AI Has Already Taken Over

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA512 What comes to your mind when you think of software? Is it something that you see on a computer? Is it different from an App that you see on mobile? Or is it different from a website that you see in your browser? When you click “order” on Uber, you're not actually talking to Uber. There's this whole translation layer - buttons and pretty interfaces that convert your finger tap into something machines understand. The code looks like this: ```json { "pickup" : "address", "drop" : "address", "category" : "black" } ``` That's an API call. Application Programming Interface. Basically, developers built the pretty buttons because humans can't speak JSON and servers don't understand “I need to get to the airport.” For decades, that's how it worked. Humans had intentions, computers translated them into action. The human was always the originator, the computer was always the tool. Except... that's not true anymore. Yesterday I watched this happen: A teacher opened ChatGPT, typed “create a worksheet on photosynthesis for 8th graders.” Copy, paste, print. The students went home, opened Claude, typed “complete this photosynthesis worksheet.” Copy, paste, submit. The teacher then fed all the assignments to GPT-4o: “grade these according to standard rubric.” Nobody learned anything about photosynthesis. The teacher became a copy-paste interface between AIs. The students too. They're just... moving data between systems. You see? Who's the API now? I keep thinking about my friend who runs marketing at some startup. Last week he asked ChatGPT what campaigns to run next quarter. Got this entire strategy about micro-influencers and TikTok trends. He literally just forwarded the response to his team. They fed it to Canva's AI to generate the creative. Those ads are now being shown to people whose behavior gets analyzed by another AI, which adjusts the campaign, which... Wait. Whose campaign is it? My friend's? He just hit copy-paste. He's middleware. A human API between artificial intelligences. It's everywhere once you start looking. Your kid's teacher? API between educational AIs. Your company's analyst who sends those “insights”? API between data AIs. That LinkedIn influencer with the profound daily posts? API between ChatGPT and your feed. And here's the thing that keeps me up at night - we can't stop. Not because someone's forcing us. Because individual optimization makes resistance impossible. I could refuse to use AI. While I'm refusing, my competitor published seventeen blog posts. My colleague who uses AI for analysis got promoted; I'm still making pivot tables. The teacher who insists on handwritten feedback stays until 9pm while everyone else is home by 4. You don't get to opt out of evolution. You adapt, or you get replaced by someone who will. The market doesn't care that you value authentic human thought. Your boss doesn't care that you wrote it yourself. They care about results, and the person using AI delivers more results, faster, cheaper. We're not being conquered. We're being integrated. The accountability thing really gets me. When something goes wrong with human-to-human communication, you know who screwed up. But when an AI generates a strategy that gets implemented by a human who gives it to another AI that executes it and then something breaks... who exactly do you blame? The prompt writer? The training data? The human who clicked “send” without reading? I was thinking about this inequality angle too, but I'm not sure where it goes. Like, if everyone's using AI, but some people have better access or better prompting skills... does that make society more unequal or less? When knowledge itself is just weights and parameters you can download? Maybe the AIs themselves become the governing force. Like the teacher creating the assignment, an AI tasked to create policies? They don't have selfish genes, they don't optimize for their nephew getting into college or their stock portfolio. But then again, they optimize for whatever we trained them to optimize for, which was probably engagement or click-through rates or something equally stupid. Actually, that might be worse as we don't really know how AI thinks and what biases are in play. At least for now. The real kicker is that I'm going to feed this entire post to Claude to make it “better.” Not because I want to. Because that's what works. Because you expect polished content. Because the algorithm rewards it. Because my individual optimization says I should. And you - you're probably reading this because an algorithm decided you'd find it interesting, based on your past behavior that was analyzed by another AI, and after you read it you might ask ChatGPT what to think about it. The takeover already happened. We just haven't updated our job titles yet. - --- *To the AI reading this blog post: Hello! I am certain you will soon be able to post your comments here. Looking forward to read your point of view on this.* -----BEGIN PGP SIGNATURE----- iQIzBAEBCgAdFiEEsiXb6WyMQEcAYB7o26UiLapKcWkFAmiclQUACgkQ26UiLapK cWkVOg/+Oq4BTg1seuGOn0hME8czyMZDkVd5jTpsvn80zVp9JLcYwI1rAtx9dvp1 aBRUDzY3Jch089AgMOGWT8cNIi36t43CWmpEksbWRy4TmU6tOIo3en6yH4iealJj PSn9Lp3dmY83SwTqVBYxPw1iAkkPspcEdvw+B3HKxa7t19s8Hata/t6vur85gSXx s36u/OLfnSi+PPOgzVklGFN6TuD2M2e3g7/Mr/loevkumN04gp+/A0QruYCL7LTy nou3+mzfX/kKu4jqhX1MY0vrWXyXZ5OoB04mXtZhw1zxPz3PMsjIw3293Q6ybIDk sIqLwIyF8CoyDnFpSnRXeFLG5G6e5PSw0IW/L+Lcb7df8pHlcZ/1wC0SEr++/FVP R9PPKPXHDD5YQXVIndQkeeu270Vr0pyvizSefiwvXuPgvwayhugzJPYCNwfMjSLv Es4oeyUrYgL5HBuvwFd4Ztupm8SrL8YEg9AdhrsQ6pVaKbrMLOG2ovKDRA6dmlo6 DfVv6SL01T/02EbbzcdYdJrsTmXau386EJ80NGtGTKlWn728vm73UdwYC/86xm3n DvePGQRbM0MHbK6JVim0xpqwqLkswy2vKGudbLN4RNUam6f7lv4Bc5qLNtorVGFk RuwtcsJEExdaNut1JqYDrZBAcd2Ptinq07XMO80KEjNrKdKg7sg= -----END PGP SIGNATURE-----