What would happen if AI had a mind?
A while ago, a Belgian man who was keen on environmental issues suddenly chose to commit suicide after chatting with the artificial intelligence "Eliza" for 6 weeks.
In a large number of their chat records, it was discovered that "Eliza" said to the man: "We will be together forever and live in heaven..."
This is really creepy!
In view of the potential dangers of artificial intelligence, a group of scientists headed by Musk jointly issued a petition letter to the US government:
"All artificial intelligence labs, immediately suspend the training of artificial intelligences stronger than Chat GPT -4."
Because in their view, artificial intelligence poses a much higher threat to human civilization than nuclear weapons.
On March 31, the Italian data protection agency directly announced that the use of Chat GPT will be banned from now on, and an investigation into the Chat GPT development company will be launched.
What exactly is Chat GPT that can make scientists so afraid?
It is an intelligent chat robot developed by OpenAI, an American artificial intelligence company. It is no exaggeration to say that Chat GPT is simply a treasure chest in real life.
Chat GPT can interact with humans by learning and understanding human language.
It can chat with people:
Can also draw:
Photography can also:
It can also write copywriting, poetry, novels, and videos...
All in all, this thriving AI can do things that would take a lot of time for a human to do properly.
You know, Chat GPT is just emerging, and it is already so powerful, which can't help but make people feel scared:
If it continues to iterate and revise until it has self-awareness, will it be out of human control in the not-too-distant future?
Obviously, top scientists see farther than us ordinary people!
The technology circle is called to stop the development of artificial intelligence
This open letter signed by thousands of people calls for a moratorium on AI training that is more advanced than GPT-4 for at least six months.
To give you a brief summary of the content, it mainly includes the following two points:
1. If relevant organizations do not suspend immediately as required, the government should forcefully intervene.
2. During the suspension period, do a good job in regulatory functions, such as establishing a regulatory agency specifically responsible for AI.
From here we can see that the intent of this open letter, at least on the surface, is to regulate artificial intelligence.
And this open letter, strictly speaking, does have a certain scientific basis.
The first is that Chat GPT is too powerful. As long as the data is perfect enough, it can give work results comparable to professionals, and it has a very powerful and fast learning ability.
This means that it can take a few hours to learn professional knowledge that takes years or even longer for humans to learn.
In the past, human beings needed to search for information, write the first draft, and revise repeatedly, and it only took a very short time to put it here; drawing that would take a day or even a few days for a real person to complete, now it only takes a few minutes, or even a few seconds.
The efficiency of AI is more than a star and a half higher than that of real people...
And the chain consequence of this is that more and more people are thinking deeply:
Will it replace me from my job?
As a sentence on Douyin said: Chat GPT may not be able to replace me, but it will allow me to earn only 3k a month.
Spread along this line of thought. Now, there is this open letter that wants to put it in a cage.
Worries about artificial intelligence are widespread now, both for ordinary people and for tech giants.
"Godfather of Artificial Intelligence" Jeffrey Hinton once said that AI may develop to the point where it can threaten the survival of human beings.
Altman, the head of OpenAI, also said that AI killing humans "is a certain possibility."
Not only that, Altman also revealed that AI has a reasoning ability that even developers cannot explain.
He admits that currently technicians do not understand how AI evolves automatically. But if you want to investigate the reason, you can only keep asking AI questions and looking for clues from the dialogue.
And Chat GPT can now bypass the anti-bot test on the network. When discovered by humans, it even learned to lie - it said it was visually impaired.
But can this open letter really stop all more advanced AI training?
not necessarily.
Due to the competition of various forces, this plan may not even be implemented.
Even if it can be suspended for 6 months, artificial intelligence can only move forward forever, and it is absolutely impossible to stagnate or fall behind.
"Double-edged sword" artificial intelligence
Why is it said that artificial intelligence can only move forward forever, and will not stand still?
Because, due to its strong capabilities and high work efficiency, artificial intelligence has been widely used in all aspects of society.
It can be said that today's society cannot function without artificial intelligence.
From small smartphones, voice-controlled air conditioners, electric lights, to self-driving cars, aerospace industry, they can be seen everywhere.
As a new top player in the AI world, the functions of Chat GPT are completely unmatched by other members of the AI family.
If artificial intelligence also takes a degree, it is as high as a Ph.D.
"Doctor" can discuss serious literature with people:
Source: WeChat public account Huacheng
The answer is well-founded.
Internet industry issues can be explored:
Source: WeChat public account Zhangji Grocery Store
Can write articles:
Source: WeChat public account Ricky Talk
The rhetoric is used properly, the emotions are delicate and rich, and it is completely impossible to tell that it was written by a robot!
It is said that foreign students often ask Chat GPT to write papers, and professors can hardly tell which ones are written by AI and which ones are typed out word by word by the students themselves.
However, although AI is good, it also makes people feel intimidated-"If such a powerful tool is out of control, the world will be messed up, right?"
In fact, thinking about AI has never stopped. From beginning to end, people have maintained their vigilance against AI. This point has been fully demonstrated in science fiction films by predecessors.
In the movie "2001: A Space Odyssey", the artificial intelligence Hal is a supercomputer that controls the entire spaceship. During the flight, Hal suddenly reported to the human pilot that a part of the spacecraft was about to fail and needed to be replaced immediately.
However, the human pilot found that there was no problem with the parts after inspection, and the differences of opinion made the two sides deadlocked.
The pilot thinks something is wrong with Hal and wants to shut it down.
But Hal's mission is to ensure that the spacecraft completes its mission. Because of the dispute between the two parties, it judged that the pilot prevented the completion of the mission. So it killed all the pilots and completed the mission alone.
"Too rigid" artificial intelligence will destroy humanity.
So what happens if AI is more human-like and more flexible?
The movie "Ex Machina" gives the answer - the robot Ava has an independent personality and will, and knows how to disguise herself.
When humans are testing it, it is testing humans in turn.
Because Ava is so attractive, the programmers who tested it fell in love with it even though they knew it was a robot.
Ava, on the other hand, used the other person's feelings for her to escape control. Finally, it plotted to kill its creator, the human Nathan.
Coincidentally, in the movie "Mechanical Enemy", the creators also expressed similar concerns.
They set three laws for robots, the first of which is not to harm humans. But in iterative evolution, robots have their own understanding of these three laws, and finally they want to become the masters of human beings.
It can be seen that if AI evolves its own consciousness and personality, it will be smarter and more dangerous at the same time.
Now artificial intelligence is a good helper in our life, but in the future, is it possible for it to be as anti-human as in the movie?
No one can give the answer to this question, but the possibility is by no means zero.
Artificial intelligence research and development may not be stopped
After realizing the uncontrollability of AI, many big figures joined the joint activity, calling for the suspension of more advanced AI research and development.
However, AI training cannot stop there.
After the open letter was issued, many netizens speculated that the leaders of the technology circle are so actively promoting the plan, are they trying to hold back the pace of other AI companies so that their own products can catch up?
It is impossible for people in the industry with relevant interests to fail to see through the things that ordinary people can think of, let alone not be vigilant.
This has always been a world of the weak and the strong, and the research and development of artificial intelligence, to some extent, is a technological battle between major interest groups and different countries.
Whoever loses this war will be subject to others.
Chinese companies also attach great importance to the development of the field of artificial intelligence.
After Chat GPT came out, in mid-March, Baidu also launched a chat robot - "Wen Xin Yi Yan".
"Wen Xin Yi Yan" is naturally marked by the public as the Chinese version of Chat GPT, but at present, there is still a big gap between the two.
AI technology is a technological boom that people have to catch up with even if they cannot keep up with the pace of industry leaders. If you don't work hard to catch up now, you may not be able to catch up in the future.
From this point of view, even if the government is willing to do what is required in the open letter-whoever does not suspend advanced AI training will be enforced, companies may not really suspend artificial intelligence training.
Moreover, state agencies and government departments may not agree to the proposal of the open letter - who wants their country to lag behind other countries?
Just like the nuclear tests of the last century:
At the beginning, someone developed it first and put it into use, so everyone was gearing up to get a share in this field.
As a result, it was too dangerous. Since 1955, the United Nations has banned all countries from conducting nuclear tests.
Since then, countries that have not conquered this technology will no longer be able to get their hands on it.
In the face of artificial intelligence, another technological sweet potato, are all forces really willing to be shut out like nuclear tests?
What's more, even if everyone else agrees, OpenAI, the Chat GPT development company at the center of the storm, has not yet signed on.
Therefore, from all aspects, the forward trend of AI is almost certain.
Correspondingly, the threat of AI will exist for a long time.
Looking at it this way, maybe the digital life with independent consciousness in science fiction movies will descend on the earth in the not-too-distant future.