Software Engineers: “Do No Harm”

Let me first clarify that I’m not opposed to AI. I’ve personally architected, coded, and trained many models. I use ChatGPT, Stable Diffusion, and other generative technologies more far more than the average person. These are powerful tools that absolutely have their place- AI helped me edit this post, for example. I’m cautiously optimistic that with the right guardrails and careful product design, AI could augment human creativity, improve education, and even support mental health. Without those guardrails though, today’s AI is a real dystopian shit show.

Meta recently debuted AI profiles. Fortunately, there was intense backlash (score one for humanity), and the feature was shut down. What troubles me about this is not the use of AI itself, but the intermingling of AI and social relationships: the explicit intention that bots should replace our friends. Indeed, when one friend of mine showed these profiles to her dad, he didn’t understand that these were not real people. It should be obvious to any product designer that a tiny “AI managed by Meta” disclaimer is not enough to clearly differentiate real humans from fake ones. Make no mistake: these were intentionally designed to be mistaken for humans, and that should infuriate us. I previously lamented about the decline of society and technology’s role in that. We urgently need our technology to support communities and (real) human connections, not surreptitiously replace them with lookalikes!

I’m more familiar than most about how these kind of things get built. At different points of my career, I’ve been the software engineer, AI expert, product manager, UI/UX designer, QA, management. My wife previously worked on LLMs at Meta. So, I gotta wonder: who were the folks who built this and thought “yup, seems like a great product. No concerns here- ship it!”. Where are our ethics in tech?

Leave a Comment