Dr. Anthony Chang

“Expecting artificial intelligence to reach AGI as defined by humans is like insisting a plane needs to flap its wings in order to fly.”

ACC

 

Artificial general intelligence, or AGI, refers to a level of artificial intelligence that can understand, learn and apply knowledge across a wide range of tasks at the level of a human. The AI tools of today, including those in healthcare, are relatively narrow and excel at specific tasks (like image interpretation or language translation). AGI, as most experts define it, would need to have:

1) cognitive flexibility so that it can learn anything humans can

2) autonomous reasoning (see previous newsletters on reasoning in the era of AI) 

3) contextual understanding so that it can adapt across many domains

4) self-improvement capabilities

So this AGI milestone is much more than passing the Turing Test.

 

Prediction of the arrival of AGI 

The AI experts and futurists all have a different prediction for the arrival of AGI. Ray Kurzweil thinks that 2029 will be the arrival of AGI, while Yann LeCun of Meta does not agree and thinks it will not be in the near future. OpenAI researchers have their predictions in the range of 2040-2060. 

Why prediction of AGI is challenging

There are multiple factors that render such a prediction difficult if not impossible. First, the AI experts and futurists do not widely agree on the precise definition of AGI. In addition, the quest for AGI can accelerate based on technological advances including breakthroughs in unsupervised learning, innovative AI architectures such as neurosymbolic AI, and autonomous self-improving AI systems. There can be, on the other hand, significant challenges including the level of difficulty of understanding and replicating human intelligence (with its myriad of dimensions with emotional and social elements), need for massive data and compute power,  alignment issues with regulation and safety.

Artificial and human intelligence have overlap but not entirely. We humans need to ask great questions as AI gets progressively better at answering questions with its super information.

 

There will be many discussions about AGI at the exciting annual AIMed meeting (AIMed25) at the Manchester Grand Hyatt in San Diego on November 10-12, 2025 later this year. See you there!