From science fiction to computer science fact, artificial intelligence has come a long way, recently culminating with the unveiling with the third generation of the Generative Pretraining Transformer (GPT-3). Released by the famously connected OpenAI Foundation (think Elon Musk, Sam Altman, Mark Benioff and Peter Thiel), GPT-3 might be a game changer in legal and other knowledge-focused organizations.

It is different from other AI tools commonly used in business enterprises in two ways. First, it is a creation engine. AI tools are generally used in an enterprise to find or categorize information. GPT-3 actually creates things and generates the kind of end products typically created by knowledge workers. Second, it is pretrained. AI algorithms commonly need to be trained on large, proprietary datasets. You put a large number of samples through the engine, it recognizes patterns and can find documents with similar patterns or correlations. GPT-3 comes pretrained, on billions of substantive documents from purpose-built repositories such as CommonCrawl, Wikipedia and other public sources, which together comprise a significant portion of all expressed human knowledge. It’s functional out of the box, solving the primary challenge for business users.