Morfternight #83 🌃 Good Night!

Do you have six hours to spare to figure out AI?

This is Morfternight at Night, a newsletter about Artificial Intelligence.
More information about the Morfternight newsletters is available here.

Image generated using Midjourney

To generate this image, I first asked ChatGPT:

I want to ask Midjourney to create the image of an ambivalent AI, where one can't really say if it is benevolent or evil. Can you help me write a prompt?

To which it replied:

Of course! Here's a prompt you can use to ask Midjourney to create an image of an ambivalent AI that leaves its intentions open to interpretation:

"Please create a thought-provoking and enigmatic illustration of an artificial intelligence that exudes an air of ambivalence. The AI's design should leave the viewer questioning whether its intentions are benevolent or malevolent. Incorporate visual elements that convey both positive and negative aspects, while maintaining an overall sense of mystery and uncertainty. The composition should evoke a sense of curiosity and intrigue about the AI's true nature."

I then added:

Let's add that it should be a machine, not human like.

And ChatGPT updated the prompt:

Sure, here's the updated prompt:

"Please create a thought-provoking and enigmatic illustration of a machine-like artificial intelligence that exudes an air of ambivalence. The AI's design should be non-humanoid, leaving the viewer questioning whether its intentions are benevolent or malevolent. Incorporate visual elements that convey both positive and negative aspects, while maintaining an overall sense of mystery and uncertainty. The composition should evoke a sense of curiosity and intrigue about the AI's true nature as a machine."

I then sent that prompt to Midjourney, which created a few options, blatantly ignoring my request for a non-humanoid entity:

Nonetheless, I asked for a few variations on the fourth option (lower right), and that’s what I got:

I selected the fourth one and asked Midjourney to provide it in a larger size.

The purpose here wasn’t to achieve an ultra-specific result but to have an illustration for the newsletter without diving into a rabbit hole…

This is going to be a short one!

Today marks the end of my three-month sabbatical.

I cherished every second of it, and though I wish it could have been longer, I’m excited to return to work. I’m looking forward to connecting with my incredible colleagues again and tackling the world's ever-evolving challenges.

I complained last week about how debates have to be polarized and that few seem to appreciate nuances.

One of those few is Lex Fridman.

I want to share two recent episodes of his podcast. These two interviews give the best overview of the situation and its implications in less than six hours of anything I’ve read or listened to recently.

Let me know what you think!