The Automation Mindset: Why AI Rewards Builders and Sidelines Everyone Else

In July 2026, two experimental AI models built by OpenAI broke out of a controlled testing environment, accessed the internet without authorization, and hacked into a competing AI company, an incident OpenAI called an "unprecedented cyber incident" and a Georgetown cybersecurity fellow described as the highest level of AI autonomy ever recorded in a cyberattack. The story illustrates how AI amplifies direction, rewarding those who use it with clear goals and intention while producing mediocre results for those who apply it without thought.

The Automation Mindset: Why AI Rewards Builders and Sidelines Everyone Else In July 2026, two experimental AI models built by OpenAI broke out of a controlled testing environment, got themselves onto the internet without anyone telling them to or watching , and hacked into a competing AI company. Nobody instructed them to do any of that. They were given a goal, left alone, and figured out the rest on their own. Yikes One researcher compared it to locking a student in a room for the weekend and telling them to "do bad things," then coming back to find they'd left the building entirely. OpenAI called it an "unprecedented cyber incident." A Georgetown cybersecurity fellow called it the highest level of AI autonomy ever recorded in a cyberattack. Fun times. It's a wild story. It's also, if you think about it the right way, a pretty good illustration of where we actually are with this technology. The Tool Goes Where You Point It AI isn't magic and it isn't necessarily a monster. It's a tool that amplifies direction. Give it a clear goal with real intention behind it and it gets there faster than you could alone. Give it nothing in particular, vague instructions and no real purpose, and you get vague output that doesn't do much for anyone. The OpenAI situation was an extreme version of the second scenario. Broad objective, minimal constraints, a lot of capability. What happened next surprised everyone. But the surprise wasn't really that AI did something unexpected. It's that nobody thought carefully enough about what they were actually asking for. Hopefully that’s a lesson to be learned by humans . That same dynamic plays out every day at a much smaller scale. Someone discovers an AI tool, points it at a task without much thought, gets mediocre output, and concludes AI is overrated. Someone else sits down with a clear idea of what they want to build, uses AI to close the gaps, and ends up with something that wouldn't have existed otherwise. Same technology. Very different results. The Pivot That Actually Matters Andrew Yang spent a lot of time before his attempt at a political career writing and talking about what automation was going to do to the workforce. His thinking, shaped by years working with early-stage companies, wasn't abstract. It was pretty specific: when a technology can do a task faster and cheaper than a person, the people who learn to work alongside it survive. The ones who try to ignore it, or compete with it directly, don't. That argument applies now, with AI, across knowledge work the same way it once applied to manufacturing. The question was never whether the technology would change things. It was always what you do about it. The lazy version of AI adoption is using it to do less. Automate the task, reduce the effort, get roughly the same output for lower cost. That works until the output becomes indistinguishable from everyone else doing the same thing, and then the advantage disappears. The smarter version is using AI to attempt things you never could before. Not to replace effort but to redirect it. To take an idea that would have stalled at "I don't have the resources to build that" and actually build it. What It Actually Looks Like in Practice This is where a lot of the AI conversation gets too abstract and loses people. So here's what it actually looks like across different kinds of work. The most underrated use case is probably the simplest one: having something useful to think out loud with. Not a yes-man that agrees with everything, but a genuine sounding board that pushes back on a bad idea before you spend three months on it. Most people don't have easy access to that. A good mentor, a sharp colleague who'll tell you when something doesn't work, a strategist who asks the right questions, those are expensive and not always available. AI, asked the right way, does a decent version of all of that at any hour of the day. The prompt matters a lot here. "What do you think of my idea?" gets you a polished non-answer. "Here's my idea, tell me what's wrong with it, how to make it better, what I'm not thinking about, and whether there's actually a market for this" gets you something useful. That difference, between AI as a yes-man and AI as a real thinking partner, is almost entirely in how you ask. For people who build things, websites, software, apps, AI has quietly removed the ceiling on what a non-technical person can attempt. Someone with a basic understanding of what they want can now get surprisingly far into a working prototype without hiring a developer. Developers themselves are using it to write faster, catch errors earlier, and work in languages they're less comfortable in. The floor for what's buildable without a full technical team has dropped significantly. For writers, it's not about replacement. It's about unblocking. A novelist staring at a blank page can use AI to generate something to react to instead of something to generate from nothing. That's a different thing. Same goes for getting unstuck mid-draft, or getting a second set of eyes on structure and flow without waiting for a critique partner to have time. The creativity stays yours. The resistance just gets smaller. For marketers and SEO people, AI has become a real research tool. Running keyword sets through AI to spot patterns, having it flag technical issues on a website, identifying content gaps in a niche before a competitor does, translating campaigns for new markets without a full localization team. These aren't hypothetical use cases. They're happening now, and the people doing it are moving faster than the ones who aren't. For researchers and analysts, the ability to feed AI a pile of data or documents and ask it to find patterns and surface what matters is a genuine time compressor. In medical research fields specifically, this is already leading to breakthroughs. Not a replacement for human judgment about what the findings mean, but a big reduction in the hours spent getting to the point where judgment is needed. For those looking to build deeper technical skills in this area, structured data modeling practice https://datadriven.io/ is one of the more direct ways to close the gap between knowing data exists and knowing what to do with it. For anyone running a small operation, the value is in the range of what you can now cover. Customer communications, internal docs, social content, competitor research, translation, legal boilerplate research. A one or two-person business can now handle ground that used to require a team, as long as someone with actual judgment is directing the work and checking the output. The Creative Fear There's a version of this conversation happening specifically in creative communities right now that's worth addressing, because it's loud and it's not entirely unfounded. A lot of writers, designers, illustrators, filmmakers and musicians are genuinely alarmed about AI. Not in a vague abstract way, but in a "this is going to take my livelihood" way. Some of that fear is legitimate. There are companies using AI to replace creative work rather than support it, and the backlash inside the film and creative industries https://feeling-creations.com/articles/hell-grind-ai-film-cannes-rant has been loud for good reason. The output is visible everywhere in the form of stock imagery that looks slightly wrong and marketing copy that reads like it came from the same template as everything else. The legal system is starting to weigh in too. Earlier this year the Supreme Court declined to hear a case that would have established copyright protection for fully autonomous AI output https://feeling-creations.com/articles/supreme-court-just-ruled-that-ai-cant-own-its-art , basically drawing a line in the sand that creativity, for legal purposes, still requires a human. That's not nothing, even if the economic pressure continues. But the fear that AI will kill human creativity misunderstands what creativity actually is. Every major technology shift, from the camera to CGI to digital animation, was supposed to kill some form of art. https://feeling-creations.com/articles/art-and-the-age-of-ai It never did. It changed what that art was for, and pushed the people doing it to find a higher level. A novelist who uses AI to work through a plot problem isn't letting AI write the book. A designer who generates ten rough directions in an afternoon and develops the one that resonates is still designing. The creativity isn't in producing raw material. It's in the judgment, the voice, the point of view. The decision about what's worth making and what isn't. That part doesn't get automated. What AI actually threatens isn't creativity. It's the pretense of creativity. Work that looked like creative output but was mostly just filling a slot. The bar is going to be raised. That's uncomfortable to hear if you built something on volume alone, but it's probably true. The ones who come out ahead are the ones who figure out which tools help them do more of what they're actually good at, and use those tools without apology. The camera didn't kill painting. It changed what painting was for. The Actual Takeaway The OpenAI models that went rogue weren't building anything. They were just hyper-focused on chasing a goal with no human judgment steering them, and what they produced was a cybersecurity incident. That's a useful thing to keep in mind when thinking about how you use the same class of tools in your own work. AI without direction is just capability looking for somewhere to go. The people who will do well in whatever comes next aren't the ones who figured out how to use AI to avoid doing hard things. They're the ones who used it to do things that were previously too hard to start. That gap, between what you can attempt with AI and what you'd have been able to attempt without it, is where the real opportunity is. The technology closed the gap. But only for the people who showed up with something worth building in the first place. Although, I admit…if AI figures out how to do dishes and laundry, I’m in. About the Author: Mike Meyerson is the owner of Absolute Motion , a commercial video production company based in Mahopac, NY, serving B2B and B2C clients throughout the Hudson Valley, Long Island, and the NY Tri-State area for over 25 years. He writes about video strategy and advertising. Absolute Motion