[kwlug-disc] On usage ... on simplistic usage of GenAI in Node project

[Khalid Baheyeldin]

There are a lot of things wrong with the current LLM, starting with
them being called "AI" without any qualification (Generative,
...etc) and in the general public's mind it is THE AI, while it is a
data-retrieval/simulator/extrapolator.

There are many people who trust the output of LLMs blindly.
They don't consider that it may invent false information (the mildly
called 'hallucinations').

Recent examples:

A real estate agent told me that he told a client that a certain house
is worth $650k, but the client insists it is $550k because AI told him
so. Even when the house sold for $660k, he didn't change his stance.
The broker decided to fire that client.

A neighbour in a discussion said: "AI explains everything to me".

There are lawyers who are submitting court documents citing
non-existent precedent cases.

There was a commissioned report on education in Nova Scotia that cited
non-existent research.

The consulting company McKinsey had to retract a report because it
contained a lot of false information.

Add all that to what Doug said about maximizing engagement (which
social media has been doing for 1.5 decades).

The monopoly of attention is the major issue, in my opinion (first
social media, and perhaps AI being the next domain)