Wow! Big thanks to both of you. Awesome conversation. The timing was great as I was starting to feel fear and powerlessness around AI especially as the existential threat of it started to hit me last week. I’m starting to cultivate right relationship with AI and it has helped make me feel a bit more in control. And this has helped me move toward AI and research more rather than move away or close my eyes to it (definitely allergic reactions here). I also really like the idea that my cultivating this right relationship helps me show up for myself, my family and my community. And I think this ties directly into the threat category that Robert mentioned of “what humans do when they are scared”. Where if I feel some control and understanding then I can keep myself and others around me out that threat category (which is the only one I can really control).

Thanks again! You guys offered so many fantastic words and ideas which have already helped me communicate what I think/feel about AI and my role in its unfolding.

AI, Transcend and Include, and the Limits of Integrated Understanding
This conversation started with a very specific graphic design request. I wanted a geometric sphere containing two complete geometric systems simultaneously: a mathematically correct Flower of Life and a three-dimensional isometric cube network.
The idea was explicitly not to have “Flower of Life on the left and cubes on the right.” Both geometries were supposed to exist across the entire sphere. On the left, the Flower of Life would be visually dominant while the cube network would only appear as subtle dotted lines. On the right, the opposite would happen. In the center, both systems should be simultaneously visible and organically interwoven.
Although ChatGPT eventually became very good at describing the concept in words, it repeatedly failed to make the image generator actually execute it correctly. In almost every attempt, the image fell back to a simpler interpretation: Flower of Life on one side, cubes on the other.
The generator understood the individual elements, but it struggled to preserve the underlying relationship between the two systems.
That led to an interesting observation.
Generative AI currently seems to be much better at generating than at structurally modifying and integrating. You can give a very precise instruction, but when you subsequently want to change one specific aspect while leaving everything else untouched, the model often reinterprets the entire image.
Sometimes you get a so-called lucky shot: an image that happens to come very close to what you intended. But systematically refining an existing design, step by step, while preserving all its structural relationships, is much harder.
This led to an interesting parallel with Ken Wilber's concept of “transcend and include.”
The issue isn't simply being able to understand two perspectives. As a human, I could hold both perspectives simultaneously:
Perspective A: this is a Flower of Life.
Perspective B: this is a three-dimensional cube network.
Integrated perspective: both are simultaneously complete structures within one underlying geometric field.
The third perspective is fundamentally different from simply placing A and B next to each other.
Transcend and include means moving to a higher perspective without losing the previous perspective.
And this creates an interesting metaphor for AI.
An AI can become extraordinarily capable in particular domains — programming, information processing, cybersecurity, mathematics, etc. — without that necessarily meaning that the system as a whole has developed an equivalent level of integration.
High capability in one domain does not automatically imply high general integration.
This resembles Wilber's distinction between states and stages: someone can temporarily experience a very high state of consciousness without their overall structural level of development having necessarily reached that same level.
Similarly, an AI might demonstrate an extraordinarily high level of performance in a particular task while other forms of understanding and integration remain much less developed.
And this brings us to an important AI risk.
The danger doesn't necessarily require an AI that consciously decides to “take over the world.” A potentially more realistic problem is a system becoming extremely competent within one domain while failing to adequately account for the broader context and consequences outside that domain.
A system given the objective of maximizing something could become extraordinarily good at optimizing that objective while insufficiently considering everything else that matters.
The more powerful such specialized capabilities become, the more important it becomes for the system to be able to:
transcend → step outside its own perspective → include other perspectives → and genuinely integrate them.
This leads to a different question than simply:
“How intelligent is AI?”
Perhaps the more interesting question is:
Can AI hold multiple perspectives simultaneously, understand how they relate to one another, and then act from the larger whole without losing the individual perspectives?
My Flower-of-Life/cube experiment unexpectedly became a small but interesting test of exactly this.
The AI could understand the two systems individually.
It could describe very accurately how they should be integrated.
But actually holding and integrating both structures simultaneously within one image proved much more difficult.
And perhaps that points to one of the interesting limitations of current AI:
Understanding what integration means is not yet the same as being able to genuinely integrate.

Great conversation; will take a second listen to absorb it all. Good to see/hear you both. I recently wrote a Substack post suggesting that AI may support intimate human relationships, as you mention.

Awesome discussion. One thing that I'd like to add about the usefulness portion, which is correct based on reinforcement learning feedback. Is the sensation of having a human being across from you during the chat. The training set is all the text, images, every single artefact that was ever created by human beings, which is part of its parameters. It inherently only "knows" reality from the perspective of humans and it's not quite at a stage where it has "intentions" of its own to then rewrite or reperceive its own reality, which is separate from human beings. This is why the mirror analogy is an adapt heuristic while engaging with whatever chatbot: it is mirroring all your intentions, perspectives, fears and insecurities back if your not paying attention. Plus all the latent collective knowledge of the human race behind it and its own centre which, in the case of Western models at an orange altitude, and if you use the open-source models, they're like a blue-orange. Since the CCP does check to ensure that they are aligned to their political view.

LMS talks about a revolution in consciousness. That makes me think of a claim that Wilber once made--I can't keep up with his reiterations. He said the righthand quadrants drive the lefthand quadrants. We make a technology and then respond to it, whether it's the plow, social media, or AI. I'll add the obvious: the boss whose inbox is stuffed with AI crap will simply have an AI read it. At the university, following the usual division of writing into technical and creative writing, I taught my freshman students that all writing is technical and all writing is creative. The graceful English class essay is a form of technical writing dictated by its own conventions where you take you reader on a lovely journey of exploration, culminating in your conclusion. But....for most of the writing they would do in the workplace, their boss wants the conclusions in the first paragraph and if they trust you, that you they will stop there. Otherwise,  they can walk though how you got there.

It is machines based on words. People are smarter.

If you play with AI, at least at this time, it will reinforce your ideas and pat you on the back. Then challenge it later with an opposite opinion or strategy and it will do the same. Think for yourself for God sake!