We do not understand most of what we talk about

“This new iPhone is awesome”

“Sally had a good meeting with her team”

“The rocket went to space”

“My son is learning how to play the game of chess”

“We need to downsize the infrastructure team to deal with the recession”

These are all statements that contain words representing objects or concepts that the vast majority of us do not fully understand. Yet we don’t dispute that these are meaningful, or even true statements: Language is useful regardless of our level of understanding of the underlying reality.

But the associative and inductive reasoning we use to construct these descriptions of reality are generally described as a manifestation of human “Intelligence”.?

When chatGPT does it, is it any less “intelligent”? What does it even mean to ask if an AI has “understanding” of the things it speaks of? Does it even matter?

Another debate I see on this topic is whether chatGPT provides the “correct” answers. But then consider the above statements and how humans disagree on the premise of each:

“Android is better”

“Meetings are stupid”

“Which definition of Space? NASA’s definition or the Karman Line?”

“The IOC says chess is not a game, it’s a sport”

“Define recession. Are we even in a recession?”

So, should we stop discussing these topics because we can't agree on the underlying concepts? No and neither should we avoid using AIs because of it.

Lastly, another common debate is about how these emerging AIs will “destroy” learning by providing students an unfair advantage.?

Did calculators destroy maths? Did word processing destroy novels? Did written language destroy reasoning? Isn’t it lazy to write down our thoughts instead of using our brains to organise and store them instead?

Clearly not. In fact the scientific consensus is that language as a tool to organise thought is the foundation of intelligence: Calculators, pencils and words are all tools that allow us to make more efficient use of our brains.

And so it will be with "AI" - whatever THAT means. The fact that we will soon no longer need to be software engineers to write code, marketing professionals to write copy or proficient at calculus to work out orbital velocities - these aren’t problems, they are shortcuts to the next level of intelligence and productivity. We will spend less time answering questions that have already been answered and more time asking new and better ones.

This wasn't written by chatGPT but it probably could have.