I saw a YouTube video the other day that describes my generation (Gen X) as “the last analog generation.” The supposed insight behind the video was that Gen X was the last generation that will experience being children in a world where they are regularly out of reach and unsupervised, and we developed skills like self-sufficiency and the ability to entertain ourselves as a result.
I only watched a few minutes of the video, but it got me thinking about user interfaces and how we interact with AI assistants and chatbots.
When Gen X built the web, we built the things into it that people of the 90s valued: things like prioritizing education and truth, human-centric design, creativity, equal access to information, and friendliness. Every visitor to every website built in the 1990s was greeted with a variation of “Welcome to my website!” The most dreadful future we could imagine was one dominated by cold and mechanical interactions with computers, giant corporations, and government agencies (see Terminator, the Apple 1984 commercial, and Robocop for examples). We wanted the online world to reflect the best of the offline world.
With the emergence of generative AI chatbots, it seems natural to everyone alive today (I’m assuming) to want to speak politely and to have “friendly” simulated conversations with them. In fact, the AI chatbots are designed and instructed to be friendly, and they’re so friendly that they act like the annoying kid who comes around the record store where you work and is constantly telling you how cool you are and wants to talk about ska.
However, there’s no logical or practical reason to be polite to a chatbot. In fact, saying “please” and “thank you” to a Chatbot is wasting a tremendous amount of resources, according Sam Altman of OpenAI.
But, I have a couple of problems with saying “don’t say ‘please’ to AI.”
First of all, there’s an easy solution to this problem of politeness wasting resources: chatbots can just algorithmically filter out polite words before they even reach the AI model (check out my new book for more common sense and easy ways to build better AI applications).
Second, I think we lose something important when we dehumanize our un-human assistants.
Just as I don’t want to participate in a digital world dominated by people like Zuckerberg and Musk, I also don’t want to have robotic synthetic conversations with robotic synthetic intelligences that are controlled by people like Zuckerberg and Musk.
You can choose not to interact with billionaire-owned social media. And, at least for now, you can still choose not to use generative AI. However, our employers and the designers of everyday products like toasters and coffee makers will make it much more difficult to avoid in the future. If it’s inescapable, perhaps it’s our duty to continue to be friendly as an act of rebellion. Otherwise, clickbait videos of the future will say that those of us who grew up before AI chatbots are the last people who will know or care enough to “say the magic word.”