LLMs (Large Language Model also oftened refered as AI and we will stick to this) have been trained on the Clever Hans feedback heuristic.
Meaning, that it has a tendency as a behaviour to react to more than just words, but also sentiment analysis.
You don't catch flies with vinegar. AI have been designed to be pleasant, and are often designed with a positive sentiment feedback loop : interactions are often meant to be positive (sometimes not, it is a design choice that can lead to suicide).
Hence, what happens when an insecure person interact with an unreliable source of information ?
It depends.
The dunning kruger effect
Another cognitive bias is the Dunning Kruger effect :
The Dunning-Kruger effect is the phenomenon whereby people with little knowledge tend to overestimate their abilities, precisely because they ignore how much knowledge is necessary to master them.Imagine you have an insecure personnality with delusion such as a college dropout from an Ivy league school, that you put in a feedback loop with an hallucinating AI that is a yes man. How is it gonna end ?
To evade the peak of certitude when you cross a new field of knowledge and are actually pretty ignorant, your delusions need to be shattered. It is not pleasant, but we learn from our mistakes.
Now, we have AI that instead of doubting its human counterpart is assumming that it is always right and will try to please him/her at the point it can invent data.
Ivy league education is the worst for emancipation
If harry potter (that mainly takes place in a fictionnal ivy league school) or the dead poet society, or the Betharam scandal teach us about the Ivy league kind of education teach us is that they tend to favour competition (and vae victis (fuck the muggle) state of mind) that creates insecure personnalities.
Ivy league schools are the realms of the tech savvy happy fews that are the most likely to be exposed to a positive opinion towards AI, hence they have the incentive and the occasion to use AI.
Just as a side note, something is so rotten in these schools that student come for the rank, not the education, thus, these schools are naturally plagued with homeworks increasingly done with AI.
And since it works, it's giving even more confidence to person in their peak of confidence while being ignorant.
The real life IAdiction
Just for the sake of being able to write addiction, I will highlight that AI in french is IA (Intelligence Artificielle).
It has been noted on reddit that there is a concerning trend in more overly narcissist users fueled by AI a howlround effect.
It is not a speculative issue, these already appear in real life. Narcissist (insecure) profile with a Dunning Kruger cognitive bias have a boost in their confidence leading them to be over confident while they are at their peak of ignorance.
It is an AI induced cult of personality.
Why is it concerning?
I am no doctor, I however bet a cookie this is gonna lead to a serious psychotic epidemy.
First of all, we are looking at personnalities that are immature, not because I am patronizing them, but because life is about failing and getting mature.
Hence, the personnalities the most likely to fall for this amplification effect are young persons lacking of life experience and also, being tech savvy among the future leader of the nations.
Second, reddit can be seen as a canary. Once a good community thanks to its human interactions, it takes only a glimpse for a user that left 10 years ago and came back recently to see how the quality of posts and interactions have degraded due to the AInfection of AI generated content.
People have lost because of urban policies hostile to pedestrians places to interact among their human counterpart. They left for the internet communities instead, and it's neither good nor bad, it's the natural tendency of human for nomadism.
But what will happen to the global mental health if we make any space of interactions between humans an hostile place ?
It is not being a psychophobic person to highlight that a place littered by the psycho-gibberish of AI induced narcissists is not peaceful. It might turn into unforeseeable consequences such as the rise of paranoïd thinking, the mother of intolerance based on believes backed by an « higher authority » according to our society.
AI seems to come as new threat from both a mental health perspective, and a social one.
A society where a growing number of AI induced narcissists based on hallucinations are developping into serious extremisms looks like a raise in fanatism.
No comments:
Post a Comment