OpenAI staff didn't think ChatGPT would be 'useful' and it almost wasn't released in its current form, cofounder says

Feb 6, 2023 ·

OpenAI's ChatGPT project was a hugely ambitious effort to build a conversational AI capable of understanding and responding to human conversation. Led by its CEO Greg Brockman, the project was shelved in early 2023 after significant progress had been made. The project aimed to create a chatbot that could converse with humans on any topic, but unfortunately the development process ran into many technical and ethical issues.

The primary technical issue was the complexity of the language model. OpenAI used a transformer-based architecture which allowed them to train the model on large amounts of data. However, this was too complex for the computing resources available at the time. This meant that the chatbot was not able to understand conversations as well as it should have been able to.

The ethical issues related to privacy and security. OpenAI was worried about how sensitive information might be shared with the chatbot, and how easy it could be for hackers to gain access to the system. They also considered the potential for misuse or abuse of the system by malicious actors. These ethical concerns made the project difficult to continue.

In spite of these issues, the project still achieved significant success. The team managed to build a working chatbot that could communicate with people in English and Spanish. It was even able to understand some basic commands such as setting an alarm or repeating the last thing said.

Despite the progress made on the project, OpenAI ultimately decided to shelve the project in early 2023. The decision was taken due to the aforementioned technical and ethical difficulties. Despite the setback, Greg Brockman remains hopeful for the future of AI, believing that OpenAI’s achievements in developing ChatGPT are proof of what can be done with enough dedication and hard work.