This week, Pew Research Center published a landmark poll showing that only 16% of Americans believe AI will have a positive impact on society. 16%. Meanwhile, 49% of American adults say they use AI chatbots, up from 33% in 2024. A quarter of them use it daily. Those two facts are not in tension with each other. They are the exact same story because nobody really is anti-fire. But if someone keeps lighting fires in your living room and telling you it's for your own good, eventually you stop trusting the person holding the mattress, not the fire itself. And right now, the people trapped in the room with those mattress are the exact people you'd least expect, creating a data paradox that makes absolutely no sense on paper.

The Pew survey is based on 5,119 US adults conducted in February 2026. 40% of Americans say AI will have a negative impact on society over the next 20 years. Only 16% say positive. 63% think AI is advancing too quickly. 71% believe it will make their personal information less secure. 67% have little to no confidence in the US government to regulate AI effectively. And this is up from 62% in 2024. The most striking finding has nothing to do with the overall negative sentiment, but rather the age breakdown. Gen Z adults, ages 18 to 29, are the most hostile to AI of any age group. 48% believe it will have a negative impact on society. And yet, they are also the generation that uses AI the most at 66%. They're using it more than anyone and hating it more than anyone. It is a beautifully toxic relationship. Gen Z is essentially using Chad GPT to write the corporate emails they are actively disassociating from while reading them. Nobody here is scared of technology. Nobody here is confused about how computers work. Gen Z is the most digitally literate generation in human history. They understand this technology better than any previous cohort understood the technologies that shaped their era. They know exactly what they're rejecting. This is informed contempt. They've used the product. They know what it does. They know what it doesn't do. And they know what's being promised versus what is being delivered. The 30 to 49 bracket and the 50 plus bracket are actually more closely aligned at 39% and 37% negative, respectively. The people who've used AI the most are the most hostile. The people who've used it the least are the most ambivalent. That's not really a pattern that any amount of marketing can really fix.

Much of this usage is not voluntary. People are using AI because it's being embedded into tools they already depend on. Their search engines, their email, their workplace software. They're using it because their bosses mandate it. According to previous Pew Research, about one in five US workers now use AI at their jobs, with managers consistently more enthusiastic about the technology than the people actually using it. People are hopping on the top of the AI hype wave out of sheer terror of getting left behind, not because they actually want to ride. AI is being shoved down people's throats whether they want it or not. Management wants to touching that little dangly thing right at the back of their throats until employees are gagging on mandatory adoption they never consented to. That is coerced adoption, not really organic demand. And an industry built on coerced adoption has a structural problem. And that is that once AI has been embedded into everything, extracting it becomes politically, technically, and commercially difficult. And users know they're being locked in. But they're not just trapped in a bad contract. They're trapped in an escalating comedy of technical errors. Because when you try to force a massive unready infrastructure into everyday software and life, things don't just glitch, they violently reject the host.

If you want to understand why the public has turned, you don't need a poll. You just need to look at what the industry has actually been doing. Slob was Miriam Webster's word of the year for 2025. The definition is low-quality mass-produced AI generated content. AI generated articles now make up over half of all English language content on the web. According to SEO firm Graphite, that's the internet. The single most important information infrastructure in human history now majority generated by machines that do not really understand what they're writing. And the companies responsible for this are not small scrappy startups. They're extra large and extra hard with ambitions to put AI right in your face whether you ask for it or not. Microsoft's aggressive integration of co-pilot into every conceivable product got so intrusive that users coined the term microsop which went viral across social media in January 2026. CEO Satiana published a blog post out in public making a scene urging the industry to move past the slope debate and it only made the term spread faster. Microsoft eventually published a kulpa titled our commitment to Windows quality and started quietly removing co-pilot branding from apps renaming AI features to advanced features. Some observers call this stealth slob AI that hasn't disappeared because apparently the industry's pullout game is weak. It just rebrands, puts on a fake mustache, walks back into the room, and introduces itself as Mr. Advanced Innovation. We see you, Sadia. We know it's you. Meta replaced Instagram search bar with an AI chatbot. Users wanted to search for people and hashtags. Instead, they got a conversational AI assistant extracting interaction data nobody asked for and nobody wanted. As Neielson Norman group skate more put it, you believe a search bar does a certain thing and then all of a sudden when you start typing in there, you're talking to an AI chatbot and you didn't want that. Meta backtracked. That's what happens when you try to park a big Mac truck right in a little garage. The thing doesn't fit, something breaks, and then you have to reverse out while everyone watches. Google's AI overviews told users to add glue to pizza and eat at least one small rock per day. It claimed that Barack Obama was a Muslim president, which really forces you to consider what kind of dark web Reddit fora Gemini was training on to conclude that the foundational elements of a balanced Italian diet involve Elmer's woodlue. These answers went out to millions of people through the most used search engine on Earth. That's what happens when you pipe unvetted AI outputs directly into an information service that billions of people trust. The Commonwealth Bank of Australia fired 45 customer service employees and replaced them with an AI voice bot. The bot failed. Call volume surged. Team leaders were pulled back into answering phones. Staff worked overtime. The bank publicly apologized and offered employees their jobs back. The finance sector union called it a cynical cover for cost cutting. I'd call it something simpler. The AI got a taste of real customer service. and now the whole operation is diabetic.

And in the developer community, the frustration is equally acute. A study on AI slop in software development found teams drowning in AI generated pull requests, 30 per day with just six reviewers in one case. Management is telling them to bring a bucket and a mob just to clear out the slop, drowning in workflows forced on them by a seauite copying AI output directly into every technical problem. One developer described it perfectly. They're literally just using you to do their job, critically evaluate and understand their AI slop and give it the next prompt. In the food chain of AI assisted development, the human reviewers are the ones getting eaten. And somehow that makes them the bottom feeders of their own profession. This is what the Pew numbers are measuring. not hostility to technology, but instead hostility to having technology shoved into every product, every workflow, every search result, regardless of whether it's ready, regardless of whether it works, regardless of whether anyone asks for it. The pressure comes from shareholders who want to see AI powered on the earning slide. Users never ask for much of this. As the Logitech CEO said about AI hardware, what's out there is a solution looking for a problem that doesn't exist. The industry doesn't cook. It doesn't clean.

If the public has turned against AI, you might expect the industry to listen, recalibrate, and rebuild trust. Instead, it's doing something else entirely, spending tens of millions of dollars to buy the political system that would hold it accountable. According to the Los Angeles Times, political finance groups linked to AI companies, including OpenAI and Anthropic have distributed a combined $37 million on 2026 midterm election campaign so far, a number that is expected to skyrocket as November approaches. And they're not picking a side, they're buying both. Open AI is heavily linked to the American Mission Pack, which has donated $8 million to Republicans, and the Think Big Pack, which has spent $14.1 million on Democrats. Anthropic is linked to the jobs and democracy pack and defending our values pack, giving $11 million and $5.2 million to Democrats and Republicans, respectively. The operating logic is essentially simple. If they got some money, then they're worth it. They talk their positions, but behind closed doors, they're hedging every bet. As former Google public policy executive Adam Kavachovich told the Times, AI companies are becoming comfortable with using their power to achieve a political goal. And the political goal is clear. Crush anyone who might regulate them. Al Oveski, a grassroots Republican incumbent in Montana, got obliterated in his primary after a super PAC affiliated with OpenAI's co-founder spent nearly $900,000 backing his opponent. Of told the Times, there was no way as a grassroots person that I could compete with that kind of money. I got crushed. Imagine explaining to your local campaign volunteers that you didn't lose because of your tax policy, but because a trillion dollar algorithmic lobby swiped your nose like a credit card and decided your political career was an optimization error. By the time voters got a look at the policy, the money had already made its mind up before it came. Beat it up AI catch a charge, spending aggressively, daring regulators to respond, and betting that nobody has the resources to fight back. Meanwhile, Democratic operatives are being warned not to upset the AI lobby, even though anti-AII positioning polls as a slam dunk with voters. This is what scale without responsibility looks like. These companies have enough money to spend $37 million on elections, but apparently not enough sense of obligation to the people whose data they're training on, whose jobs they're automating, and whose search results they're filling with rocks and glue.

And the surveillance dimension makes it tangible, too. 404 media has documented more than a dozen cases of police officers using Flock automated license plate reader cameras to illegally stalk people. In Orange City, Florida, an officer ran his ex-girlfriend's license plate through the Flock system at least 69 times. He searched for her mother's play 24 times, her father's 15. Even a colleague noticed and told him to stop. Officers who were supposed to enforce the law were using surveillance infrastructure like a private stalking dashboard, checking ex-girlfriends, relatives, and personal acquaintances through systems citizens never consented to being tracked by in the first place. And in many cases, for months, there were no consequences. The whole system went in dry and came out soggy. Cities that regret installing flock cameras are literally covering them with trash bags because they cannot figure out how to cancel the contracts. We have achieved peak tech utopia, everyone. Trillion dollar state-of-the-art AI powered license plate recognition infrastructure defeated by a 50 cent hefty bag. Get your boots and your coat because this is the landscape. An industry that surveills without accountability, spends without restraint, and deploys without readiness. And the 71% of Americans who think AI will make their personal information less secure are looking at this evidence and drawing the only rational conclusion available.

Look, I need to be direct about something. I am a computer scientist. I have spent my career working with data, algorithms, and computational systems. I am not anti-AI, and I don't think I could live with myself if I were. AI is one of the most powerful technologies humanity has ever developed. It accelerates drug discovery. It improves diagnostic accuracy in medicine. It enables real-time translation across languages. It can identify patterns in databases that would take human analysts years to find. The technology itself is extraordinary. What I need, what the public needs is not a garter snake of token regulation. We need a king cobra. Real oversight with a hook in it. Audits, fines, disclosure rules, liability, and consequences sharp enough to make companies lean over the paperwork and actually comply. Right now, tech regulators are walking into Silicon Valley boardrooms holding a damn pool noodle, asking politely if the multi-billionaires wouldn't mind being slightly less invasive. Spoiler alert, they mind.

Good deployment doesn't look flashy. It's not on an earning slide, and it doesn't require firing anyone. A customer service team that uses AI to handle routine inquiries and frees human agents for more complex cases, that is a genuine efficiency gain. A diagnostic tool that gives doctors a second opinion on imaging scans that saves lives. A coding assistant that handles boilerplate while developers focus on architecture. That's productivity. The key distinction is this. AI doesn't have to be 100% accurate to be valuable. If something gives you a 20 or 30% improvement, that is still something worth using. It doesn't mean you replace the human. It just means you augment the human so that they can be freed up to do something more valuable and more humanlike in nature. There are workplaces, by the way, that adopt exactly this kind of approach and they tend to be the ones where employees actually appreciate the technology rather than resenting it. Credit where it's due. When AI is deployed properly, the results can be genuinely transformative. But that's not what most of the industry is doing. Most of the industry is not carefully evaluating where AI adds value and deploying it with discipline. Most of the industry is gobbling up data, swallowing compute budgets, and letting the cost drip down the side of quarterly earnings reports while hoping nobody notices. They're trying to force adoption before the technology is reliable enough to justify it, degrading actual human experiences, actual products that work perfectly well just to be able to say that they're using AI. There is an element of delusion to this. Companies are so desperate to demonstrate AI integration that they'll actively make their products worse in the process. It leads to a world where AI powered becomes a warning label rather than a selling point. And based on the pew data, that world has already arrived.

There is a difference between having a small operation and running something that touches billions of lives as well. It's similar to growing a YouTube channel. Actually, when you have 500 subscribers, you're just doing your own thing. The stakes are personal. But when you have 5 million subscribers or when your platform has 4 billion users or when your AI is embedded in the search engine that half the planet relies on, you carry a different kind of weight. And it's very easy to stay in your own little bubble to treat the company as if the implications are just your own to manage. But at that scale, you are responsible for the livelihoods of the people you employ, the accuracy of the information you serve, the integrity of the elections you're spending money to influence, and the safety of the surveillance systems you're selling to police departments that use them to stalk their ex-girlfriends. Scale creates obligation. Not legally perhaps, though given the flock prosecutions, the FTC complaints, and the election spending disclosure gaps, the legal dimension is coming. But ethically, undeniably, if I hire 10 people, I would feel the weight of those 10 families acutely. I'd know that I need to bring in business or those families don't get to eat. That sense of responsibility doesn't decrease as the numbers get larger. It's the opposite. It just increases. and the apparent comfort with which trillion dollar companies discard thousands of employees, degrade millions of user experiences, and spend tens of millions buying political protection, suggests a big D, which stands for big demeanor, a confidence that is strikingly disconnected from the reality of what they're actually delivering. You can't hurt the public's feelings anymore. They've already absorbed the layoffs, the slob, the surveillance, the broken promises. What they want now is honesty, not another layer of fake reassurance. The Pew data says so. The subreddit say so. The comments on this channel say so. The people who told researchers they would pay more for a product that didn't lie to them say so. If it doesn't hang together if the product can't do what it claims it can do, then you can't bang on about transformation and revolution and the future of humanity. People can see right through it. And when people ask whose technology is it? Whom does the AI belong to? Who benefits? Who bears the cost? The answer the industry is currently giving through its actions, if not its press releases, is essentially it belongs to the shareholders. The benefits flow up, the costs flow down, and if you don't like it, there's a super PAC ready to make sure your preferred candidate never reaches office. Fire in the right hands warms. It cooks. It lights the way. It has done so for hundreds of thousands of years making human civilization possible. Fire in the wrong hands just burns things down. And the 40% of Americans who believe AI will be negative for society are not making a statement about fire. They are making a statement about the people holding the mattress. The question has never been about the technology.