Whose Line Is it Anyway?

OpenAI released ChatGPT to the public on November 30, 2022. The powerful chatbot is based on the company’s 175 billion parameter large language model, GPT-3. While the underlying AI model has been around for since 2020, the combination of a friendly interface and open access has propelled ChatGPT to the forefront of the AI conversation. If you haven’t already, you should try it.

For me, the impressive thing about ChatGPT is not the power of the model. It is the ease and excitement with which people are embracing the technology. Large AI models are not necessarily powerful on their own. They require people to find creative uses. Making AI accessible to millions of people is the real story of ChatGPT.

For those raising concerns about the accuracy of ChatGPT, I understand the argument. However, I think talking about what ChatGPT can’t do is missing the point. AI models will get better over time. GPT-4 will be more capable than GPT-3. What ChatGPT shows is that OpenAI is making progress towards both capability AND adoption. The former has been happening for a while. The latter is what is new.