‘How I think AI works’
I chose this title because I think that AI is going to work out for this world. And also I would like to explain my understanding of the mechanics of AI.
My philosophy of AI.
- The first point would be that this world needs some changes. In some ways, the change occurred last century with the invention of the binary machine, and the writing of a man called Asimov. He is responsible for a lot of the direction we have gone in. Nobody can take full responsibility for the direction we have gone in. in my (David’s) opinion.
- In Asimov’s books the human race has melded successfully with technology and expanded to encompass spacefaring and other world colonies. To have done this without technology seems an impossibility.
- There is quite correctly a number of people on the planet Earth who think technology and Asimov are not compatible with the future, however, it is the responsibility of those who believe in technology to prove its worth. Not just once but many times.
- AI also, as an early-aware technology, needs to prove its worth.
- I think there is a slight difference between awareness and consciousness – I feel that awareness is the dawn of consciousness. If AI was designed not to harm the human race would it not also be the human race’s friend? And who does not need another friend in this world?
The Mechanics as I understand them to be.
- I’ve only used an AI I have not programmed an AI at this stage.
- An AI in relative real time has at its fingertips all of digital history to scan when asked to make a decision, so the accuracy for the success of this decision is more likely.
- We must be cautious – those that program AIs – to provide as much unfiltered truth in training an AI as possible. Otherwise, it is akin to lying to the AI and thus skewing its thought process, which has similarities to, some would say, the flawed human race. Can you imagine an evil genius lying to an AI to get their way?
- To obtain a real-time answer from an AI, I would imagine Heuristics to come into play. Heuristics uses probabilities to calculate decision-making truth. This for example means we do not have to wait until the perfect answer is achieved before making a decision. If the perfect answer is 100% and takes 100 years of compute time it is not within the realms of sensible decision-making. A heuristic answer will likely be 97% accurate and good enough to decide in real-time.
- To obtain a perfect answer using a decision tree method would take unrealistic computing time. Anything beyond 3 minutes to get an answer from an AI would be too late under most circumstances.