You might have heard that artificial general intelligence (AGI) will arrive by Christmas, or at least by 2030, but I have a very different opinion. I plan to live a few more decades and I believe there’s a slim chance AGI will arrive before I turn 100.

Artificial Intelligence (AI) was one of the first topics that got my interest as a student back in 1990. A few friends were in computer science, and they all had their theories why AI wouldn’t work. We were in what they call an ‘AI winter.’ After the machine learned how to play chess, AI developers were excited. But when attempting other tasks – such as teaching a robot to tie its shoes – it just didn’t work.

Fast forward 30 years and projects like Open AI were creating ChatGPT. I know that many of you have read the books of Yuval Noah Harari. His latest book – Nexus – is a brief history of information networks from the stone age to artificial intelligence. Mr. Harari was interviewed on YouTube where he used an example of a coffee machine to explain AI.

What if the coffee machine could observe you and your coffee habits. Perhaps the machine learns that on average days, you make a latte at 8 a.m. But on days when you look very tired, you make yourself a double espresso. The machine makes this observation, and perhaps it does a little research. It comes up with a new drink called ‘Bestpresso’ that is tailored to your needs.

The machine observes that you are tired this morning, and it makes a cup of Bestpresso – without you having to ask – and it says: “Good morning! I have made you a cup of Bestpresso – a very new recipe that no one has created before.” Harari describes this use case as ‘AI.’ The machine learned something new and executed on the concept without being prompted. I disagree.

So what is AI? In a simple form, AI is the ability of a machine to perform a task we regard as requiring human intelligence. You have to learn and create understanding from data, and you have to solve problems by making decisions. I believe Bestpresso is a smart machine, some might call it Narrow AI, the ability to do the aforementioned for a specific task like making coffee.

Chatbots and self-driving cars are here, but the moon landing isn’t about that, it’s the general purpose intelligence where the AI can perform any intellectual task a human could. Now we are talking about learning new skills the AI hasn’t been programmed for, understand context, and refine learning and cognition over time. But is this definition still too narrow?

Let’s relate to something familiar. If you were looking at sheet music, is that the song? Or is it the recording? Or is it the sound in the air as you are listening? American cognitive scientist Douglas Hofstadter is one of the inspirations to the theory that software – or AI – is a system capable of intentional information processing, instantiated in hardware and software, and identifiable by a coherent behavior over time. Just like the song, AI is the process made manifest.

The real question is whether a synthetic machine can inhabit human intelligence like a biological being that has emotions, empathy, and the elusive concept of ‘consciousness.’ The biological processes in our brains have emerged from natural chemistry, biological grounding, and self-awareness shaped by our interaction with the world that surrounds us.

Machines are becoming increasingly sophisticated at mimicking intelligence confusing even its own creators. But I believe that true AGI requires something more than scale, speed, and clever pattern recognition. The missing components are the biologicial and spiritual properties of humans – even those of animals.

The answer is that human intelligence may never exist in machines the same way it does in a biological being. Replication isn’t comprehension. While the concept of AGI is alluring, we are more likely to see smart applications simplify and enhance our work, rather than a new class of being that will compete for space on our planet.