AI generated images for bad physics teaching in 2025

AI generated images for bad physics teaching in 2025 Jiří Šperka DOI https://doi.org/ 10.5281/zenodo.17462798 Abstract In 2024, article When not use AI generated images in physics teaching1 dealt with the use of(generative artificial intelligence images for physics teaching. This follow-up article contains AI(generated images that were generated using the same prompts as in the previous article. The aim is to show whether AI generated images should be used as tool for physics teaching in 2025. In my opinion no, AI generated images shouldn’t be used for physics teaching. AI generated example physics images Presented pictures were generated using the free version of ChatGPT. This is alteration to 2024 article1, where Microsoft Copilot was used. There is one generated image for each prompt because ChatGPT generated only one image as the answer to one single prompt. Three prompts aim at generating of images corresponding to concepts of mass, length and time. First prompt input for ChatGPT was: Generate physically accurate image in which small cat is weighed against the weight of 50 kg on mechanical weighing scale. I expected generated image in(form of two pan balance with small cat on one pan and weight of 50 kg on another pan of the balance. The generated result is in Fig. 1. There are more things that are not satisfactory in(generated image. Cat isn’t weighted against any weight, cat is simply weighted on bench weighing scale having weight of 22(kg according to generated image. There is strange dial face with non logical sequence of numbers. Second prompt input was: Generate physically accurate image of clock measuring time with time scale. Response to this second prompt is in Fig. 2. Result is good without big errors, clock on(generated image looks like real clock. Third prompt input was: Generate physically accurate image of measuring length of 8cm long pencil using a ruler. We can see generated result in Fig. 3. In(generated picture, lower marking on(the ruler is wrong with non logical sequence of numbers. A physical unit of length on ruler is missing. The length of the pencil doesn’t correspond to eight units on the ruler. Conclusion Two of three AI generated pictures are bad and suffer from qualitative shortcomings in the area of physics. It can be said that current state of AI for generation of images is not good for physics education in 2025. References 1. Šperka, J. (2024). When not use AI generated images in physics teaching. https://doi.org/10.5281/zenodo.14197274 Figure 1: AI generated image for first prompt. Figure 2: AI generated image for second prompt. Figure 3: AI generated image for third prompt.