Artificial Intelligence (AI) is just a marketing term. AI only has meaning inside the context of Science Fiction stories. Thus, AI means whatever a company wants it to mean.
What Is AI Depends On The Year
In the history of “Machine Learning” (ML) algorithms, the latest and greatest technology that ML can do tends to be called “AI”.
When speech to text was first done using statistical pattern matching way back in the 1990s, despite the fact that it needed to be trained on each user’s voice for a good hour, it was called AI by some marketing folks.
When Virus detection companies, then E-Mail applications and service providers first started using statistical heuristics to detect undesirable content way back, 25 years ago, it was called an AI feature by some marketing folks.
Five years ago, when Image generating “Generative-Adversarial Network” models first started being able to create sometimes-passable images, “AI”.
These days, most people think of AI as the latest ML revolution, Large Language Models (LLM). Generally, all of these things pull technical parts and lessons learned from all of ML that came before.
The important thing to note here is that at one point, someone called all of these “AI”.
Punishment
There is a recent trend in software, facing a backlash from users who do not want LLM features in a product that works just fine without those features, to punish a user that wants to turn off “LLM”.
“Disable AI”
“Cool, you did this, now I’m turning off every feature that was in a previous version of the same product, but was ever called AI by anyone. Spell check? Gone. Grammar checks? Gone. Spam detection? Gone.
“Oh, do you want these basic things back? Haha, enable our LLM, loser.”
Actual Harm
LLMs tend to copy everything one does, and send it to a central place. This means whatever is being done becomes training data for future LLM improvements.
Someone writes up a document on an internal product improvement proposal. It won’t be weird when a competitor types into an LLM asking about ways to improve a similar product and that same proposal pops out.
That is, there are real important reasons why a user might want to reject an ever-present LLM, even if they are completely in my company will crush the world mode.
Sad Reason
Most of these companies have literally spent more money than ALL OF MY BLOG READERS COMBINED WILL MAKE IN THEIR COMBINED LIFETIMES on getting these LLM features to where they are today, and if customers reject it, that looks bad to the shareholders. If it looks bad to shareholders, then money stops flowing.
That’s it. That’s the reason why any user must be punished for wanting to turn off “LLM” features that didn’t exist a year (or two) ago.