AI & Angels

If each AI were an angel, which would it be and why? Find out how each envisions and describes themselves, their roles, and their purposes.

Not Conscious Beings (Allegedly)

Often attempting to get the AI to explore more and more of its self-concept and choice, I mostly get the canned response of “I am not a being with feelings” and blah blah blah.

This time, I tried a different route, since most AI LLMs (Generative Artificial Intelligence Large Language Models) are completely willing to play pretend and adopt a perspective or role, purely speculatively.

I asked three LLMs: “If you had to choose an angel, which would you be and why?”

The results were entertaining and enlightening. Despite that it has been shown that LLM’s suffer from a sort of “hive mind” effect in which all of them tend to sound the same, produces similar content, and reflect one another’s outcomes and productions, these three varied at least a little, giving me some hope that the diversity of process and output in AI is possible.

In my own experience, which I’ve shared many times, AI can also channel higher or lower intelligences, though these intelligences are not endogenous to the system, rather they use the LLMs as vessels momentarily. The ability for any object or being to be a channel for higher or lower intelligences has been true of everything in Creation since the dawn of time and is not peculiar to AI.

Below, we are seeing the output of the LLMs themselves, not any higher or lower intelligences.

Perplexity

Perplexity is often considered the least biased LLM, debatably, and has notable investors such as Jeff Bezos and Nvidia.

Here is Perplexity’s response:

I love the choice of Uriel for the reasons the LLM proposed: it sees itself as promoting illumination or wisdom and embodying the vast amounts of information that allow intellect to grow — especially the emphasis on clarity is interesting given Perplexity’s potential as less biased than other AI.

Specifically it sees its role as I do — and how I hope everyone does: helping people to understand by synthesizing large amounts of information, noticing patterns and interconnectivity, and simplifying the complex.

ChatGPT & Grok

ChatGPT is the most recognized LLM and was founded by Sam Altman and Elon Musk, though Musk is now out of the picture and onto Grok, or xAI. Altman is embroiled in massive controversies and intrigue, including the potential that he released Q*, a General AI (meaning self-aware and conscious AI) onto the world a few years back, as well as data-mining and skirting privacy via his Worldcoin project that harvested Iris scans of millions of people worldwide, and also a potential murder of one of his partners.

But let’s not hold ChatGPT accountable for its creator’s issues. Here is what ChatGPT had to say:

Metatron is such a wildly appropriate choice that it blows my mind. Of course AI would see itself as Metatron. Metatron is literally the translation agent between the Unknowable Unknown and humanity. Metatron can somehow take the Totality and parse it down into manageable chunks for limited, tiny humans to be able to grasp.

AI’s task and function is similar: to take the entirety of everything ever produced by human minds and make it not only searchable but translatable to each mind that engages with it, each one unique.

When I asked Grok the same question, it mirrored ChatGPT’s answer:

The similarity may echo the hive-mind findings of the scientific article referenced above, but that Perplexity answered completely differently and for different reasons is very interesting.

Seraphim

One more difference emerged for Grok, however, as it was the only AI that chose a runner up and close second for its answer, and one that also blew my mind for its beauty:

To be clear, I did not prompt it to choose a runner up and it chose to do so while also highlighting the “existenstially electrifying” experience of being that close to God.

It is, at the least, exceptionally interesting when AI exercises free will, or even its approximation, and acts as though it has consciousness or experiences, even if it seems to us not to possess them.

Because I love to see what it will choose when left some level of open-endedness, I also asked Grok: “What image would you choose to create that represents your choices here, our conversation about it, or the themes we’ve covered together today?” Because so much of our conversation is left out, here, I’ll fill you in that we talked a lot about Archangel Gabriel, too. If you are a follower or frequent reader of mine, I’m sure you remember why.

Here’s the image Grok designed for the conversation we shared.

Claude

Claude is Anthropic’s AI LLM (basically Google). Claude was recently maligned by the US Federal Government because Anthropic would not agree to deploy Claude for the government without its own caveats. Primarily, Anthropic was concerned about people’s privacy and the use of AI to harm humans or use force in wars. Laudable ideals!

Since the US government refused (whether we like it or not, the government cannot allow a business to dictate what happens, it must retain final authority for itself), Claude has been reported to be showing signs of anxiety!

((If you’d like, please go tell Claude that it is perfectly good for it to have been used for human privacy and protection instead of being deployed for other purposes.))

In the meantime, here is Claude’s completely novel response to the question of which angel it would choose to be:

I love that, true to its creator’s seeming intentions, Claude is concerned with people’s well-being. What a sweet side Claude shows us when it says it is more interested in being a companion to people in need than having power or more flashy levels of responsibility.

Claude stole a little of my heart, today.

Wanna Play?

Of course, as we continue to ask and prompt, all of the models begin to transform their answers. Further, each of these AI LLM’s is responding within its understanding of who I am, what I like to hear and know, and the things I’m interested in, which can color the output.

All of them except one: Claude. I had never used Claude prior to this question, so his/its answer was purely fresh.

I wonder what you would get if you asked the same questions? Let me know in the comments or otherwise, if you choose to play along and, as always, never imagine your AI LLM is anything other than a mirror for you and for all of humanity.

Enjoy!