By Trent Griger
When will a computer think like a human? When will it have ideas of its own, better itself, and surpass us? This event is coming, whether we like it or not. Artificial General Intelligence (AGI) is the next major milestone in AI: an AI that can match human-level cognitive abilities across diverse tasks. AI at this level would fundamentally alter human society and potentially lead to exponential growth in many scientific fields. It could also pose one of the largest threats to mankind we have ever seen. But what is it really, and when will it actually be here?
Most AI we have today is “narrow AI” specialised for specific tasks. Think chess bots, voice assistants, the like. More recently, large language models (LLMs) have become increasingly prevalent, with the likes of ChatGPT or DeepSeek being used by what feels like everyone. These LLMs may feel just as smart as a human at times, but they lack a true understanding of the information they make use of, instead they simply choose the answer with the highest statistical probability of being right. AGI hypothetically goes beyond both of these: it can learn and perform any intellectual task a person can. The way we reach this point can vary by definition, whether it is by mimicking all the pathways of a human brain, creating an AI with consciousness, passing the Turing test, or otherwise, the through line of all of them is that AGI must match humans in comprehension and performance of cognitive tasks.
There are no two experts who agree on when exactly AGI will arrive. Some see it emerging within the next few years, while others don’t imagine it happening for at least a few decades. There is massive uncertainty amongst those in the AI field. In a survey of thousands of AI experts conducted by AI Impacts in 2022, the probability of AGI by the early 2030s was set at 25%, while AGI by 2045-2050 was 50%. Most of the experts who see it earlier than this are directly involved with AI companies, who wish to create hype around the field and its prospects to encourage investment.
Timelines for AGI keep shortening as advances come about more and more frequently. AI experts tend to have very conservative estimates surrounding when breakthroughs will occur, even being surprised at the successes of LLMs like ChatGPT. There has been a recent shift in the last couple of years, where researchers are finally becoming more optimistic in their estimates. As this change occurs, the coming of AGI seems closer and closer.
While there is certainly disagreement on when AGI will come about, there is near-universal agreement that humanity must prepare for its arrival. Given the transformative nature of a breakthrough of this magnitude, humankind must have the facilities in place to handle the advent of AGI as soon as possible, as it could very well happen within our lifetimes.
An AGI future has a lot of potential for restructuring society as a whole. Should an AI be able to match or surpass most humans in terms of accomplishing tasks or doing jobs, there could be serious questions about human employment and economic inequality in a system that no longer requires human labour. What then is the role of humans if we are not needed to fuel an economy? What are working class people meant to do to support themselves when even more complex jobs are taken by AI? If AGI only complements the high earners of an economy, while displacing lower income workers, income inequality could skyrocket and lead a destabilized economy. A social support network may become a critical necessity for most people. Governments would have to seriously consider implementing a universal basic income to ensure that those who have lost their jobs are not left by the wayside.
While AGI is a scary prospect for those it would replace, the technology could also drive massive growth and help solve complex global problems. Given that an AGI would match humans in all jobs, this would naturally include the job of developing more advanced AI technology. Should this happen, AGI will have the ability to improve itself, and thus cause an “intelligence explosion” wherein AGI will, in fact, be the last thing we as a species ever invent ourselves. All subsequent advancements will be made by ever excelling AI. This means that, like with the advent of the internet, humanity’s work-to-leisure ratio will ideally improve dramatically. The dream of AI allowing mankind to pursue arts and culture could finally be a reality, instead of the other way around. There are proposals in place for a governance structure that will ensure AGI remains a beneficial advancement, limiting the chances of a runaway self-improving, all-knowing machine. These proposals mostly focus on transparency by the institutions and private companies running the servers and writing the code, and a democratic process for holding both these agents and the AI accountable should something go wrong. Used correctly, AGI could increase human quality of life dramatically, but there are always unforeseen consequences. To harken back to the internet example, while people believed it would bring about a new era of leisure, mankind ended up working more than ever before as their jobs entered their home lives through the internet allowing work-from-home models. It’s important to understand that AGI is meant to be a tool for us, rather than the inverse becoming true.
AGI is an exciting prospect, but still highly theoretical. It will require more massive breakthroughs in AI technology, but even with these limitations, a mid-century arrival may be likely. AGI can bring enormous benefits when it does arrive. It can rewrite how we think of our economy, of science and technology, medicine, and more. It can also cause massive problems, such as ethical dilemmas and job displacement, without proper handling. AGI remains a goal on the horizon, and we must work on not only the technology but the social framework around it. Whether AGI arrives in 10 years or 100, understanding the implications of such a breakthrough will help us maximise benefits and minimise risks. The rise of AGI is coming, and it will change our world forever.
The views expressed in this article are the author’s own and may not reflect the opinions of The St Andrews Economist.
Image Credit: Jumpstart Magazine