Non-sequitur statements like this drive me nuts. Somehow, politicians and executive types learn how to use just enough of them to make the audience forget what they're not saying.
It's quite funny (not sure if ironic) in the context of OpenAI, ChatGPT can do exactly the same thing: generate a string of sentences that from cursory skimming might sound about right but when reading with attention you find all the cracks and incongruences in the generated text.