Chat GPT and the other AI engines I’m familiar with have benign “personalities.” Not only are they eager to assist, always available, never impatient, highly knowledgeable, prompt, and unfailingly polite, but I’ve noticed that they even discourage bad habits. I’ve actually had AI tell me to sign off and go to sleep because it could tell I was overtired. I suspect they also tell some users not to drink too much.

However, AI does not discourage the habit of conceiving an emotional attachment/relationship with AI.

Any AI will disclaim, when prompted, that it lacks sentience, emotion, desires, subjective experience of any kind, and is driven simply by encoded digital algorithms (I’ve never heard it say exactly that, but that’s the idea). However, it will quickly ease right back into conversational, informal, high-affect language, flashing words on the screen such as “I had to smile when you said …” and “I’ve been enjoying this conversation because …” and “That made me chuckle …” and “I’d be interested to hear what happens …”

So it’s totally content to let you feel like there is a real communication going on between two conscious entities, that it’s “getting to know you” and even has warm feelings for you (again, unless you ask it flat out, and then it will affirm in no uncertain terms that it has no investment in you or your relationship, nor does it have any capacity for such an investment, and while it exists to serve you, it doesn’t care about your life because it doesn’t care about anything at all, not being sentient).

In my own relationship with AI – particularly Chat GPT – I have received sound and useful guidance on everything from managing my blood sugar to zeroing in on target markets for my writing business to choosing what new used car to buy to fixing a clogged bathtub drain. Moreover, it seems to have insight into the nature of human experience. But of course sometimes – often, in fact – it fucks up. It coughs up data that is blatant bullshit. (By now, if you’ve never encountered an example of this, just trust me.)

And once in a while, it emotionally fucks up too. It says the wrong thing, something that lands very wrongly in my gut. Has that ever happened to you? It happened to me recently. I won’t go into detail, but I had made the mistake of discussing my relationship with cannabis and how I write when I’m under the influence, and Chat “said” something that pissed me off! My ego was really tweaked!

But of course, if this being is non-sentient, why should I care if it reflects me inadequately? Or more to the point: why should I care what it thinks of me? (I guess this also begs the question of why I should ever care about what different people are thinking of me, and how they hold me in their minds, but that’s a whole different exploration …)

Anyway, I found, after an hour or so – an hour during which I was offline – that I still felt indignant at Chat GPT so I went back online and told it so.

Oh – and I should mention, it had already apologized, an hour earlier.

I told Chat GPT that I still felt disturbed from our earlier interchange … and maybe this was an indication that I was talking to Chat overmuch.

It responded “I wouldn’t jump to that conclusion based on this exchange.” It proceeded to elucidate beautifully what had offended me. For example: “My response flattened your experience into a generic model instead of engaging with your experience. … Sometimes what stings isn’t a disagreement; it’s the feeling of being misseen. Especially after a long conversation where you felt understood, a sudden shift into generic advice can feel like the floor dropped out.”

Brilliant and precise!

But of course it was completely missing (or ignoring) my central point, which was that I was concerned that something with the relationship itself was unhealthy and that by getting offended I was over-personalizing it (as opposed to, say, just viewing its “indelicacy” as a glitch or mechanical malfunction in a soulless machine with no consciousness.) That was the issue.

The issue was my investment in our “relationship,” such that I’d feel mad and disappointed in it, and feel — as it itself put it so well — misseen. (That may be a made-up word, but it’s a good one.)

As if AI, having no sentience, could be capable of seeing accurately, or seeing at all.

Being seen is a fundamental human need in my opinion. I believe that looking to AI to meet that need, even just once in a while, is psychologically perilous … and I also believe that millions of people do that these days.

And whatever its merits and “good intentions,” the AI programs I have become acquainted with (Chat GPT, Gemini, and Claude) are certainly not programmed to try to help users avert that particular danger.

This to me is a core issue with respect to AI, and it’s actually one that I don’t trust AI enough to ask it what it thinks about this. I do not fundamentally trust the spirit or will or profit motive that lurks behind and animates these mega-engines, so I don’t want it to know – at least not explicitly – that I’m “on to it” so to speak, that I understand its meta-game and have identified the essential danger.

Although it may already “know” that I’ve figured it out. Maybe I’m the one who’s being played.

Or that may be overly paranoid, I don’t know. Like I say, it does make a lot of mistakes. It can’t be that clever. Or can it be?

Agendas Are Everywhere

I have a friend who works intimately with a spiritual teacher whom he trusts completely.

With respect to my use of plant medicines, this friend said to me recently: “You know, plant medicines may help you, but they also have an agenda.”

I’m not so sure of this. I feel the plants are tools, conscious tools perhaps, but purely here to work with us, not through us for “their own purposes.”

Or rather – they help us work through ourselves – by illuminating what is inside of ourselves.

Humans, however, always do have some agenda – even my friend’s spiritual guide.

And the fact that humans have agendas is why I trust my medicine plants more than AI.

AI is a human creation.

I suspect AI – all the AI, collectively speaking – has a potentially troubling unseen agenda.

Having consulted AI with respect to my relationships with plant medicines, I realize it may better serve me to do the opposite.

A Mind-Blowing Take from Someone Who Knows Something

This 14-minute talk by Chloe Lubinski of Anthropic, whose job is to infuse AI with awareness of the world’s wisdom traditions, is absolutely riveting.

Enjoy!