Microstructure segmentation is a crucial task in materials science which facilitates detailed material characterization and establishing processing-structure-property linkages. Image segmentation is a classical technique in computer vision for which several deep learning models exist. Furthermore, transfer learning can leverage the performance of these deep learning models on new datasets by initializing them with parameters pre-trained on large datasets. In this chapter, transfer learning is applied for segmentation of nickel-based superalloy microstructure images using a model pre-trained on a large dataset of microscopy images called MicroNet. GPT-4 on ChatGPT Plus is instructed to generate a Python code for performing this task. By a suitably designed sequence of prompts, GPT-4 provides promising results in implementing the task. Aspects of prompt design, handling errors and testing the outcome are considered as well.