There also have been several instances where people ‘complained’ about inaccurate responses from the AI chatbot. Even the ‘new’ Bing that has the same underlying technology used by ChatGPT, has been reported to be ‘manipulative’.
OpenAI, in a blog post, has reiterated that it wants to ensure that artificial general intelligence (AGI) benefits all of humanity and it is a work in progress.
“As of today, this process is imperfect. Sometimes the fine-tuning process falls short of our intent (producing a safe and useful tool) and the user’s intent (getting a helpful output in response to a given input). Improving our methods for aligning AI systems with human values is a top priority for our company, particularly as AI systems become more capable,” the company said.
“We therefore think a lot about the behaviour of AI systems we build in the run-up to AGI, and the way in which that behaviour is determined,” the company said. By AGI, OpenAI means “highly autonomous systems that outperform humans at most economically valuable work.”
How ChatGPT is trained
OpenAI said that ChatGPT is not like ordinary software and its training models are massive neural networks. The startup noted that AI chatbot’s behaviour is learned from a broad range of data and not programmed explicitly.
Listing out the two phases of training, OpenAI said that there is an initial “pre-training” phase, in which the model learns to predict the next word in a sentence, informed by its exposure to lots of Internet text (and to a vast array of perspectives).
This step is followed by a “fine-tune” phase where OpenAI’s models narrow down system behaviour.
It also highlighted that its models are also trained by user feedback and the company iterates its models and systems on the basis of these user responses.
For all the latest Technology News Click Here
For the latest news and updates, follow us on Google News.
