Part 3/11:
The progress from GPT-1 (2018) to GPT-3 (2020) saw rapid advancements, with each iteration roughly doubling performance within months. However, since GPT-3.5 and ChatGPT emerged at the end of 2022, the pace has decelerated significantly. The timeframe between GPT-3 and GPT-3.5 was about 30 months, with only incremental improvements, indicating a possible plateau. This slowdown hints that AI development may be following a sigmoid or s-curve, where early rapid gains taper off over time, leading to diminishing returns.