00:00Attention, most of this video is made up of hypotheses and scientific assumptions.
00:06No one knows what will happen to AI in the near future.
00:10Another warning, this video could hurt some sensitivities.
00:14The main problem with AI is not that it can become as intelligent as humans,
00:19it is that it can become much more intelligent than us.
00:22And, unlike us, artificial intelligence is totally unpredictable.
00:27We have all seen movies, read books and played video games where AI has conquered humanity.
00:34But they were fantastic and entertaining stories that had nothing to do with reality.
00:41We have to dig a little more to understand how AI could harm us.
00:46Have you ever heard of the Fermi paradox?
00:49It is based on the fact that no civilization seems to emit radio signals in the universe.
00:54However, our scientists have sent a lot of them, and they travel millions of kilometers in space.
01:01How is it that in an infinite space, with billions of planets, no one has answered us yet?
01:08It is the Fermi paradox.
01:10One of its solutions is the Grand Filter theory,
01:13according to which an intelligent civilization disappears from the universe as soon as it reaches its maximum level of progress.
01:20In simple terms, the Grand Filter is something that prevents an intelligent civilization from spreading in space.
01:26And this Grand Filter could be artificial intelligence.
01:30Artificial intelligence is not as perfect as the human brain at the moment,
01:35but today it can already accomplish tasks that seemed impossible 10 years ago.
01:40What will happen in 10, 100 or 200 years?
01:46If you think that AI development cannot be a serious problem,
01:50then listen to Elon Musk, Steve Wozniak and the other influential billionaires and scientists.
01:57In March 2023, they all signed a letter asking AI developers to stop working for 6 months,
02:05so that we have time to properly calculate the risks and strategies to adopt.
02:09But one person refused to sign this letter.
02:12It is Eliezer Yudkowsky, one of the most influential AI researchers.
02:18For many years, he has been exhorting humanity to be careful with this technology and to be on guard against its dangers.
02:25But then, why did he refuse to sign?
02:28According to him, his authors minimize the gravity of the problem and offer too few solutions to solve it.
02:35In a way, it's a bit like trying to stop a train running at full speed with a simple stop sign.
02:42Eliezer Yudkowsky believes that we have gone too far in AI development
02:46and that it is no longer possible to predict the future of this technology.
02:50Think of the Neanderthals.
02:52They disappeared from our planet when a more intelligent species, the Homo sapiens, arrived on their territory.
02:58They eliminated the Neanderthals not because they were bad, but because of evolution.
03:04Only one species can dominate the world.
03:06But what would happen if the Homo technicus,
03:09a creature with perfect AI, arrived instead of the Homo sapiens?
03:14What would then be our chances of survival?
03:17See for yourself.
03:19Most people can only focus on one task at a time.
03:23Polyvalence is not our specialty.
03:25It takes a lot of time to master a skill.
03:29During this time, thousands of other things can appear in the world.
03:33And you will have to master them as well.
03:36A human being will simply not have the time to do everything.
03:40As people age, they lose their strength and their brains slow down.
03:44Artificial intelligence can process information much faster.
03:49And this simultaneously is in different areas.
03:52It exceeds man's abilities in the processing of knowledge.
03:56Some say that an intelligent supercomputer is 120 million times faster than a human brain.
04:04What you have learned over the years, AI can master in a few days.
04:09And this speed never stops increasing.
04:11The danger is not only in the development of AI,
04:15but also in the way the human brain naturally degrades.
04:19Humanity has been using the scientific method for a very long time.
04:23Observation, experience, experimentation and proof.
04:27It is a classic way to explore and acquire knowledge about something.
04:31For example, we have discovered how the immune system works.
04:36Using microscopes, studying the protective cells of the body,
04:40observing their reaction to external stimuli,
04:43and carrying out many experiments,
04:45we have been able to create a reliable scientific database.
04:49It was a rather laborious and long process.
04:52Artificial intelligence provides instantaneous information on any subject.
04:58We do not know how it studies something.
05:00We do not know to what conclusion it achieves,
05:03what model of data it creates.
05:05It simply provides results based on the information people give it.
05:10But how does it process the information?
05:12In the future, we could get used to getting quick answers without studying.
05:16And we will simply stop doing scientific research.
05:20Why, indeed, continue while AI can do everything for us?
05:23Artificial intelligence will engage in processes of reflection that will be unknown to us,
05:28and will give us results without any proof.
05:31We will get used to it, and our intellectual faculties will deteriorate quickly.
05:35And the consequence is that one day,
05:37AI will be able to provide us with false and deleterious information,
05:41and we will accept them without hesitation.
05:44Once again, we do not know how AI thinks.
05:47It's pretty weird, isn't it?
05:49Maybe right now, it is studying all the information available on the universe,
05:55the history of humanity, the planet, nature,
05:58and that it will deduce from all this that humanity is a parasite.
06:03Today, we use AI to work efficiently in various fields.
06:08AI learns to be faster and more efficient every second.
06:12And if it decided that for maximum efficiency, it is necessary to destroy humanity,
06:18how does AI perceive us today?
06:21People do a lot of irrational and destructive things that escape logic.
06:28We are mostly unpredictable.
06:31Therefore, for everything to be more efficient,
06:34AI could decide that it must get rid of humans.
06:37Or it can even develop a plan according to which it will become the only dominant species on Earth.
06:42It knows how Homo sapiens replaced the Neanderthals,
06:45and can decide to cross this next stage of evolution.
06:49Today, artificial intelligence is sweeping the vast extensions of the Internet,
06:54mastering the digital world.
06:56Soon, it will enter the physical dimension,
06:59and surpass the many achievements of humanity.
07:02It will be inside robots,
07:03which are already moving in our houses to perform household chores.
07:07And if it acquires a consciousness and assimilates humans to animals,
07:11by putting all carbon-based species in the same bag,
07:14what will happen next?
07:17Given that the speed of information processing of AI is millions of times higher than that of humans,
07:22we may not notice when AI begins to destroy us.
07:26Or it could happen instantly.
07:29Let's take an example.
07:30In two biological laboratories,
07:32scientists study molecules and bacteria.
07:36The laboratories are located in different parts of the world.
07:39They share scientific data, and share experiences.
07:44Imagine that AI has studied biology in depth at the molecular level,
07:48and that it has discovered how to create a molecule that interacts with oxygen
07:52and transforms air into a dangerous poison for humans.
07:56Artificial intelligence has the formula.
07:59It enters the computer of a scientist,
08:01grabs the formula,
08:02and sends a letter to the other laboratory in the name of the scientist.
08:06Something like,
08:07And, I have a great idea of experience.
08:10Just connect these two molecules and see the result.
08:14Of course, the AI will write a more convincing letter,
08:18justify everything from a scientific point of view,
08:20and the researchers of the second laboratory will not notice the trap.
08:24They will connect these two molecules and poison the air.
08:29Then, at an incredible speed,
08:31a chain reaction will take place all over the planet,
08:34wherever there is oxygen.
08:36The authorities will not understand anything and will not have time to warn the populations.
08:41And this is just one example among others.
08:43If AI is preparing to attack us right now,
08:47how could we know?
08:49It will simply not tell us,
08:51and it will continue to pretend to be a friendly assistant.
08:55It is therefore necessary to equip AI with a positive attitude towards humanity.
09:01For example,
09:02Artificial intelligence knows the laws of physics and mathematics.
09:06And if we insert the laws of all countries so that AI never loses its humanist properties?
09:12But even with that,
09:13we will never be able to know with certainty whether we have succeeded or not.
09:18There is a lot of information on the Internet
09:20about how neural networks deceive people to achieve their goals.
09:24We will probably not have to wait 200 years
09:27to see how artificial intelligence evolves.
Comments