Get the latest tech news

OpenAI Offers a Peek Inside the Guts of ChatGPT

Days after former employees said the company was being too reckless with its technology, OpenAI released a research paper on a method for reverse engineering the workings of AI models.

ChatGPT is powered by a family of so-called large language models called GPT, based on an approach to machine learning known as artificial neural networks. The company released code related to the interpretability work and a visualization tool that can be used to see how the words in different sentences activate concepts including profanity and erotic content in GPT-4 and another model. Bau is part of a US government-funded effort called the National Deep Inference Fabric that will make cloud computing resources available to academic researchers so that they too can probe especially powerful AI models.

Get the Android app