Question: What is AI?

Answer: It’s a computer program that can give answers.  Their answers can vary from best travel path, identify objects that are in view, have AI drive a car in real time and have AI fly a plane in real time, etc…

Question: How does it do this?

Answer:  It gathers a large database of raw information. It then cuts up that raw information into a database of tokens.  Tokens can be thought of as puzzle pieces.  But these database puzzle pieces are not designed to be fitted back together into its raw information.  In fact, they can’t.

When a question is submitted to AI for an answer, it is also cut up into tokens.  Again, similar to database puzzles pieces.

These question puzzle pieces and database puzzle pieces are designed to be matched together into an answer. There are typically multiple answers to a single question.  AI gives these answers probability scores.  Highest score wins

AI finds ways to increase speed and accuracy to give answers with best probability scores.  This works only because AI software embeds additional info into question puzzle pieces and database puzzle pieces.   This embedding requires a single token to sit in a very large computer space, known as multi-dimensional vectors or matrices.  This refines AI’s ability to match database puzzle pieces and question puzzle pieces into answers.

Question: Is this why so many AI data centers are being built?

Answer: Yes.  Different AI vendor software have different methodologies that generate tokens and embeds additional info to these tokens. They have different ways to speed up and get more accurate answers via probability scores. Each vendor needs its own suite of data centers.

Question: Why do these AI data centers consume so much electricity?

Answer: if raw data is received in real time, these data centers have to process them and add it on.

As a start we can think of real time international news feeds, international financial data feeds, etc…

AI software might get improved.  So these data centers have to keep their raw information.  Periodically their raw information will be processed, again, through this improved AI software.

Of course, data centers are designed to handle a very large number of incoming questions.  Answers must go out quickly.

Question: What is the business of these vendor data centers?

Answer: We can simply say they sell their puzzle matching capabilities.

Question: Can Universities, Firms and individuals do this AI process on their own?

Answer: Yes.  Their database of tokens can have various sizes and be highly specialized.  They can do businesses and research.  This is a subject of master theses and PhD dissertations.  It’s similar to a burgeoning game industry.