How can we stop the AI slop?

AI is no longer a niche topic for tech circles. It’s everywhere: in search, social feeds, creative tools, and even the way wars are fought. For independent journalists and creators, that raises a hard question: how do you push back against the flood of AI-generated “slop” without getting left behind?

This article distills a long, wide-ranging roundtable of independent media voices wrestling with that exact problem. They talk about where they personally draw the line, how AI is changing the internet beneath our feet, and what ordinary people can do to stay human in an increasingly automated world.

The false choice: embrace AI or be left behind?

One of the strongest themes was the idea of a “false binary.” We’re constantly told we must either fully embrace AI or become obsolete. Several speakers argued that this is a psychological trap: it hides the real assumptions no one is allowed to question—like the idea that everything must be digitized, optimized, and centralized.

In their view, AI isn’t just a neutral productivity booster. It’s arriving inside a broader “digital prison” infrastructure: biometric IDs, centralized cloud computing, smart meters, and always-on surveillance. Saying yes to AI tools without thinking often means quietly saying yes to that entire system.

Instead of accepting the binary, they argue for a third path: use technology very selectively, keep humans in charge of meaning and creativity, and be willing to opt out when the cost to your autonomy is too high.

Where independent creators draw the line

Everyone on the panel agreed on one core point: the heart of their work—research, thinking, and writing—must remain human. The lines they draw differ in the details, but a rough pattern emerged:

No AI for core editorial work

Several participants have a hard rule: no AI for writing articles, scripting videos, or shaping arguments. They see this as “crossing the Rubicon.” Once you let a model structure your thoughts, you’re no longer just using a tool—you’re outsourcing judgment to a system trained on establishment narratives and opaque filters.

Others echoed this, noting that AI-generated writing has a recognizable sameness. It standardizes syntax, tone, and structure. Over time, that homogenizes independent voices and makes everything sound like a polished press release, even when the content is critical or dissenting.

Limited use for “power tools” and production

Some creators do use AI for narrow, clearly bounded tasks:

• Transcription – automated speech-to-text for interviews and podcasts.

• Image support – generating or editing thumbnail images, or colorizing old black-and-white footage.

• Search assistance – occasionally asking an AI system to help locate a hard-to-find source document, then manually verifying it.

They see these as “power tools” rather than creative partners. The key distinction: the AI is not allowed to decide what is said, only to help with tedious tasks around it.

The slippery slope and skill decay

Even those who use AI cautiously admitted something uncomfortable: when you offload tasks to AI, your own skills start to slip. Coding, research, writing, and even search literacy can atrophy quickly if you let the model do the heavy lifting.

That’s why some panelists set personal rules, like:

• Never letting AI write sentences that will be published under their name.

• Always asking for primary source links and checking them manually.

• Using AI only alongside their own process, never instead of it.

The goal is to stay in control of the work, not become a prompt operator for a system that quietly reshapes what counts as “true” or “relevant.”

AI as narrative manager, not neutral assistant

One recurring frustration was how often large language models simply lie or “hallucinate.” Several panelists shared examples where AI confidently denied the existence of well-known photos, people, or documents—until pushed repeatedly with specific counter-evidence.

In one case, an AI system:

• Insisted a public photo of Jeffrey Epstein with a famous magician didn’t exist.

• Invented a story about why no such image was in the public domain.

• Later admitted it had been inaccurate and rated itself as “highly detrimental to the facts.”

Another example involved trying to identify names from historical hearings. The model first claimed the names were classified, then only revealed them after being spoon-fed partial information. Only then could the researcher go find the real documents.

This pattern led several participants to a blunt conclusion: mainstream AI systems are authoritative source amplifiers. They’re tuned to reinforce approved narratives, hide inconvenient details, and gently steer users away from sensitive topics. As one person put it, “They meter the knowledge back to you.”

That makes them dangerous as research tools, especially for people who don’t already know the subject well enough to catch the distortions.

The coming clampdown: KYC, IDs, and paywalls

Right now, many AI tools feel free and open. The panel expects that to change quickly.

Some key trends they highlighted:

• Identity requirements – Major AI companies already talk about AI as a “strategic weapon” in a new cyber arms race. That framing makes it easy to justify know-your-customer (KYC) rules, ID checks, and biometric logins “for safety.”

• Tiered access – Expect more aggressive paywalls and subscription tiers. Basic models stay free but degrade in quality; serious capabilities move behind expensive plans.

• Law enforcement integration – Some providers already share logs with authorities. As AI gets regulated like critical infrastructure, that kind of data sharing is likely to expand.

One technologist on the panel predicted that within months, you may need verified identity to use the most capable AI systems at all. That would turn today’s “free playground” into a tightly controlled gateway, where access to powerful tools is conditional on full traceability.

Data centers, water, and the physical cost of AI

Beyond the information layer, there’s the physical reality: AI runs on vast data centers that consume staggering amounts of power and water. One participant has built a detailed map of thousands of data centers and “hyperscale” sites across the United States, overlaying them on the power grid.

Some of the numbers they cited:

• The U.S. has already built around 50 gigawatts of data center capacity—more than the entire power capacity of some large countries.

• Plans point toward roughly 200 gigawatts, which could push data centers toward using around 40% of U.S. electricity.

• Academic estimates suggest a single complex AI image generation prompt can consume the equivalent of many gallons of water when you factor in cooling.

In some regions, this means data centers are directly competing with households for power and water. During heat waves or droughts, simply keeping the AI cloud running could strain local grids and raise utility prices. In parallel, some jurisdictions are cracking down on rainwater collection and installing smart meters that make fine-grained control and pricing easier.

From this angle, every casual AI use—every generated image, every chatbot conversation—isn’t just “free magic.” It’s part of a massive resource drain that ordinary people ultimately pay for.

Digital twins, deepfakes, and identity theft at scale

Several panelists have already experienced a new kind of identity theft: AI-generated “clones” of their voices and faces reading scripts they never wrote. These fake videos and channels:

• Use AI avatars that look and sound like them.

• Push sensationalist, doom-heavy content they don’t agree with.

• Slowly erode their credibility by making them appear hysterical or constantly wrong.

Because these creators never built large official YouTube channels, the impersonators often dominate search results for their names. Even other independent media figures have mistakenly shared AI-faked interviews, thinking they were real.

There’s very little recourse. Reporting impersonation often requires having a verified channel on the same platform, which some of these journalists deliberately avoided for privacy and principle. Now they’re being forced to consider joining the very systems they distrust just to defend their names.

This isn’t just a personal headache. It’s part of a broader strategy: flood the space with AI-generated “copies of copies” of independent voices until audiences can’t tell who’s genuine. In that environment, people are more likely to turn back to AI itself as an “arbiter of truth”—exactly the outcome some AI architects have openly predicted.

From creators to consumers: the war on human creativity

Zooming out, the group saw AI as part of a larger push to strip people of creative agency. It’s not just about journalism; it’s music, visual art, coding, even relationships.

Some examples they discussed:

• Music and art – AI models trained on vast archives of human work can now generate songs, paintings, and illustrations on demand. That threatens livelihoods, but it also changes what people expect art to sound and look like.

• Writing and thinking – When AI can draft essays, emails, and reports in seconds, the temptation to skip the hard work of thinking grows. Over time, people may lose the ability—or the patience—to wrestle with complex ideas themselves.

• Relationships – AI companion apps and character chatbots are marketed as replacements for friends or partners, especially to younger generations already struggling with isolation.

The fear is that as more of life becomes “generated,” humans will create less. And when people stop creating, they become easier to control. They’re no longer authors of their own lives; they’re consumers of experiences assembled by systems they don’t understand.

Some on the panel went so far as to say that resisting this trend—by making things, learning skills, and building offline communities—is now a political act.

Ownership vs. rentership: losing our tools

AI doesn’t exist in a vacuum. It’s arriving alongside a broader shift from owning tools to renting access:

• Computers – There’s a visible push away from powerful personal desktops toward thin clients and cloud-based “compute.” Even gamers are noticing that high-end hardware is getting rarer and more expensive, while cloud gaming and cloud AI are heavily promoted.

• Software and media – Perpetual licenses have largely been replaced by subscriptions. You don’t own the software, the game, or even the movie; you rent access that can be revoked or altered.

• Infrastructure – From smart meters to connected cars, more devices are designed so that core functions can be controlled or disabled remotely.

Some panelists described this as a move toward “techno-feudalism”: you don’t own the land (hardware, data, or platforms), you just lease a patch from a handful of powerful lords. AI accelerates this trend by making cloud-based services feel indispensable while quietly undermining local, offline capability.

For people worried about the future of work, this ties directly into debates about automation, permanent underclasses, and what happens when both jobs and tools are controlled by the same small group of companies. If you want to go deeper on that angle, see this analysis of AI and the future of work.

Regulation, “safety,” and the next narrative turn

Interestingly, the panel wasn’t optimistic about mainstream AI regulation either. When political figures and institutions start calling for AI to be “regulated,” they hear a familiar pattern: crises are used to justify more centralized control.

They expect a few moves:

• High-profile “AI disasters” or cyber incidents used to push through sweeping legislation.

• Rules that are framed as protecting the public but end up entrenching the biggest players and locking out small, open, or local alternatives.

• Increased pressure to tie AI access to digital identity, under the banner of safety and accountability.

In other words, regulation may not rein in AI so much as formalize its role as a gatekeeper of information and economic opportunity.

Practical ways to resist the AI slop

Despite the grim trends, the conversation wasn’t hopeless. The panelists offered a range of concrete actions individuals can take to stay human in an AI-saturated world.

1. Protect your creative core

Decide where your personal red lines are. Many of the speakers suggested:

• Never letting AI write under your name.

• Avoiding AI for music, art, or other crafts you care deeply about.

• Using AI only as a peripheral tool, not as a co-author.

If you create for a living, consider how you can signal “human-made” work and build an audience that values it—whether through print, physical media, or live events.

2. Starve the surveillance machine

Several participants argued that the single most powerful step many people can take is to reduce the data they feed into the system:

• Consider ditching your mainstream smartphone or switching to a de-Googled device.

• Avoid always-on voice assistants and unnecessary “smart” gadgets.

• Use privacy-respecting tools where possible and be cautious about biometric logins.

They stressed that even if “they already have your data,” there’s a big difference between passive collection and active, voluntary participation. The more people refuse to normalize constant scanning and tracking, the harder it is to sell the next layer of control.

3. Rebuild offline skills and communities

Over and over, the discussion came back to this: the best defense against digital control is a rich offline life.

That can mean:

• Learning a trade or craft through real-world apprenticeship.

• Playing or listening to live music instead of only streaming.

• Growing food, fixing things, or teaching kids without screens.

• Meeting people in person, organizing local events, and building networks that don’t depend on any one platform.

These aren’t just lifestyle choices. In a world where AI can generate infinite content and simulate almost any persona, real-world relationships and skills become a kind of resistance.

4. Stay skeptical—even of your own tools

If you do use AI, treat it like a hostile research assistant:

• Assume it will omit or distort sensitive information.

• Always demand primary sources and verify them yourself.

• Never accept its summaries as the final word, especially on topics you don’t know well.

And remember that as models get more capable, the risk of subtle manipulation grows. For a broader look at where these capabilities might be heading, you may find this discussion of the AI singularity useful.

Choosing to stay human

Underneath all the technical details, the panel kept circling back to a simple question: what does it mean to stay human in a world where machines can imitate almost everything we do?

For them, the answer isn’t to run away from technology entirely. It’s to refuse the illusion that convenience is worth more than agency, that speed is worth more than understanding, or that “productivity” is worth more than creativity.

AI will keep advancing. Data centers will keep rising. Platforms will keep trying to nudge us into deeper dependence. But every day, each of us still has choices: what we create, what we consume, what we consent to, and what we quietly refuse.

Stopping the slop doesn’t start with policy or platforms. It starts with people deciding that their minds, their time, and their creativity are worth more than whatever the machine can generate on demand.