSS desktops, typically GNU/Linux; people starting to self-host what they need etc, are becoming common. At a more macro level, digital public administration is becoming common, so fewer people are needed there too. With it, the question "does this procedure make sense today?" becomes common for all those operations that required many steps on paper but make no sense in digital form, as the data is immediately ready and shared. This trend means the giants feel the risk of people gradually rediscovering the Xerox desktop model; after all, even today, in every innovation, we tend towards it, at a "crayfish pace" compared to commercial innovation.

Now, we've been able to have this model since the time of Xerox, that of the Augmenting Cognition of back then. Today, the price of a desktop assembled in the West is within reach of almost every family, just as we have internet connections almost everywhere. We've had IPv6 for ages, or at least we can have it, and for the most part, we're starting to get it due to the lack of IPv4 addresses. So, having a global per host, knowing what a global is and therefore what to do with it, is roughly within everyone's reach.

Today we could have, and without "interventions" we most likely would have, a generation that starts to discover the power of computing and obviously won't want to lose it once discovered and conquered.

AI, pushed the way it is being pushed, is the key to preventing this from happening: "hey, you can't train a model at home", "hey, you can't have enough storage or computing power to have an efficient and useful ML system at home", "you MUST live on the giants' systems, as a service for which you pay a subscription and over which you have no control". "Hey, you don't want to learn how to use a computer; you want it to learn how to be used by you." The same narratives used back then, just repackaged for today's society.

In fact, today we could even eliminate banks, not because of cryptocurrencies, but simply because we could have just one public central bank and operate with it. Today we could avoid a sea of paper and everything that goes with it, while still possessing information at levels even better than paper. That is, today we could avoid many clerical jobs and yet not have a tsunami of desperate people. But this goes against the 2030 Agenda's where "you will own nothing" according to the WEF, because IT is the nervous system of society. A society where individuals form the nervous system and, by individually owning their part, are components who know the value of possession and don't become smiling slaves to the master of the day, the Canon model in China. Therefore, there is an urgent need to invent a narrative that averts this danger for the oppressors of the day.

This is where we are, and why it is imperative that a sufficient mass understands and rebels, imposing a change of course. This doesn't mean eliminating LLMs, which are good search engines implementing Conrad Gessner's dream of the Universal Library/Library of Babel, but rather a FLOSS society on open hardware, digitised for the common interest, not that of a tiny few against the many. Because that's how evolution works: it advances; if a tiny few do it, they do it for themselves; if many do it, they do it for themselves, and the interest of the many generally coincides with the interest of humanity.

The first phase is understanding the concept of digital private property (one's own data, hardware in one's own home, domain name, global IPv6 address, etc.) and that of the common good (FLOSS, distributed and decentralised networks). It is from there that we move somewhere positive for the many, and lethal for the few kleptocrats.