Introduction (or “What has convinced me to write on the matter”)
Somewhere in France, two months ago : a ten year old boy plays with his mother’s (Android) smartphone and ends up on Gemini which reads something like this :
“Hello $NAME, here is what you can do near $LOCATION […]”
- How do you know we are in $LOCATION ?
- [Gemini’s irrelevant and pointless redacted answer]
- NO !!!HOW. DO. YOU. KNOW. WE. ARE. IN. $LOCATION ???
Welcome to dystopIA. Youngsters simple (thus naive) approach of life against huge tech giants steamrollers…
Politics
As a previous Intel CEO (Andy Grove) said in early 2000s :
“High tech runs three-times faster than normal businesses. And the government runs three-times slower than normal businesses. So we have a nine-times gap… And so what you want to do is you want to make sure that the government does not get in the way and slow things down.”
… we cannot expect and wait for slowly-evolving legislation to enforce regulation for the sake of citizens (and eventually humanity).
My only hope (for the top-down approach) is that even on the right-side of the economically-political spectrum, liberals begin to understand that this new “technological revolution” won’t just cause massive layoffs and hence unemployment, but also totally destroy the “labour value” so important to them.
How to “emancipate through work” (if you happen to believe in it) if one is enslaved to a black box prompt copying/pasting someone else’s work that has been mixed up with others’ ?
How to “benefit from the social elevator” if most of your studies consist into giving course slides and exercises to a machine only good at respectively summing them up and resolving them for you ?
How to “innovate” (to save life on Earth, as techno-solutionists evangelize) if your only tool is a huge regressive equation, which is only good at dwelling on what has already been discovered ?
You know the situation is completely f*cked up when the ones actually driving on the highway are simultaneously calling for speed regulation (for their public image) as well as fearful about their own behavior.
Energy (or “Houston, we have a problem”)
Infrastructure
We have piled up millions of lines of codes, terabytes of binaries, and now with this new paradigm we keep still : we need anti-crawling bots to keep our websites up, some even feed them trash to “slow” them by wasting computational resources, and we build concurrent models to “fix” inherent security issues… Cats and mice game continues, but planet loses.
As network backbones had to scale up to deal with modern usages (e.g. a whole city streaming 4K of different VoDs at the same hour every day), our power grids and energetic mixes have to go down the same path. And don’t fool yourself : it’s our usages that design infrastructures, not the other way around. Companies and engineers won’t spend money (nor time) on unworthy projects. I keenly invite you to read on the “path dependence”, and here (french) applied to AI.
We reached a point where a datacenter built in a neighborhood¹ almost consumes what the residents do. Imagine.
In the end, the situation is so lunar that we’re encouraged to pee in the shower skip greetings…
Brain
I’d like to mention Luc Julia (co-founder of Apple’s personal assistant Siri) and more particularly his take on why AI doesn’t exist.
The key note is : our brain consumes pretty much nothing compared to what neural networks require, and it usually does better.
So you will tell me human-beings cannot possibly work 24/7, need medical insurances and (what a shame !) happen to fight for their rights. Indeed, we aren’t machines.
But let’s be honest, a “performant AI” is a “well-trained AI”. And training at the end of the day, comes from humans !
What actually happens is : we’re paying few cents per hours some people (for their low brain energy consumption) on the other side of the planet to “train” (very high energy intensive) machines on ours, so as not to pay a (salaried) human-beings decently. More on the “hidden (social) conditions” behind AI here (french).
Earth loses. Humanity loses. Tech parties “wins”.
Economies
It has been some years we haven’t experienced the explosion of a speculative bubble… We had Internet around 2000, real estate that begins to collapse in China, cryptocurrencies that happen to more or less follow “real-economy” trends between two insider trading infringements, …
Investors needed a new horse to bet their extra-money on, and AI appeared. Once they will understand that “garbage in” usually and eventually gives “garbage out”, be ready and brace for impact…
The same and rather old patterns apply : competition, predation, economic headlong rush (not to spoil the property owners) : our Occidental unindustrialized (and thus massively unemployed) societies have come to an end (and are slowly declining).
Owners and renters can still put some money on the table, wait for the donuts to grow (because others follow), and gather the differences (often positive for them).
Trivially, we could say that “owners” have interest in such a “technological revolution” to happen in order to maximize the expansion of what they already have. On the other hand, “workers” will be compelled to embrace yet another enslaving companion (for “productivity”).
Relationship (or “What we’re actually dealing with”)
Would you value one’s speeches if you knew each one of their sentences next word were computed using probability ? But you do if you use generative AI.
Would you seriously value a discussion with a delusional/mythomaniac friend of yours ? But you do if use generative AI.
Would you share some personal/confidential data with someone publicly known to use it for their own benefits (and that may make them available for the rest of the world afterwards) ? But you do if you use (a public) generative AI.
Do you know Crocker’s Rules ? If not, you really should.
What would be the point of feeding bullet points to an LLM² for it to generate a well-written e-mail, if on the other end, your recipient also uses an LLM to sum up your block of text down to… bullet points ?
So you’d have expanded-then-compressed some data, which at best is useless, and at worst the essence of your key notes may have been lost during the process.
Here’s a situation : you have a problem to solve, and a quick query using your favorite Web search engine and relevant keywords gave no results : don’t lose time asking a generative AI, it will hallucinate and you will just lose time (and consume more energy for nothing).
Philosophy
A neural networks acts more or less as a linear regression. It will try to minimize what we mathematically call a cost function. So it’ll be a “performant” machine, for a given set of tasks. But it won’t be able to create new things.
If it’s the end of innovation, it’ll be the end of Societies (at least as the forms in which we know them today).
No more artists. Thus, all artists ? Creating Generating content relying on NFT opened to speculation, on a daily-basis ?
A choice needs to be made : do we want to keep a “social usefulness” ? Or at least, the possibility to have one ?
Is a machine only performing syntactically-correct answers sufficient ? Try to e-mail a well-written but factually wrong text to your boss and you’ll eventually see it isn’t…
Is “Chat” a GPT (Global Persistent Threat) to humanity ? Yes because, by paraphrasing and translating Emmanuel Dockès³ :
“If you happen to only free people that can stand thralldom, you only free the strong ones”.
If you think users should be blamed for blindly following definitely wrong instructions, try to analyze who operate the service, and why.
“Tool” dependency
What will you do when advertisements will directly be included in prompt responses ?
You will pay for the service your work (life ?) depends on ? As you already paid for your music, VoD, game, Internet access and other entertainment-thus-optional services, while the persons feeding you daily kill themselves ?
Does a plumber uses free tools ? Does a plumber would even rent them ?
We shouldn’t start relying on tools we do not master, trust, or that we cannot replace (I’m looking at you Office 365 
).
Giant tech companies offering “smart” assistants for free are currently performing a massive “technological dumping” that is never discussed.
As it started before with “clouds” and e-mail providers, we are giving them away all of our personal data, and they will eventually be able to make you pay for their services.
You don’t care about privacy ? But the ones you care about may !
I’d like to remind you of a previous popular take on e-mails : Google Has Most of My Email Because It Has All of Yours
Unless you’re called Robinson and live on an island, your (in)actions have an impact on society (particularly on your close circles). If you let predatory companies entering your life, they also enter lives of the ones you meet, spend time and discuss with, make love to, … In the end, our ideas, our voices, our emotions, even our intentions go to them. And that’s scary.
Don’t forget that everything you type on a platform that belongs to a third party, 1) doesn’t belong to you (that were the terms of service you didn’t read when signing up) and 2) it’ll be a matter of time before it ends up being publicly available on Internet.
Conclusion
We were arguing whether history is written by the victors, and maybe we’ll soon debate whether past facts existed at all (french).
I don’t want a world where the only particularity of human-beings would be their thumb opposability, that will serve machines (somewhat already close to warehouse worker conditions, for you to receive your packages on short time notices…).
I don’t want either a world looking like something between Idiocracy (2006) and Don’t look up (2021) (already crossed with Her (2013)), where average IQ went drastically down due to massive assistance or billionaires deciding how we should live (or die). If it doesn’t end up as in Terminator 3 (2003), once we’ll have had connected the first AI to military infrastructure to fight… itself.
Will we be able to live without a personal assistant ? Optarim verius, quam sperarim. In the meantime, this post has been written without any.
#StepAIside, #WalkAwAI, #OptForBrAIn
Footnotes
1 : La Courneuve, France
2 : Large Language Model
3 : “Si vous ne libérez que ceux qui ont la force de refuser la servitude, vous ne libérez que les forts”. p302, Emmanuel Dockès (2017), Voyage en misarchie. Éditions du Détour