Converting the text in an image to a Google Doc with a lot of help from AI

The photo, circa 2018?

Years ago, in a different classroom, I had a glorious long whiteboard. It stretched for nearly the full length of one side wall of my classroom. One of my favorite things to use it for was listing sentence frames my 9th graders could use to help them craft body paragraphs of an essay. To take a picture of it I had to use the pano option on my phone, but I am so glad I took that picture. 

Fast forward to this past spring when my teaching partner suggested we revive the comparative analysis uint we used to do. I knew I would need these sentence frames again, but my current classroom (while beautiful) has only an eight foot whiteboard—not long enough to recreate this helpful scaffold for my current students. 

I decided a digital document could suffice and also be available to students outside the classroom through our LMS. Arguably an improvement. But, it would mean accurately retyping these sentence frames into a table of a Google Doc. Not something I was excited to spend time on. 

Can AI help with this?

I wondered if Claude.ai could do it for me. I was pretty sure it could probably extract the text, even with the distortion of the panoramic photo, but I also wanted to pivot the text so that the yellow banner at the top was a column on the left side with the sentence frames in a column on the right side. This would be a better orientation for a document. 

The prompt I used with the picture attached: 

Even though I suggested Claude ask me more questions about the task, it didn't need to and just produced the result. 

What Claude.ai did from my photo

Note that the result is in 'markdown,' a kind of text language that uses symbols to specify formatting. 

Enable Markdown

I've met lots of folks who are frustrated that when the copy/paste results from AI it comes out with lots of extra ## and other symbols—meaning time wasted cleaning up the text. 

For a Google doc there is an easy way to fix this issue. Got to Tools/preferences and then check the box that says "Enable Markdown." Then choose "paste from markdown" in the edit menu. Control+V should also work. 

Do we need AI for this?

Converting hand written material to typed text is not new. Optical character recognition has been a thing for years now, but I've never found a truly smooth work fro for it. Also, in this case, the pivot was critical to making the text work well on a doc. 

I haven't ever seen a tool for converting handwritten charts to text in any of the widely available teacher-facing AI tool boxes. So, I'm guessing this isn't a high demand use case, but it sure saved me a lot of time, and made the material quickly available to my students. 

I make a lot of charts in my classroom. I archive photos of them on my Classroom Charts blog, so that I can add them to lessons easily. Now I'm thinking I'll leverage AI to convert more of them to text that will be more accessible to screen readers, easier for students to copy/paste from (when I want to let them do that) and easier to share as Google docs. 

Could I do that by retyping and formatting them myself—yes, of course. Have I ever done it—no. But with AI assistance to recreate the charts quickly, I will likely digitize more in the fall. 

PS AI did not write any of this post—I just love love the em dash.