Simple AI-Generated Imagery Programs

Cross-posted from my blog at https://ronkowitzllc.com

 

Artificial intelligence scares some people. When AI takes over human roles and takes the form of a human, that is quite scary. AI writing essays scares teachers and pleases students. It also scares people whose income comes from writing. None of us want to be replaced by a machine.

I have found that AI used to create images seems to be less scary. Oh sure, if I were a graphic designer, artist, or photographer, I would be frightened that my career could be replaced by AI. And you have heard the horror stories about "deep fakes" of photos and videos. But I have been playing with a few simple programs for generating images using AI. These are all free - for now. I do some drawing, painting, and photography myself with my hands, but I would not rate them as professional and this AI method is just so easy. 

One of the first things I tried was Microsoft Bing  bing.com/images/create/  These programs are very easy to use. You input text - a prompt - describing what you want to see created. The more detailed the prompt, the more likely it is that you will get a result that matches the one in your imagination. Beyond that, you don't have a lot of control. You can't choose colors for small parts of the image but you can describe a style (like a Renaissance painting, or anime, etc.) and add details.

For example, you could ask for an illustration of a woman. That is very broad. Let's try again - an Asian woman on a computer monitor. How about a man creating an image of an Asian woman using his phone and Bing and projecting it on a monitor. Okay, just as a test of its capabilities, let's add a small teddy bear.

Here is the result:

Search and you'll find lots of these programs. www.craiyon.com is another one. I also liked experimenting with deepai.org/machine-learning-model/text2img. I went down a similar path asking variations on a woman standing by a wall, sitting by a lake, etc. 

I'll keep posting things I create on this site and I will be using them for a lot of my posts online.

I would like to see more control over these simple tools - such as being able to start with a photo of my own - and they are not perfect in following prompts. Did you notice that in the image at the top, Bing misspelled Bing?

4 variations on a monochrome man walking in the woods prompt

Can Bloom's Taxonomy Teach Us Anything About AI?

spiral model
Image gettingsmart.com

 

When I was studying to be a secondary school teacher, Bloom’s Taxonomy often came up in my classes as a way to do lesson planning and a way to assess learners. Recently, there have been several revisions to its pyramid stack. An article on www.gettingsmart.com suggests a spiral might be better, particularly if you want to use it as a lens to view AI.

The author, Vriti Saraf, opines that the most important potential of AI isn’t to enhance human productivity, it’s to enhance and support human thinking, and that looking at AI’s capabilities through the lens of Bloom’s Taxonomy showcases the possible interplay of humans and machines.

It is an interesting idea. Take a look.