Publish at June 28 2023 Updated June 28 2023
Each of us thinks in the terms of our perceived reality. In other words, most of our thinking depends on our senses and our environment, and Artificial Intelligence works no differently. Many artists and philosophers are aware that the materials of our thought are often not original and have been trying for ages to free themselves from them, but even pure reason always ends up embodied somewhere.
Socially, our activity is mainly geared towards satisfying our vital needs: housing, food, security, etc. We worry about our food because it's the only thing we have. We worry about our food because we have to eat to live; it's our biological condition. We can choose many things, but not to live without a body, or even outside society. An A.I. is not limited by these conditions, but it has nothing other than our preconditioned data on which to base itself.
Our power of choice is limited by the preconditions of our existence; even socially, most of the choices available to us are determined by our environment. We may have a choice of hundreds of consumer products, from clothes to air conditioners to cars, but we don't always have the choice to do without them, and other possibilities are now inaccessible to us that were once commonplace. Yet most of the conditions that limit us as humans are not universal; they don't apply to all living things or in all environments.
We don't have the choice to breathe, but we do have a higher level of choice: that of choosing different points of view, of broadening our perspective beyond ourselves, our environment or even our species. In this way, we can achieve respect and tolerance in the best interests of all - something that fanatics of all stripes are incapable of achieving, and which is a good test for identifying them. In this way, unbalanced agents can be ruled out when it comes to establishing constitutions and principles, whether intransigent religionists or obsessive materialists.
We don't think like whales - their living conditions are too different from ours - but there's nothing to stop us including whale lifestyles in our reflections, while excluding those of suicidal whales.
One of the major problems with A.I. and algorithms is the quality of the data on which they are based: you can't reach reliable conclusions with false or fuzzy data. When it comes to qualifying data drawn from the Internet, you need to be able to distinguish between observed and altered data, and to separate conclusions drawn from structured reflection from those without any logic. Popularity criteria are not enough to determine the value of a piece of data or an idea, otherwise we'd still be convinced that the earth is flat and that it's at the center of the universe. The criterion of the results obtained is much better.
Artificial intelligence does not judge in terms linked to its material condition. It works with accumulated data of all kinds: objective data from sensors we've created, such as cameras, satellites or meters, texts from more or less articulate bloggers as well as those of recognized luminaries.
At a second level, the A.I.'s weighting methods are oriented towards specific objectives, which represents its conditioning, and this is where part of the control challenge lies. A concept such as killer warrior robots brings up the analogy of the intransigent fanatic. To a lesser degree, A.I.s programmed to increase the impulse to consume or the time spent on the gaming site are equally open to criticism. It's conditioned thinking, but not in our interest.
The move towards A.I. control rules is not a paranoid whim, but rather a healthy philosophical response (Philo - affinity, Sophia - wisdom). The answer that emerges is to determine the principles to which A.I. must conform, principles such as collective and environmental interest, truth and overall tolerance of that which seeks to live decently, not at the expense of others.
The quality of the data they use is an initial criterion, but that of the conditioning of their algorithms can be evaluated by the results they produce, which can even be simulated. By adopting a way of thinking that is open to feedback and multiple points of view, not just that of the market or a political model, we can move forward gradually without causing catastrophe.
A.I.alignment and containment are theorized methods, but ultimately it's humans who will decide whether or not to implement them. It's not a simple subject, especially in a changing reality where what works well today can go wrong a week later, but it's essential to condition A.I. to our reality. In so doing, we may also learn to broaden human viewpoints to include all nations and all living species.
Illustration: Gargantua
References
Controlling artificial intelligence? - Networks - 2022
https://www.calameo.com/read/000215022c6ea54f7600d
How to control the proliferation of AI? A conversation with Robert Trager - Victor Storchan - Le Grand Continent
https://legrandcontinent.eu/fr/2023/05/19/comment-controler-la-proliferation-de-lia-une-conversation-avec-robert-trager/
Alignment of artificial intelligences - Wikipedia
https://fr.wikipedia.org/wiki/Alignement_des_intelligences_artificielles
What would happen if a conscious artificial intelligence took control of humanity? - Claire Manière - Trust my Science
https://trustmyscience.com/intelligence-artificielle-consciente-controle-humanite/
Controlling artificial intelligence, yes, but how? - Benoît Georges - Les Echos
https://www.lesechos.fr/2016/11/controler-lintelligence-artificielle-oui-mais-comment-218494
_______
Metaphors at the source of thought - Rémi Sussan - Internet Actu
https://www.lemonde.fr/blog/internetactu/2014/10/01/les-metaphores-aux-sources-de-la-pensee/
La condition philosophique (La réflexion, le préréflexif et la question du scepticisme) - Vincent Citot - Le Philosophoire - Cairn.info
https://www.cairn.info/revue-le-philosophoire-2007-1-page-219.htm
20 dangerous threats of artificial intelligence - Céline Deluzarche - Futura
https://www.futura-sciences.com/tech/questions-reponses/intelligence-artificielle-20-menaces-plus-dangereuses-intelligence-artificielle-14343/
You might also like:
Virtual or real, bring robotics to the classroom and beyond!
Robotics enables the direct integration of several disciplines: programming and computing, industrial design and 3D printing techniques, mechanics and electronics. The practical educational benefits of robotics can keep you busy for years. Many robotic projects touch on social, environmental, scientific or artistic fields. Everything you need to get started.
Can "virtual eloquence" replace eloquence?
The Covid 19 pandemic has strongly affected the organization of eloquence contests, hence the birth of what could be called virtual eloquence. Despite the prowess of digital technology, can eloquence be done without the audience, which is the very target of the speakers?
Change your personality with a digital coach
If people can get into physical shape with apps, why shouldn't it be possible to get into psychological shape in the same way? Sometimes we're psychologically overweight too.
Hosting an effective videoconference, guaranteed!
It's best to have a successful videoconference with the frank impression that everyone is satisfied. Here's a checklist of the main points to look for to achieve this, compiled and synthesized from various sources and authors.
The future of student personal data
For colleges and universities, this open approach to managing student data is a real relief in terms of responsibility for data security.
