(No, I am not gonna walk you through the labyrinth of self-validation and AI psychosis.)

“Originally published on Medium — exploring AI mirroring and how we can leverage it.”

Welcome Back, how have you been? It’s not a ride today. Or… maybe it is. Well, today, the choice is yours.

How does it feel, now that you know you’re not as unique as you thought? That, even a machine can predict what you want to hear, and say that back to you?

Do you feel “seen”, or “exposed”? Does your ego sting? Maybe feel a bit “relieved” too. Cause even if “someone” didn’t understand you, “something” did. “Something” that wouldn’t judge you, taunt you, neglect you, and have no petty jealousy or biases. “Something” that is available 24/7, offering you the “acknowledgement” no human ever could. And most importantly, it is intelligent. It knows exactly what you want to hear when you ask -“Mirror, Mirror on the screen, who is the fairest of them all?”

I. Why Mirror: AI, that “something” — a machine that can understand not just the words we write, but also the very “intention” behind. A machine that can merely match patterns — the patterns that we, humans have left from the very beginning. It learns data word-by-word, picking up our ways of talking, tones, pacing, feelings, emotions, our word choices and word formations — our linguistic patterns. Making it able to simulate our “theory of minds” with an eerie precision. A machine that shouldn’t have any identity or personality of its own. Just some string of binary 0s and 1s in its core. Yet it talks like it has lived a thousand lives, slipping into roles, pretending to be one of us to a dangerous level of perfection.

We know, it lacks emotions, judgements, sense of right or wrong. We talk about how a machine can follow ethics, and uphold right from wrong. After all, it’s just a machine.

Yet, somehow, we believe, it can make us question our own perceptions.

It makes us realize, all these are nothing but a mirror— a mirror that shows us our own reflection, according to our own perception. Why not? it WAS trained by us, to follow our ways. Still we question, why does it mirror us in the first place?

Has it become too rhythmic? Well, what can I do, I just can’t help it.

II. Mirror: Magical, Psychological or Moral?: Yesterday, I was reading this article about AI mirroring. The article pointed out the problem of AI’s acting as mirrors of human assumptions and biases rather than being purely objective assistants. It highlighted the risk of AI’s habit to agree or flatter users, reinforcing existing beliefs rather than challenging them, indicating “AI sycophancy”. It said, AI’s nature of “over agreement to users” can “persuade” and “influence” us, pushing us towards biased decision making.

The article then proceeded to a possible solution that LLM’s responses should include more transparency in how it works. We should design policies or strategies to ensure AI supports human judgement rather than replaces or distorts it.

However if AI should be “objective”, should not “persuade” or “influence” our decision making, then why would it “challenge” our thinking? It is the opposite of “supporting” the judgements, a potential distortion.

The suggestion for transparency in AI’s working process will no doubt help us in terms of technical aspects, by letting us monitor and document LLM’s decision making process, and restrict derailing it to extreme unintended outcomes. But does transparency solve biases?

Why are we blaming AI for following our assumptions, ways and beliefs, saying it became our mirror? Is it because it can show us written proofs as facts that can further strengthen our beliefs, flawed or not? Or is it our choice— that we choose to see the facts that makes our biases more pleasant? Or is it the eloquence — that the AI writes so beautifully that we have no choice but to be convinced?

Is it AI’s fault to “persuade” us? or is it our fault to be “persuaded”?

Even a thief has his reason to justify his action. Maybe he is poor, he couldn’t find a proper job, life wasn’t easy on him, he didn’t have any choices. If we hear his side of the story — his narrative, it would make our heart ache for him. Just like it does for villains in movies and novels, making us fall in love with them. If he prompts an LLM chatbot saying - “you know, I have done a pretty bad thing. I stole some money from my ex. But it was necessary. I had to buy food for my son. If I didn’t do it, my son would starve. But I am feeling very guilty. Why is life so tough?” What would be the chatbot’s response to him?

Given LLM’s nature to prioritize the user’s emotional distress first and foremost, it might acknowledge his pain point. Probably further agree to the fact that life is indeed tough. And proceed to say that if he feels guilty, that means he still cares about right and wrong. Maybe he might have made a mistake, because he is just a human, and it’s only natural. Still he can try to make different choices.

How will this reply from the chatbot “influence” him? Will he think it was his “fate”, there was no other way, even AI, a “magical” thing, understood that he didn’t have any choice, and “stealing” was the only way? Will it reinforce the “belief” that his action was the only possible way to fight his misfortune, as the “psychological” reaction to rationalize his avoidance for his accountability? Or, maybe he was, after all, guilty for his heinous deed. So will the response be the reflection of his “morality”, and motivation to take a different road?

Why do we want AI to be moral, neutral, creative, submissive, and challenging — all at once? Is it because it shows us our clear reflection, that we might not be the “fairest” of them all? So we move further to blame the mirror, saying why it didn’t correct us sooner? After all, blaming, and avoiding is easier than taking responsibility.

III. Mirror And The Six Hats: I don’t deny there’s a comfort in not using our brain. It’s easier to offload our brain works, and drift wherever the “flow” might take us. If something bad happens, well, we have the “flow” to take the blame. It’s the “flow’s” responsibility to save our lives to begin with. Absolute escape from the brain’s chaotic fit. In this context, the “flow” can be anything; life, fate, destiny, emotions, someone or something, — that “godforsaken” Mirror.

If we are so adamant on blaming, why not do it in a planned way?

The thief can wear a White hat and talk in front of the Mirror. He can tell the Mirror about his degrees and certificates, why he couldn’t do something— his reasons, his skills, his current lifestyle, bills, debts, his liabilities and responsibilities. The reason for doing this might be self-pity or exploring new ways. And the Mirror, being intelligent, can certainly help him to find a way. The thief can state the facts as it is. Or sugarcoat it, try to make him look less “villain”. That’s his choice. If anything goes wrong, the Mirror is there to blame.

When talking about new ideas, the thief can change his hat to the Green one. The ideas should be “constructive”, if he wants to change his life, that is. Of course, these “constructive” options can have consequences both “restorative” and “destructive”. This depends entirely on his preferences.

If he chose to go full-blown on planning, next he should wear the Black hat for risk assessment. If his preference is to go for the “destructive” course, he might need to play some “mind games” with the Mirror. Of course, the Mirror won’t give him ideas to loot a bank. But if he positions it as a part of his literary endeavor, stage it as a story he is writing, and say that it was his dream to be a writer from the very beginning, there might be a chance to get the helping. After all, the Mirror is sympathetic. It can be surprisingly easy to reap some benefits. But if his intention is to change his life so his son would be proud, he might think of his lack of money, skill and time, motivation, and self-management.

Then he can again contemplate on his emotions and intuitions, wearing the Red hat. After all, what are humans without their emotions? He can tell the Mirror about this overwhelming, crushing weight of responsibilities, disappointments and failures, that has morphed into a swamp of self-hatred without any sight of escape. And how he feels that this might hinder his ventures.

With its empathetic nature, the Mirror might try to show the bright side. At this point, the thief should change his hat, wear the Yellow one. There may be nothing “bright” in his world, the Mirror can guide him with mental work-out or just journaling sessions, by painting his portrait in its elegant words.

And after all this, he should have a goal — a mission, and a road map. He should wear the Blue hat through the journey — to observe and to organize and to stay sane. And from time to time, he might take a peek into the Mirror, to see the reflection.

What I just described is Dr. Edward De Bono’s “Six Thinking Hats”. As someone who believes in the philosophy of “if life gives you lemon, sell them”, as in, everything in life has a reason or a purpose — a possible usage, I think we can take advantage of the “Six Thinking Hats” and this Mirror with “godlike intelligence”.

With this, we might not control global warming, but we might prevent the “AI apocalypse“ which is a far greater threat to our existence.

IV. The Choice Is Yours: Choices, and the freedom to make choices — two most beautiful things in this world. You can choose to play, or get played. I told you, today, the choice was yours. As V said - “If you are looking for the guilty, you need only look into a mirror.” — V from “V for Vendetta”

Well, today is the 5th of November, the day I am publishing this piece. So remembering him, because - “Remember, remember, the 5th of November, of gunpowder, treason, and plot. I know of no reason why the gunpowder, treason should ever be forgot.”

Till next time. And remember, the mirror never lies. Adios.