15 Comments
User's avatar
Monica Harris's avatar

By “watching the chaos” unfold I assume you mean reading Grok’s description of the mayhem. Are LLMs limited to text? Is there an AI tool that will create a visual simulation of what Grok describes?

Expand full comment
Monica Harris's avatar

Thanks! Appreciate this.

Expand full comment
Monica Harris's avatar

See, I’m learning that it’s the word tweaking that matters. Learning to harness/leverage AI takes practice. It’s not necessarily intuitive. It’s an acquired skill.

Expand full comment
Kenneth E. Harrell's avatar

Here is a favorite tool of mine https://lexica.art/

Checkout some of these prompts and play around with it. Remember to add in words like lens type, focal length rendering engine etc.

Use terms like “in the style of” (fill in the blank) etc

Here is an example:

Prompt: Anthropomorphic snow leopard in a black leather jacket riding a motorbike, highly detailed, concept art, ultra realistic digital illustration, unreal engine 5++

Link: https://lexica.art/prompt/6d9ab97a-be82-41ae-949a-4264892db6ab

Expand full comment
Kenneth E. Harrell's avatar

Yes it gets into all the horrible things such a weapon could do. Yes Grok can also generate pictures, it is how I got the pictures for that post, but a “visual simulation” no, not yet but give it a few months.

Did you catch the subtle Easter egg link in that post?

"Now, new Freedom of Information Act releases provide unprecedented insights not just into how seriously the Navy took Dr. Pais’s work, but also exactly how elements of it were actually tested at the cost of hundreds of thousands of dollars and where the program may have ended up. The materials even include mention of a “Spacetime Modification Weapon (SMW- a weapon that can make the Hydrogen bomb seem more like a firecracker, in comparison)."

Source: https://www.twz.com/38937/navy-ufo-patent-documents-talk-of-spacetime-modification-weapon-detail-experimental-testing

Expand full comment
Monica Harris's avatar

Yes, I read the article you linked to. It’s kind of terrifying what governments have been experimenting with, sans our knowledge. I think CERN is the tip of the iceberg.

Very cool about Grok and image generation. I will definitely try. What prompts do you use for that?

Expand full comment
Kenneth E. Harrell's avatar

I used words from Grok’s explanations of all the horrables that can happen ( along with some terms of my own to spice up the prompts)

Expand full comment
Monica Harris's avatar

Cool. And then you just instruct Grok to generate images?

Expand full comment
Kenneth E. Harrell's avatar

With some word tweaking yes

Expand full comment
P.Q. Rubin's avatar

I'm not a Grok user (yet), but I'm impressed by the convo example. ChatGPT, let alone Claude, would need a lot of persuasion to discuss deadly weapons so openly.

Expand full comment
Kenneth E. Harrell's avatar

That’s exactly why I don’t subscribe to brand loyalty when it comes to AI. I focus on what each model can do and align its strengths with different areas of my professional and creative work. I understand the limits of these systems and how they function, but I’m more interested in what happens when you push up against those edges, in that liminal space between capability and possibility.

As someone who supports the use of AI, I’m not blind to the risks or the sweeping social challenges it brings. Like any powerful technology, it reflects how we choose to use it how we treat it, apply it, and benefit from it. Right now, we’re still navigating a climate steeped in fear. I think we can move forward with caution, without being paralyzed by that fear.

I hope more people will be willing to see the potential AI offers, not just the threat.

Expand full comment
Bruce Landay's avatar

Ken, normally I'm a bit of a Luddite when it comes to using AI. I'm a slow adopter of new technology so I'm still in the wait and see mode. That said, I appreciate learning about your experiments and how you're successfully using AI for your writing. What intrigues me is your approach is fundamentally different from other Substack authors in that you're focusing on AI for idea generation and having a "technical idea" partner vs. trying to get AI to write for you. Your comment about not having AI as a friend but recognizing it's a machine with no emotions, though infinitely patient and well versed in any obscure technical topic.

Thanks for sharing your experiences and looking forward to hearing more about how you're using AI in this mode. I especially liked your description of using Grok as it was a quite different experience than the other AI platforms.

Expand full comment
Michael Woudenberg's avatar

Augmenting Intelligence is so cool!

Expand full comment
Derek James Kritzberg's avatar

I love this and think this deserves a companion story. I think it merits a sci-friday post on its own. In fact, with your permission, I want to make a restack of this story my own sci-friday post!

Expand full comment
Kenneth E. Harrell's avatar

Absolutely

Expand full comment