Wow. Great ideas, thanks for sharing. I'd like to see image>text ai so it can explain what it's seeing, for driving, etc...