Artificial General Intelligence by 2040?

Jorge Costa Oliveira

Despite the tendency to designate everything that is technologically advanced as AI, there are still only a few cases of true AI (i.e., autonomous agents and not mere instruments), and there is still no Artificial General Intelligence (AGI).

The concept of AGI is not defined in the same way by everyone who has spoken publicly about it. David Silver, principal research scientist at DeepMind, explains that AGI refers to a type of AI that possesses human-like intelligence, with the ability to understand, learn, and apply knowledge across a wide range of tasks – from solving complex problems to creative thinking and social interaction.

Unlike specialized AIs, which are limited to a single function (medicine, finance, engineering, or games), AGI aspires to mirror human adaptability and general problem-solving capability.

Elon Musk, meanwhile, defines it as “an artificial intelligence smarter than the smartest human,” which he predicts could emerge as soon as 2026 (though let’s not forget that he has been promising “next year” unmanned robotaxis since 2016…). Nvidia CEO Jensen Huang has predicted that AI will “match or surpass human performance on any test” by 2029.

So when is AGI likely to appear? Just a few years ago, before the rapid progress in large language models, scientists predicted it would arrive around 2060. A 2023 AIMultiple survey of 2,778 AI researchers found that many now expect AGI to arrive around 2040. Some entrepreneurs are even more optimistic, forecasting it could happen as early as 2030.

According to Silver, while AGI is a long-term goal, achieving true human-level intelligence will likely require several incremental breakthroughs and will develop gradually over time. Ilya Sutskever, co-founder and Chief Scientist at OpenAI, predicts AGI could emerge within the next five to ten years, though he acknowledges the uncertainty about that timeline.

With a few exceptions (AIs unbeatable at chess or Go), today’s specialized AIs still lack the versatility, adaptability, and depth of understanding found in human intelligence. AGI would represent a transformative leap in AI research – from narrowly focused capabilities to a more holistic and adaptable intelligence. Despite accelerating technological developments, the steps toward its creation may unfold more slowly than many theorize.

The shortening of time estimates seems optimistic and most likely stems partly from the need to sustain high investor expectations so that substantial funding continues to flow.

In any case, it’s impossible to ignore that many experts consider AGI’s creation inevitable. Companies developing AGI will keep emphasizing the wonders it will enable – and that is probably true. However, it is imperative that civil society also debates the risks it poses – and that policymakers recognize the importance and urgency of preventing those risks through well-considered analysis and regulation.

linkedin.com/in/jorgecostaoliveira