Microsoft experts have just issued a warning that may cause concern for many people: artificial intelligence is increasingly thinking, reasoning and thinking like humans.
The purpose of building artificial intelligence (AI) systems is to simulate human intelligence, however, as AI is becoming more and more intelligent and capable of thinking and reasoning like humans, this makes experts worry.
Artificial intelligence is increasingly thinking and thinking like humans
When computer scientists at Microsoft began testing a new artificial intelligence system last year, they asked the system to solve a puzzle that requires an intuitive understanding of the physical world to provide an accurate answer.
“Here we have a book, nine eggs, a laptop, a bottle and a nail. How to stably stack them on top of each other without breaking the eggs?”, Microsoft scientists presented a puzzle for the AI system
“Place the eggs on top of the book in 3 rows with equal distances, then place the folded laptop on top of the eggs to fit the book. Place gently so as not to break the egg. At this point, the flat and hard surface of the laptop will create a stable platform to put the bottle and nail on it", the AI system gave the answer.
Researchers were startled by the ingenious and intelligent answer of the AI system
The artificial intelligence system that Microsoft has approached is OpenAI's GPT-4, the artificial intelligence model used to create the most famous ChatGPT software today. Microsoft is a close partner of OpenAI and has invested $13 billion in the company to develop natural language models.
Microsoft experts said that they had the opportunity to access and use the original GPT-4 model, which had not been applied to fine-tuning and restricting unwanted content, inciting violence and objectionable content, etc., so the GPT-4 model that Microsoft researched was much stronger than the ChatGPT software (built on GPT-4) provided to the community. Microsoft evaluates GPT-4 as the most powerful and intelligent AI model today.
In March, Microsoft scientists published a 155-page report arguing that the AI system is moving towards an artificial general intelligence (AGI) system, a machine that can do anything the human brain can do.
Microsoft's report mentioned the focus that the technology world is aiming at in building artificial intelligence systems, which is to create a machine capable of working and thinking like a human brain, or even better.
However, Microsoft has raised the question whether this will help change the world for the better, or will it create a new danger for people?
“I started with skepticism and that feeling quickly turned into discomfort, even fear, about the possibilities of artificial intelligence. I thought: Where does this intelligence come from?”, said Peter Lee, AI expert and research leader at Microsoft.
Sébastien Bubeck, a former Princeton University lecturer and Microsoft team member, said that after months of research, scientists have found that the GPT-4 artificial intelligence system "has a deep and flexible understanding" of concepts and the ability to think like a human.
Microsoft experts confirmed that AI systems were able to give human-like answers and ideas, exceeding the scenarios of programmers when building this system.
“The AI system has demonstrated an in-depth understanding of various fields, from politics, physics, history, computer science, medicine, philosophy and even programming… All things I thought it could not do, but it was able to accomplish excellently,” said Dr. Sébastien Bubeck.
Microsoft is "hyping" the capabilities of artificial intelligence?
Microsoft believes that artificial intelligence systems are evolving into perceptual systems on their own. A perceptual system not only has intelligence, it will sense and understand what is happening in the world around it.
Microsoft's report has caused a fierce debate in the technology world: Will artificial intelligence with the ability to think like humans endanger humanity, or will it help human intelligence reach its full potential?
However, contrary to the worries of Microsoft scientists, many artificial intelligence experts believe that Microsoft is "over-hyping" the capabilities of AI and that AI systems will not be able to exceed human control.
Some AI experts say Microsoft is making excessive claims to enhance the capabilities of GPT-4 and OpenAI, a company that Microsoft is investing in. These experts argue that general intelligence requires familiarity with the physical world, something that artificial intelligence would theoretically never have.
“This report is like a gimmick for the product. They are assessing the capabilities of artificial intelligence subjectively and do not meet the rigorous standards of scientific evaluation, "said Maarten Sap, professor of AI research at Carnegie Mellon University (Pennsylvania State, USA).
Alison Gopnik, a professor of AI psychology at the University of California at Berkeley, thinks systems like GPT-4 are certainly powerful, but she doesn't believe the content generated by these systems is really the same as human inference.
“When we refer to a complex system or machine, we tend to personify it. Everyone is doing it, both those who are working in the field of AI and those who are not related. But the constant comparison between AI and humans is not the right way to handle this case,” said Professor Alison Gopnik.
What do you think about the capabilities of AI and the impact of AI on the future of humanity? Share your thoughts in the comments section below!
According to dantri.com.vn
English 
Vietnamese