# KISS AI Safety > Source: > Published: 2026-07-12 00:47:38+00:00 One of the most important principles in engineering is the KISS principle — **Keep It Simple, Stupid**. The best engineers are the ones that rigorously adhere to this principle. The worst engineers are the ones that spawn endless complexity and write 20 microservices for a CRUD app. But the KISS principle doesn't just apply to engineering; it applies to many other things in life as well — including public communication around AI safety. I strongly believe that whenever you communicate to the public about AI safety, you should adhere to the KISS principle. AI safety people like to throw around terms like "instrumental convergence", "Omohundro drives" and "mesa-optimization" because these are the terms they are used to (including on this website). And I have nothing against those terms — technical jargon is what lets "insiders" exchange complicated ideas in a few words. But in public communication this is completely unnecessary. Because the core AI safety pitch is actually really simple, and you can present it as follows: **Big companies are currently spending hundreds of billions of dollars trying to build machines smarter than us. Nobody understands how these machines work. In fact, the evidence so far says that they lie, cheat and manipulate. If that stays true, we will either lose control of our future or all die.** This is a really simple pitch as far as pitches go. For example, the climate change pitch is much more complicated — you have to explain the greenhouse effect, for example (quick, give me a four-sentence description of the main processes behind the greenhouse effect — you probably can't). I know that this pitch is really simple because I have tried it and it works. Even on non-technical people. That's because this pitch builds on ancient intuition. "We made something more powerful than ourselves and now we cannot control it anymore" is literally as old as humanity itself. The Bible says you shouldn't play God. Everyone has watched the Terminator. Actually, on that point: AI safety people sometimes treat the Terminator comparison as an embarrassment to be preempted ("no no, it's not like the Terminator"), but I think that this is backwards. The public already has the concept of "we built super-powerful machines, we lost control, very bad". Why not use it? You can correct a few details ("no, actually, there wouldn't be robot skeletons, they would probably create a supervirus"), but the core logic is pretty accurate, no? So KISS AI safety. You don't need the words "instrumental convergence" to explain to a layperson why this is important work. You don't need to demonstrate your fluency in the vocabulary (yes, the Gray Tribe is not immune to that status dynamic, as it turns out). You don't need to sound smart. You can just say four simple sentences.