Super AI: a future in which AI surpasses human thinking

Androids ‘Star Wars’ They have the ability to understand people and develop their own desires. Personal assistant ‘Her’ He demonstrates such developed emotional abilities that he even falls in love. Computer HAL “2001: A Space Odyssey” he rebels against his crew to take control of the spaceship. These are examples from science fiction, but they represent the best examples of technologies that may become a reality in the future: artificial superintelligence.

Artificial superintelligence is a hypothetical artificial intelligence system that could “intellectual transcendence of human intellect”, according to the IBM article. This tool will have advanced cognitive functions and highly advanced computational capabilities, allowing it to solve complex problems faced by fields such as healthcare, finance or scientific research.

The specific benefits that artificial superintelligence will bring are not yet known, but it would be equivalent to having a near-perfect supercomputer available 24 hours a day. So I could analyze and reason about any amount of data at a speed that is still unknown. “It will be ideal for tasks such as safely driving self-driving cars or assisting during space exploration. In addition, its ability to analyze data can lead to solutions that people cannot even imagine, such as a higher quality of life or longer lifespan,” they note in the IBM paper.

Thus, according to OpenAI, “artificial superintelligence will be more powerful than other technologies that humanity has encountered in the past.” That is, artificial superintelligence would help humanity to be more productive in almost everything. And because of its hypothetical reasoning ability, experts are already discussing its development and the implications of integrating AI with these characteristics into everyday life.

Different types of AI and their capabilities

Artificial intelligence technologies are classified by their ability to mimic human characteristics. The world is currently in the first of three proposed stages for the development of AI, but the foundations are already being laid for the next stages, such as those carried out by artificial superintelligence, and its consequences for humanity. Now, it should be noted that there is still open debate about each of the definitions.

• Artificial Narrow Intelligence (ANI)– It is programmed to perform specific tasks and is currently the only type of AI that exists. It is trained to perform a specific task, often at faster-than-human speed, but is unable to function beyond its assigned function. It’s also called “weak AI” and includes everything from virtual assistants like Siri or Alexa to predictive generative language models like OpenAI’s ChatGPT.
• Artificial General Intelligence (AGI): Currently a theoretical concept that refers to a state that has not yet been achieved, it refers to an AI capable of learning and acquiring new skills beyond the function for which it was programmed. It is also called “strong AI” and with it, human intervention will no longer be required in its training as it will be able to perform almost any task autonomously.
• Artificial superintelligence (ASI): Although the previous stage related to artificial general intelligence must be successfully developed first, this will be the next step. In this case, AI will surpass human cognitive abilities. It’s a theoretical concept for now, but if it materializes, it will be capable of thinking, reasoning, and making cognitive judgments beyond human capabilities. Plus, you can even develop your own emotions, needs and desires.

Artificial superintelligence debate

Although artificial superintelligence is still a theoretical concept more typical of science fiction, various technology companies, researchers and experts are already warning about its future potential. The answer oscillates between admiration for the achievements and fear that algorithms may pose some danger to people.

Shortly after the release of ChatGPT Elon Musk, Steve Wozniak (co-founder of Apple) and Jaan Tallinn (co-founder of Skype) published an open letter along with more than 1,000 researchers asking for research into advanced AI to be suspended because it is a “grave risk to humanity.” Goes in the same spirit Eliezer Yudkowskyhead of the Machine Intelligence Research Institute and considered one of the greatest experts in the field of artificial intelligence, who published an article in Time magazine in March 2023 in which he called completely stop the development of these achievements: “We cannot calculate in advance what will happen and when, and at present it seems possible to imagine that a research laboratory could cross critical lines without realizing it.”

On the other hand, other experts such as an investor Marc Andreessenco-founder of web browsers such as Mosaic and Netscape, They don’t think AI is a danger to humanity. “The AI ​​doesn’t want anything, it doesn’t have goals, it doesn’t want to kill you because it’s not alive. And AI is a machine: it has no more chance of coming to life than your toaster,” he wrote on his personal blog. Andrew Ng, the predecessor of Google Brain and director of the Stanford Artificial Intelligence Laboratory (SAIL), also does not think that AI will pose a threat: “I think the impact of AI is having a huge impact on society, and I don’t see how it could lead to human extinction. Working in the field of artificial intelligence, I feel a great ethical responsibility and keep an open mind to understand the risks involved.”

Another great guide to predicting the future of AI is Ilya Sutskever, co-founder of OpenAI and former chief scientific officer who left the company to focus on a personal project. “At some point we will have true artificial general intelligence (AGI). Perhaps OpenAI will create it. Maybe some other company will do it,” Sutzkever said in an interview published in MIT Technology Review.

In fact, OpenAI has created a safety and security committee responsible for overseeing and advising on the company’s most important decisions. While OpenAI currently occupies a prominent position in the artificial intelligence space, the potential development of more advanced artificial intelligence is not unique to the company. Other tech giants such as Amazon, Google, Meta and Microsoft They also have research departments studying the possibility of creating AGI type AI, that is, artificial general intelligence.

While artificial general intelligence is developing, other experts are already thinking about what will happen when the milestone of artificial superintelligence is reached. But when will it come? There is still no clear date. OpenAI notes that it could arrive “this decade.” Bye, Bill Gates He is less optimistic, noting in his blog that this achievement “could happen in a decade or a century.” However, the Microsoft founder identifies three questions that he believes will become more relevant over time: “What will be the goals of these AIs?” What happens if their interests collide with the interests of humanity? Should we prevent the development of strong AI?”

Meanwhile, other experts such as Yann LeCun, chief artificial intelligence scientist at Meta, believes there are still many steps to be taken before we talk about machines being able to outperform humans. “The emergence of superintelligence will not be an event. Today we have nothing like a model of superintelligent systems. At some point, we will design an architecture that will help us achieve this goal (…) We will design it in such a way that its sole purpose is to achieve human-defined goals (I call this goal-oriented AI).” This is stated on the social networkwhere he denied the possibility of creating a super-powerful system typical of science fiction.

See more: Pope Francis makes historic appearance at G7 to talk about artificial intelligence

See also: Artificial Intelligence and Data: Bots’ Crucial Impulse in Customer Service

Read more: Meta AI will not appear in Europe at the moment