The effect of ChatGPT on KM

Although valid use cases will emerge, we must remember that none of these tools, even those labeled as “document-understanding,” genuinely understand anything; they generate something believable. AI should not replace jobs, and humans should not be in the AI loop. Instead, AI may help augment and automate some human work as long as AI is the subsidiary partner in the equation. AI is in the human loop, not the other way around.

Let’s not assume that (massive) LLMs like ChatGPT and DeepMind are the end game, or even the best solutions. Much smaller, focused, and accurate LLMs may instead offer much more promise for what is going to be a fascinating and transformative technology ride ahead. As of today, ChatGPT reminds me, on the one hand, of the IBM Watson saga, and on the other of George Santos and Tom Ripley—men who completely invented their education work history and background and learned to dress and behave the part. In Santos’s case, he even managed to convince 140,000 voters in Long Island to elect him. Tom Ripley, Patricia Highsmith’s brilliant fictional character, duped and successfully murdered his way to success. All three, Santos, Ripley, and ChatGPT, have no sense of guilt, no morals or ethics, but damn it, they are all very convincing.