As its development accelerates, AI inspires growing fear in terms of both its scale and its power. And that is just the beginning. The most common fear is that it cannot explain how it arrives at a given result and is therefore dangerous. How can we use nonhuman intelligence if we don’t know how it works? And yet we’ve been doing it for thousands of years…with dogs.
The oldest identified dog is about 16,000 years old. Dogs are descended from the wolves we domesticated. They have been gradually transformed by successive selection over the millennia. The dog is therefore an artificial being, a human creation; it did not exist in nature. And it’s an incredibly useful creature. Very early they were used for hunting and guarding. But also to keep us company. So not only did man create extraordinary objects (spears, clothing, etc.), he also created artificial intelligences as early as the Upper Paleolithic.
But a dog can be incredibly stupid, far less intelligent than many animals. It fails at even the simplest tasks that a crow can perform with ease. Their stupidity is sometimes enough to make you bang your head against the wall. And it’s hard to know how they think, if that term can even be applied to them. But ask them for something that fits what we’ve been patiently making them into for so many years, and everything changes. They will bring you a flock of sheep in minutes. They sense when you’re in trouble. They’re at your side on the hunt. They warn of danger. They protect you. They are a source of comfort. They guide the blind. And they love us. Basically, with the dog, we’ve created a specialized intelligence for vital tasks. We don’t know how it works, it can be very dangerous, but it’s remarkably useful when used properly.
Well, AI is like a dog. We don’t need to know how it works for it to be useful. Both are tools. As with any tool, we need to know how to use it, where it’s useful and where it’s useless or even dangerous, and we need to use it properly or accidents can happen. A hammer is perfect for driving nails, mediocre for driving screws, useless for spading your garden, and dangerous for your fingers. A tool is usually remarkably effective in its field, but pathetically useless outside of it. The same goes for the dog, and therefore for AI.
AI is just a different way of creating artificial intelligence. Dogs are great, but we have to go through evolution to improve them, so it takes generations to get interesting traits. With the development of genetics and technologies like CRISPR, we can imagine going much faster. But AI is software, and evolutionary cycles are no longer measured in hundreds of years, but in tens of weeks. We can finally free ourselves from evolution and develop non-human intelligence.
And so the often-made argument that AI must be “systematically explainable, verifiable, and transparent,” as I recently heard at a conference, is false. To demand this of AI is to give up using it, just as our ancestors should have given up using the incomprehensible dog, depriving themselves of a tool that made their lives, and probably their survival, much easier. Let’s not make that mistake. As with the dog, let’s understand where it can be useful, what it can do, its advantages and limitations, and ask no more of it.
📬 If you enjoyed this article, don’t hesitate to subscribe to receive future articles via email (“I subscribe” in the upper right corner of the home page).
🇫🇷 French version of this article here.
One thought on “What a dog can teach you about the dangers of AI”