Rapid developments in artificial intelligence (AI) have made endless headlines this year thanks to the popular chatbot ChatGPT, search engines releasing AI chatbots of their own, and people using AI to create images, audio and video from scratch.
But some AI companies and experts are increasingly turning their attention to what is known as artificial general intelligence, or AGI — which some expect will herald the arrival of AI systems that are as smart as humans (and eventually even smarter).
The idea brings some exciting possibilities but also some massive risks, so let's take a look at how powerful some people think AGI could become, and whether we might need to stop robots from taking over the world.
What is AGI?
AGI is yet to become a reality, and it still sounds a bit like science fiction.
Basically, it is the concept of AI reaching a level of intelligence on par with or greater than that of humans.
American company OpenAI, which created ChatGPT and the popular AI image generator DALL·E 2, has described AGI as "highly autonomous systems that outperform humans at most economically valuable work".
These systems are expected to solve problems and do complex things while adapting to their environments and improving their knowledge and skills on their own.
There's also the concept of artificial superintelligence (ASI), which imagines systems with much greater intellectual capabilities than humans.
Some experts think these technologies won't be achieved, but others are both excited and concerned about how they could change the world.
Some experts think AGI holds 'extraordinary' risks for humanity
OpenAI chief executive Sam Altman has warned of AGI's "massive risks", including "misuse, drastic accidents, and societal disruption". He has said we could see AGI arrive in the next decade.
"The first AGI will be just a point along the continuum of intelligence. We think it's likely that progress will continue from there, possibly sustaining the rate of progress we've seen over the past decade for a long period of time," he said in a recent blog post.
"If this is true, the world could become extremely different from how it is today, and the risks could be extraordinary. A misaligned superintelligent AGI could cause grievous harm to the world; an autocratic regime with a decisive superintelligence lead could do that too.
"Some people in the AI field think the risks of AGI (and successor systems) are fictitious; we would be delighted if they turn out to be right, but we are going to operate as if these risks are existential."
Last month, OpenAI co-founder Elon Musk, who is no longer affiliated with the company, said AGI had caused him to experience "existential angst".
It wasn't surprising, given the concept of AGI has left experts considering potential negative scenarios, such as:
- A healthcare AGI deciding to not treat elderly people in order to optimise the health of the rest of the population
- An AGI tasked with making money for a corporation which quickly replaces all human workers, leading to mass lay-offs
- An AGI created by a nation state using its abilities to break down other countries' cyber defences
- A defence AGI deciding to take out an entire town in order to kill a high-value target — which sounds like something out of The Terminator
Some have even suggested that AGI could be used to command human soldiers in war.
In a paper published this month, Monash University's Robert Sparrow and the University of Twente's Adam Henschke wrote that because such a war was a possibility, there were ethical concerns about whether AI should be given the authority to send humans to their potential deaths.
"Eventually, the pursuit of victory may require handing over command to machines and victory may be determined by which force has the better AI," they said.
If you are imagining a bleak, dystopian world controlled by dangerously powerful supercomputers and robots right now, you are not alone.
But don't worry just yet — research and planning for AGI is increasing, so that humanity can hopefully avoid negative outcomes.
What are people doing to stop AGI from getting out of hand?
Paul Salmon is a professor of human factors at Queensland's University of the Sunshine Coast. His team is analysing the risks of AGI, and plans to make recommendations to government, developers and researchers later this year.
Professor Salmon agrees that AGI poses an existential risk to humanity, and says we need to act now so that we are prepared for its arrival, even if it is decades away.
"With AGI, if we start to worry about it when it emerges, it's too late," he says.
"The idea is that once AGI is here, we'll quickly get to artificial superintelligence, which basically means we've missed the boat — we will have no opportunity to control this thing.
"We really need to understand how to design AGI so it's safe, ethical and usable. And how to manage the risks and then see how to control them.
"These aren't risks that will arise through malicious use of AGI, or through something like a Terminator scenario. It's more when the AGI becomes prepotent or superintelligent, and in seeking to balance its different goals that we have given it, does things that will be problematic for humans."
In his blog post last month, OpenAI's Sam Altman laid out the company's plans "for AGI and beyond".
He says OpenAI is "becoming increasingly cautious" with releasing its AI models as AGI draws nearer, and says the company has a clause in its charter "about assisting other organizations to advance safety instead of racing with them in late-stage AGI development".
Mr Salmon says there are still questions around the ownership of AGI, given "one of the risks is that the first corporation to develop AGI will quickly have an advantage they can use to generate lots of profit".
Mr Altman says OpenAI, which consists of a non-profit and its for-profit subsidiary, has a cap on the returns its shareholders can earn so that the company is not incentivised "to capture value without bound and risk deploying something potentially catastrophically dangerous".
Australia's minister for industry and science, Ed Husic, says he has asked the country's National Science and Technology Council to consider how AI will evolve, and "to think through what the implications are and how governments should respond".
"We want the technology to work for us — not the other way around," he told the National Press Club this week.
The CSIRO's National AI Centre is also advising government on responsible frameworks for AI.
AI companies are optimistic, despite concerns from some experts
Elon Musk has said that while he has experienced some angst over AGI, he was excited to hopefully witness it come to fruition in his lifetime — and he's not the only tech leader with some optimism.
In his recent blog post, OpenAI's Mr Altman said AGI technology could potentially "give everyone incredible new capabilities" to help with cognitive tasks.
"If AGI is successfully created, this technology could help us elevate humanity by increasing abundance, turbocharging the global economy, and aiding in the discovery of new scientific knowledge that changes the limits of possibility," he said.
"We can imagine a world in which humanity flourishes to a degree that is probably impossible for any of us to fully visualise yet."
A few days after Mr Altman's blog post, an attorney from the US Federal Trade Commission — which works to protect consumers and competition — wrote in their own blog post that AI was "a marketing term", and warned companies to be cautious about how they used it.
"We're not yet living in the realm of science fiction, where computers can generally make trustworthy predictions of human behavior," they wrote.
Microsoft co-founder Bill Gates wrote in a blog post this week that "none of the breakthroughs of the past few months have moved us substantially closer" to creating AGI, and pointed out that developments in AI brought both "opportunities and responsibilities".
Professor Salmon said he remained optimistic that if society prepared well for AGI, it could have hugely positive impacts.
"If we take a proactive approach to designing the controls, there's no doubt that the benefits to humanity can be absolutely massive," he said.
Mr Altman appears to agree, but it is clear that creating and managing AGI is a huge undertaking.
"Successfully transitioning to a world with superintelligence is perhaps the most important — and hopeful, and scary — project in human history," he said.
"Success is far from guaranteed, and the stakes (boundless downside and boundless upside) will hopefully unite all of us."
