Part 7/9:
Through insightful testing of prompts, comparisons between GPT-4.5 and GPT-4 unveiled disparities in handling complex literary tasks and contextual creativity. For example, when prompted to write a letter from Gandhi supporting an abstract concept such as an electron as a presidential candidate, GPT-4.5 demonstrated a deeper understanding of context and character development versus GPT-4, which delivered more generalized and literal responses.
Similarly, when tasked with creating a humorous rap or episode plots, the outputs reveal nuanced differences in creativity and thematic development. Factors such as how humor is structured and character voices are represented seem to distinguish the two models, although the differences may be subtle rather than stark.