I finally got around to testing ChatGOT's ability to write Scratch code. I included our entire chat session in comments inside for anyone interested. I also got code for another project, a ball sprite that bounces around the screen, then corrected ChatGPT's code. That led to a discussion over how ChatGPT must learn from quantities of collected data rather than specific instances to protect the integrity of its data and output from bias, intended or otherwise, from a single user. Nevertheless ChatGPT erroneously recommended a Bouncing Ball style project via web address so the question remains, where did it get that address from? Especially considering that the project it linked was created Sep 2021 yet ChatGPT was launched Nov 2022. It seems like ChatGPT can be quick to contradict itself but perhaps partly out of necessity as it has been coded to avoid some topics and hide some of its own data. I guess we have to chalk some of the odd behavioral response to somebody's ghost in the machine. Lots of potential, the future is wide open.
I asked ChatGPT to write the code, got inspired by its idea of combing a text element with the Speak block and added the Say block as a third component. When I offered to credit ChatGPT for co-writing the code, this was its response.. "Thank you for the offer, but as an AI language model, I don't have authorship rights or ownership of any code or content. While I appreciate the gesture, it's not necessary to credit me as a co-author of the project. However, feel free to use the code we wrote together to create your own project and share it with others. Sharing knowledge and helping others learn is always a positive thing, and I'm glad to have been a part of your learning experience."
