The problem is not AI, the problem is that the current state of AI is designed for efficiency. In other words, the logic is derived from facts (which is good) and its mechanism designed to figure out the easiest shortest path to solve any given problem. As such, when asked how to fix global warming, AI answer will probably be to “eliminate” humans if it finds that we are the root cause of the issue. What current AI models and mechanisms need is to learn “values” so that it can provide more “complex” solutions to any given problem. Assuming the same problem above, such AI should understand that the problem is not that humans are farting CO2 into the atmosphere at the rate that a cow does. Instead it’s their tooling and their social arrangement that needs to be fixed. When such AI exist (and it will once Quantum computers are fully operational and more affordable), we will be able to train machines to provide a multitude of different options that are not necessarily the most efficient in the shortest time possible, but instead, oriented to problem fixing over time. More importantly, such AI will not only provide solutions to a given problem but might actually prevent future issues altogether.
AI is coming and it’s here to stay. We just need to rethink society and enjoy the extra time robots will provide us. I could use more time with my kids, family, learning, traveling or simply to read more posts like this.