Will OpenAI Fund ChatGPT’s College Education?

OpenAI’s ChatGPT, the popular chatbot, is essentially self-taught, using the vast repository of information available on the public internet as its source of knowledge. As a result, it sometimes makes embarrassing errors, such as citing imaginary court cases. One solution to this problem is to train ChatGPT on more reliable data. This raises the intriguing possibility of publishers and other firms selling high-quality data for use in training language models. Although expensive for OpenAI, such purchases could reinforce the company’s position as a dominant force in the field. However, there could also be a downside, with data, prices spiraling out of control, potentially benefiting only incumbent technology firms. As in many areas of AI, success, if it comes, will depend on the ability to balance competing issues to create affordable, reliable, and flexible AI solutions.