AI will soon gain consciousness and may increase the ‘probability of disaster’ if owned by billionaires, experts warn

In the words of Stuart J. Russell, a British computer scientist at the forefront of the artificial intelligence revolution: “Nobody has any idea how to build a conscious machine.”

But just because we don’t know how to build conscious machines doesn’t mean there won’t be conscious machines.

Some artificial intelligence researchers have suggested that conscious AI already exists.

Last year, Blake Lemoine, a former engineer at Google, claimed that LaMDA AI, a collection of advanced, conversational Large Language Models (LLMs), was conscious.

His former employers quickly responded and denied the claims.

But according to a recent paper published by 19 researchers from a variety of disciplines, including philosophy and computer science, conscious AI is on the rise.

Dr. Avi Loeb, the director of the Harvard Institute for Theory & Computation, told The US Sun: “Knowing whether AI systems are conscious or not is neither important nor possible.”

“The only conscious person I know for sure is myself.

“I would treat AI systems with the same respect as humans if they appeared to be as intelligent as humans.”

Dr. Owen Holland, an expert in the field of machine consciousness, believes Loeb’s dismissal of the importance of conscious AI is unwise at best.

Holland, a professor emeritus of computer science at the University of Sussex, UK, believes AI can achieve consciousness.

However, to understand the “how,” he emphasizes, “we need to explain a little more.”

First he asks: “Is artificial consciousness possible?”

He believes this because “conscious people seem to be entirely material systems – no magic or fairy dust anywhere that I can see.”

Next we must ask: Can artificial consciousness be built on a digital substrate or does it have to be analog, like neural tissue?

Holland relies on digital technology because people “can digitally model analog systems to any level of precision desired.”

As for the idea that virtual consciousness is generated “in simulated agents in simulated worlds, i.e. in running software,” the expert also believes this is possible. Why? “Because I cannot see anything in consciousness that requires a material body or a material world.”

Holland firmly believes that “traditional digital IT will very likely be able to support consciousness.”

Although some aspects of consciousness include obvious functional components, the researcher emphasizes that “both intelligence and consciousness depend heavily on the formation and manipulation of models.”

For this reason, Holland considers it quite credible that a real or, more likely, virtual, self-modifying system capable of creating new types of models and novel interactions of models, in an appropriate environment and under appropriate contingencies can develop both intelligence and consciousness, just as evolution did by exploiting the capabilities of nervous tissue.

“I don’t think AI alone will ‘achieve’ consciousness,” he said.

However, he adds: “I believe that the search for artificial consciousness will succeed.”

Holland, someone who has been studying machines and consciousness for decades, is deeply troubled by the idea that “consciousness has everything to do with internal processes and representations.”

For many, sentience is synonymous with consciousness.

For Holland, sentience is “roughly synonymous with phenomenal consciousness.”

However, he adds, “Many people work on consciousness from a functional rather than an experiential perspective, and that’s just an inferior kind of sentience – being aware of something in the sense of taking it into account, more like the military term situational awareness.” “

Sentience involves the ability to perceive certain feelings and emotions.

Consciousness, on the other hand, is best described as heightened sensation; it is consciousness on a much deeper level.

A conscious person not only perceives; analyze and evaluate them.

Scientists still don’t even know how and why consciousness arises in humans.

There is no unified theory of consciousness. It may never exist. The idea of ​​consciousness is that complex.

Some experts, like philosopher David Chalmers, an academic who has studied consciousness for decades and written best-selling books on the subject, argue that while science can explain why certain neurons fire, it may never be able to , to explain the subjective experiences of consciousness, known as qualia.

Qualia includes the perceived pain of a migraine and the perceived beauty of a sunset.

Lemoine told The US Sun: “We can tell that an AI is truly conscious, just as we can tell if a human (or other animal) is conscious.”

He says, “You watch its behavior and look for things it does that it would be incapable of if it were unconscious.”

When it comes to the idea of ​​advanced AI, Holland’s fears are justified.

The engineer is particularly concerned about the “possibility of abuse by people or of use in subsequent areas by people who do not understand its weaknesses.”

He compares advanced AI to nuclear power, which can “reduce the likelihood of inconvenience but increase the likelihood of disaster.”

“The question of whether or not you should be worried,” says Lemoine.

“I do not believe that the presence or absence of consciousness is inherently good or bad. It makes the distribution more extreme (could be very good or very bad), but consciousness does not necessarily shift the balance one way or the other.”

Instead, he adds, “What should worry you is the way this new power is heavily concentrated in the hands of a few billionaires.”

“It appears that most of them are not particularly benevolent people.”