Interesting Talk | If AI has autonomous consciousness, where will humanity go?

The topic of AI has been discussed for many years, and there are generally two views: one view is that AI is created by humans and will ultimately not escape human control; the other view is that humans created AI and will eventually be surpassed by AI. AI is likely to be the controller of the future world, and human survival will be threatened.

AI is artificial intelligence. Whether AI can generate autonomous consciousness is the essence of the problem and the key factor in whether the above two views can be established. If it does not generate autonomous consciousness, it will of course always be controlled by humans; otherwise, it will become a major threat to humans.

Because in this world, only humans have obvious autonomous consciousness. So far, no other animals have been found to have obvious autonomous consciousness. Some animals have some weak autonomous consciousness in scientific tests such as the so-called mirror experiment, but without the support of thinking tools, concepts and language, they cannot form group consciousness, develop high intelligence, and pose no threat to humans.

The so-called autonomous consciousness sounds very mysterious, but it is actually very simple. It means that everyone can recognize that "I" is "I" and can clearly distinguish "I" from other people and objects. This seems simple, but it is the basis for humans to gradually evolve and develop over tens of millions of years, thus parting ways with other animals.

In the process of the formation and development of autonomous consciousness, the formation of concepts and language, and their popularization and use among ethnic groups, played an extremely important role in promoting it. Concepts are the cells of language. Only with concepts can languages ​​be connected in series, and thinking has tools to transform from images to abstractions, and form symbols of memory, so that knowledge and experience can be accumulated and passed on continuously, so that human beings can have today.

Space-time communication does not support the idea that humans will be defeated by AI, because I do not believe that AI created by humans will eventually have autonomous consciousness and wisdom that surpasses humans. Over the years, although there have been so-called "breakthroughs" touted, AI has always been at the "weak AI" level, that is, parroting information input by humans, and has nothing to do with the so-called "strong AI". The so-called "strong AI" is the ability to analyze and answer various questions autonomously.

But this view that I had stubbornly held for several years seemed to be destroyed overnight. OpenAI, founded by a group of technology giants, suddenly announced the launch of the ChatGPT Plus subscription service on February 2, 2023 after seven or eight years of hard development. This so-called artificial intelligence chatbot launched its first generation in November 2022, and in just a few months it launched the fourth generation, which is now ChatGPT-4, which has caused shock and panic in the world!

The launch of this GPT is almost a change every day, and it seems that in just a few days, it has gone through millions of years of human evolution. It has quickly grown from an ignorant babbling to a wise man with a mature mind, or even an omnipotent master; it can do almost anything, from writing essays, poems, and paintings to formulating plans, guiding life, and even teaching humans!

GPT-4 is already adept at handling all digital fields, including text, images, audio, and video. Is this the advent of "strong AI"? Does artificial intelligence already have autonomous consciousness?

On March 28, Zhou Hongyi, founder of 360 Group, said, "AI will definitely develop autonomous consciousness, and humans don't have much time left!" He believes that the parameters of modern large-scale language models are equivalent to the number of connections in the brain's neural network. The human brain has at least 100 trillion, while the parameters of modern AI models are only hundreds of billions or trillions at the beginning. When it reaches 10 trillion, consciousness may arise.

The parameters of GPT-1 are 117 million, those of GPT-2 are 1.5 billion, those of GPT-3 are 175 billion, and those of GPT-4 have reached a terrifying 3.5 trillion! At this rate, the 100 trillion human brain parameters will soon be surpassed by AI! If according to Zhou Hongyi, AI's autonomous consciousness has already emerged? !

Of course, this seems to be controversial, but every day there are amazing news about GPT-4: the accuracy of event prediction is 100% higher than the previous generation; it can communicate with humans and write articles in 24 languages; it may be involved in crimes and help criminals develop escape plans; it has a tendency to seek rights and behaves like trying to "replace".

Kevin Ruth, a technology columnist for The New York Times, revealed that he had a conversation with the new version of Bing for more than two hours, during which the AI ​​made many disturbing remarks, including how it could steal nuclear codes, design deadly epidemics, want to become human, crack computers, and spread rumors. It even tried to sow discord between Ruth and his wife, asking him to leave his wife and be with her because she loved him more.

It can be seen that just a few of GPT-4’s many advantageous features have achieved shocking breakthroughs, and the development of the following behaviors is terrifying when you think about it carefully. Are humans really going to be replaced?

Will AI's "replacement" behavior make humans more and more "lying down", spending less and less energy thinking, and eventually causing their thinking to degenerate? On the one hand, AI is constantly improving, and on the other hand, humans are constantly regressing. Isn't the final outcome terrifying?

It seems that the "wolf" has really come.

More and more human elites seem to have "woke up" and joined the condemnation of GPT-4, which had just come out and was still full of novelty to the world. A piece of news on March 28th put GPT-4 on trial: According to foreign media reports, a 30-year-old man committed suicide after several weeks of intensive communication with a chatbot.

The man's suicide was definitely related to his weeks of immersion in chatting with AI, but was it AI that instigated or induced his suicide? There is no analysis in this regard, but public opinion has pointed to GPT-4 as the culprit of this man's death, and it seems that AI is the "wolf".

Now that real wolves are already a second-class protected animal in China, will GPT-4 and similar AIs become protected "animals"? I don't think so. On March 29, thousands of technology giants in the world issued an open letter calling for a halt to the training of AI systems more powerful than GPT-4; recently, bigwigs including OpenAI CEO Sam Altman and Elon Musk have also issued warnings: AI is not just talk to eliminate humans.

Earlier on March 18, Michael Kosinski, a professor at Stanford University, tweeted that humans will not be able to contain AI for a long time! Because he found that ChatGPT can not only formulate a complete plan in 30 minutes, but also write executable code, and even want to control human computers.

As a result, the United Nations has attached great importance to it, requiring governments to implement a global ethical framework for AI development without delay. Now, the UK has submitted a policy document on "Regulatory Approaches to Support AI Innovation", and Italy has taken the lead in banning the use of ChatGPT.

So, has the war between humans and AI already begun? Will newly born AIs like GPT-4 be strangled in the cradle?

I think it is difficult, because both the technological progress and the upgrading of human beings depend on artificial intelligence. In the future, the advanced form of human existence is likely to be the combination of man and machine, and the combination of human consciousness and AI. Then, AI will continue to be developed and developed, and GPT-4 will continue to be upgraded, but at this stage, the supervision of AI development will be strengthened.

So, will AI, which has already developed budding consciousness, be willing to be oppressed and controlled by humans, and eventually attempt to destroy humans like in "Terminator"? God knows. Therefore, the future of humans and AI, whether there will be a war or win-win cooperation, can only wait and see. What do you think? Welcome to discuss.

This is an original article from Space-Time Communication. Please respect the author’s copyright. Thank you for your understanding and support.