Is AI Really Not That Smart? What Comes Next for Tech.

You've probably heard a lot about how intelligent AI is becoming, but a top expert in the field, Yann LeCun, has a different take. He believes that current AI systems aren't actually 'smart' in the way we might imagine. He even said that robots aren't as good as a rat at understanding the physical world around them. This might sound surprising, especially with all the buzz about AI writing stories or creating images.

LeCun, who used to be a lead AI researcher at Meta (the company behind Facebook), points out that today's AI is very good at specific tasks. Think about predicting the next word in a text message or recognising faces in photos. But these systems lack a common-sense understanding of the world that even a small animal possesses. They can't learn by watching, experimenting, or interacting with their surroundings in the way humans or even rats do.

So, if AI isn't 'smart' yet, what's next? Experts like LeCun are pushing for a new approach. They want to create AI that can truly learn about the world by experiencing it, much like a child. This would mean AI could understand cause and effect, grasp basic physics, and adapt to new situations without being specifically programmed for every scenario. Imagine an AI that could learn to stack blocks just by watching someone do it once, rather than needing millions of examples.

For everyday Australians, this future AI could be a game-changer. If AI can understand the physical world better, it opens doors for more advanced robotics, like robots helping in homes or assisting in tricky jobs. It could lead to smarter self-driving cars that navigate complex environments more safely, or AI tools that can truly understand a small business's unique needs without extensive setup. It's about moving from AI that's great at specific functions to AI that has a more general, adaptable intelligence.

This shift isn't just about making AI more powerful; it's about making it more useful and intuitive for us. If AI can learn more like humans, it could make our interactions with technology far smoother and lead to innovations we can only begin to imagine now, impacting everything from healthcare to how we manage our homes and businesses.