My Sunday essay, Adapting to the AI Revolution: Don’t Just Hand Over the Tools, Rework the Workflow, is about innovating workflows and processes as aggressively as the technology itself is being innovated. I’m interested in hearing about your experience with this topic. I have found it to be the biggest challenge my clients are facing.
In the news this morning, the two New York City lawyers who used ChatGPT to file a legal brief got hit with a $5,000 fine. If only they checked their work – you know, like law firms generally do – before relying on a word calculating tool they had no idea how to use.
It is well known that ChatGPT makes stuff up. The term of art is “hallucination,” which is a terrible term, as it implies that ChatGPT is thinking. It isn’t; it’s calculating the probability of the next best word. When you don’t explicitly constrain ChatGPT’s output (with prompt crafting or prompt engineering), it will do its best to output words that make grammatically correct sentences. It is remarkable how often it gets stuff right, and it is equally remarkable how good it makes wrong stuff sound.
If you’re interested in learning how to properly prompt craft and constrain ChatGPT’s output so that it is always useful, please sign up for our free online course, Generative AI for Execs. It will help you get the most out of ChatGPT.
Author’s note: This is not a sponsored post. I am the author of this article and it expresses my own opinions. I am not, nor is my company, receiving compensation for it.