🤖 ​🖼️​ Robots, Art, and Kids

A trip down the AI-generated art rabbit hole

Out of all the luck in my life, perhaps the biggest is that I was born when my life’s work would ride the wave of the Internet and its aftershocks like mobile & cloud. To come of age at a time when the greatest economic engine ever known was being created!

Winning the economic lotto in the form of the 486, sound blaster speakers, and the HP printer.

As a parent myself now, I look back with a bit of amazement at the important and hands-off role my parents played in sparking that interest. As the son of a dentist and a teacher, I was given a computer and told to explore. It lived in my room. I was made the expert. It was my playground to expand my horizons. What a gift. (Oh, and Happy Birthday, Mom! Love you!)

Fast Forward to the 2020’s

Today the world is a bit more complex. The Internet’s been around for a bit. What the Next Big Thing might be is a bit of a judgment call. And tech’s impact is far from 100% positive (don’t think we’ll be letting our kids have a computer alone in their room anytime soon). And what was novel and created awe-inspiring amazement is now just commonplace in our digital lives.

That’s why I was so intrigued earlier this year when I saw the launch of the emerging category of AI-generated artwork. DALL-E, Midjourney, and Stable Diffusion exceeded my expectations and sparked my imagination for what technology could accomplish more so than anything I had seen in years. There was the world before I knew this was possible, and now after. It was unambiguously Frikin’ Cool.

For those that aren’t familiar with it, all three of these tools are based upon the same large language models that power text generation and speech capture. They take a text prompt and can generate a picture about anything you can dream up:

As a writer, I have more creative ideas for visuals than I have the talent to produce. So this technology is a particular gift in that it felt like it filled a creative niche in my life. Perhaps now I could envision myself not only writing but also producing the art for a children’s book.

Down The Rabbit Hole

I’ve been meaning to spend serious time playing around with the tech. I had used it to generate some artwork for a slide deck at work, but I wanted to sit down and get comfortable with it. I chose to use Dream Studio, based upon Stable Diffusion, mainly because it’s available without any waitlist.

I decided to take my kids - 5 and 8 - along with me on this journey. Perhaps predictably, this turned into a stress test of how well the AI knew about rainbows and various mythical creatures. I can confirm, gentle reader, that clearly these AI tools did include My Little Pony in their training set (look at those Cutie Marks!).

Without further preamble, here are the results we are most proud of:

Thoughts On The Experiment

It was rejuvenating to explore a new passion with my kids. So much of what we do with our kids is confined to the safe place of activities where we know the expected result ahead of time: we are teaching kids things we already know. This was different.

I was worried that I’d be teaching my kids to love “shortcuts” in life: instead of doing the hard work of a hand-drawn painting, just get a computer to do it for you. It didn’t work out like that. Our 8-year-old decided that she really wanted to have one wall dedicated to the artwork she drew herself and another wall for what she calls the “computer dreaming stuff.” And it seems to have sparked her interest: at her request, we’ve returned several nights in a row to generate more art for her wall together and she’s been thinking about various prompts and ways to work the tool to get what she thinks up in her head.

I’ve attempted to get our kids into programming or Minecraft or other forms of creation and creativity with computers; this is the first thing that stuck. Being so visual helped. So did immediate time to value. It’s led to fun conversations about what is art, who owns art, and how unique a piece of art is when it’s created for the very first time.

I regret how involved I needed to be - unlike with my own journey with computers, I as the parent was the facilitator here. It’s not yet unbounded creative explorations for my kids. This inspires me to figure out how to build on this passion and give my kids the freedom to feel like the owners and experts.

Wrapping Up: Bloopers

Lest you think what you are seeing on Instagram and in this post is the norm, the AI is far from perfect. It often needs refinement and you can see emerging behaviors here on the new way you need to “program” your requests.

Social context matters a lot, too. Despite our best efforts, we just could not get Dream Studio to give us a good picture of a unicorn answering a banana phone. ​🤷‍♀️​

Thanks for reading! And hey, drop us a line (you can reply to this mail) of what you’d like to see more of on Mind The Beet. What posts have been your favorite? What did you wish we wrote about?