AI and Machine Learning - Opportunity to disambiguate the term "AI"

[aardrian]

Yeah, that issue name is a mouthful of fancy words.

Essentially, this Note could benefit from explaining the different types of machine learning and generative AI that are in use (or proposed).

If nothing else, this can make it easier for readers to identify which technologies are tried and true, which are nascent, and which are on the horizon.

Otherwise, these terms are so loose and generalized as to be meaningless. Marketing efforts, in particular, have laid claim to "AI" for things that are no more complex than find-and-replace (#16 touches on this in particular).

I suggest expanding 1.1 to provide specific examples of each and (perhaps) dates they went into commercial use (or some other common date that shows actual uptake versus theoretical musings).

For example, I have seen the following all called "AI" at some point (and which certainly have fed into and/or support larger generative efforts) but without understanding by people using the terms (initialisms intentionally withheld):

• optical character recognition
• natural language processing
• text-to-speech
• auto-complete
• predictive text
• algorithms
• query expansion
• computer vision
• large language models
• completely automated public Turing test to tell computers and humans apart

OCR is an example of a technology that, being arguably over 100 years old, can be as mundane as audio tone changes or robust as parsing hand-written words.

[mgifford]

That totally makes sense to me @aardrian particularly when we are thinking about how "AI" is going to be used to support efforts for greater inclusion.