1. We don’t know how the developers use the feedback they get from AI based on what AI learns.
2. We also don’t know how the developers select which material AI can access.
3. AI is designed to mimic the neural net in human brains to look for “logical patterns” in data.
- The neural net in a human brain is much more efficient than that in an AI. Our brain uses a few watts of power, whereas an AI uses billions of times more power. While AI systems will improve with time, they will not be able to match the design power of nature (kammic energy).
4. Our brains (and the neural system there) help us make decisions. But just like the AI, our neural systems (i.e., the brains) are as good as we train them. While the brain does not generate thoughts, it helps the mind (located in hadaya vatthu in the manomaya kaya or gandhabba) to make decisions.
- Let us take a mundane example. Consider two people: One spends the day watching useless movies, and the other spends time learning things and building new skills. It should be clear to anyone “which brain will win” at the end.
- In the same way, a given AI is as good as the information it gets.
- This is why AI will NEVER achieve human intelligence. It may pass standard tests like the SAT (college entrance exam) or even bar exams for lawyers; it has already done so). But that mainly involves “remembering” and making “logical decisions based on what is learned.”
- However, AI cannot generate paradigm-changing ideas like Newton or Einstein, let alone the Buddha. That is what “human intelligence” really is!
