David Wojick has thought a lot about AI. He says it may be little more than a fancy Wikipedia, primarily because bad inputs yield bad results and only real intelligence can separate wheat from chaff.
I to have given much thought about AI, even messed around with it some, not to the point that some have. I tend to agree with you, it is like any other tool one would use for research. If you ask the right question enough, eventually you find the correct answer. The tricks to good research apply, know enough about what you are asking to sort out bad data and look in more than one place. I also agree that there is a danger, as you say there is always a danger of putting machines in charge. The question that I pose on that issue is this: How will I know when a machine has been put in charge?
I to have given much thought about AI, even messed around with it some, not to the point that some have. I tend to agree with you, it is like any other tool one would use for research. If you ask the right question enough, eventually you find the correct answer. The tricks to good research apply, know enough about what you are asking to sort out bad data and look in more than one place. I also agree that there is a danger, as you say there is always a danger of putting machines in charge. The question that I pose on that issue is this: How will I know when a machine has been put in charge?
Your question at the end says it all about the threat!
Thats the bothersome part, no one has an answer for that pesky question.