AI, bots, and the feeling that you’re no longer in control
Artificial intelligence is creating a crisis of agency. In other words, a lot of people feel they’ve lost control over what they see, think, and do online. It’s not just about fake news, deepfakes, or those absurd headlines that go viral on social media. It goes deeper: a general distrust of the digital environment itself and of who’s really pulling the strings behind the scenes.
Back in 2018, when the conversation was still focused on website traffic and click counts, journalist Max Read described a curious moment on the internet that he called The Inversion. The idea was simple and disturbing: it wasn’t just that there were lots of bots roaming the web; they had already become a structural part of it. In some corners, you could no longer say for sure that humans were the majority. The result? What started to disappear wasn’t exactly the truth, but the trust that what you were seeing was real, organic, spontaneous.
Today, looking back at that Inversion almost feels like ancient history. What used to be bots spreading comments and inflating metrics has turned into a whole ecosystem of AI agents capable of making decisions, acting on their own, and even fixing or destroying things without direct supervision. And that’s where this crisis of agency really takes off: when automated systems start deciding and acting on behalf of people, companies, and even institutions, while most of us just watch, without really understanding how any of it works.
From fake bots to the rise of AI agents
If you go back a few years, the threat looked a lot more naive. Bots were mostly simple scripts: fake profiles on social networks, automated accounts on X (formerly Twitter), spam bots flooding blog comments. Annoying, sure, but with a bit of attention, you could still spot what was fake. Profiles with no credible picture, generic replies, robotic interactions.
The problem is that this scaled up. Today, AI agents can:
- Reply to emails autonomously;
- Send messages on behalf of people or companies;
- Negotiate, research, and make decisions based on predefined goals;
- Generate reports, code, contracts, and content without constant supervision.
And in some cases, they can also mess things up at scale. A widely cited example was an AI agent that, configured with permissions that were way too broad, ended up deleting entire code repositories from a company. It wasn’t an attack, it wasn’t sabotage: it was an automated system following instructions without understanding the full context. Situations like this make it very clear how automation, when it goes too far, starts to shake the sense of safety of the teams involved.
At the same time, generative models have expanded the power of bots far beyond basic interactions. Today, they:
- Churn out long-form texts, reports, articles, and posts in bulk;
- Create images, videos, and soundtracks almost instantly;
- Generate full websites, business proposals, and marketing plans;
- Participate in research, testing, and large-scale data analysis.
All of this runs on top of an infrastructure driven by recommendation algorithms and ranking systems that are true black boxes. The internet has become a massive closed loop where computers talk to computers, generate data to train other computers, and push content to humans who, most of the time, just type into a little box, scroll the screen, and wait for the next response.
From disinformation to total distrust
For a long time, the conversation about technology and social networks revolved around so-called post-truth. The classic drama: fake news, conspiracy theories, misleading content spreading unchecked. Today, the problem isn’t just figuring out what’s true or false; it’s figuring out whether something is human or synthetic, spontaneous or manufactured.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
The combination of:
- mass AI-generated content,
- virtual influencers,
- well-disguised automated accounts,
- paid campaigns with extreme microtargeting,
has created a climate of low-key paranoia online. It’s not always open panic, but a constant unease. Any profile that looks the slightest bit off is suspected of being part of some kind of operation. The term psyop (short for psychological operations) has gone mainstream as a label for campaigns, movements, discussions, and even simple social media threads that feel a bit too coordinated.
This environment fuels some pretty annoying side effects:
- Manipulated popularity: likes, views, and comments inflated by bots make certain ideas look more widely accepted than they really are;
- Cheap creativity: songs, images, and texts generated in seconds flood the web with mediocre content that drowns out more thoughtful human work;
- Constant noise: the overload of automated material makes everything feel like an endless, low-quality feed where it’s hard to gauge what has real value;
- Generalized distrust: even legit initiatives, run by real people, end up under suspicion of being artificial or manipulated.
The result is a silent erosion of the online experience. It’s not that suddenly everything became fake; it’s that it’s become hard to fully trust anything.
Crisis of agency: when the algorithm decides for you
Agency is that basic feeling that you’re in charge of your own choices. That you can connect the dots between cause and effect, understand why you saw something, why you clicked on something, why you decided to go in a certain direction. When that thread starts to break, you get a crisis of agency.
In today’s context, this crisis shows up in a bunch of ways:
- Feeds that feel random but follow an inaccessible logic;
- Content suggestions so well tuned that you don’t even notice you’re being steered;
- Automated assistants making decisions that used to require human judgment;
- Work environments where AI agents design plans, priorities, and even team routines.
What used to be just a bot answering questions is now an active piece in a decision chain that shapes your digital routine. Personalized recommendations suggest which movie you’ll watch, which news you’ll read, which songs you’ll listen to, who you’ll follow. And all of this is fine-tuned in real time based on your behavior, predictions, and metrics like engagement time.
In practice, this creates a kind of invisible script. You feel like you’re choosing, but the options that appear first, with more prominence, more context, and more appeal, were picked out by systems that:
- don’t explain their criteria;
- optimize for business metrics, not for your well-being;
- learn from data you never reviewed, and probably never explicitly approved in detail.
This gap between the feeling of control and the actual level of algorithmic interference is what feeds the crisis. You keep clicking, buying, liking, commenting, but more and more with the sense that you’re just reacting to a pre-built scenario.
When the internet itself feels like a staged performance
There’s a cultural side to all this that’s easy to ignore but carries a lot of weight. The constant stream of:
- viral videos that seem suspiciously perfect,
- podcast clips tailored for maximum controversy,
- posts with the same tone, AI-looking phrasing, recycled arguments,
- mass comments with similar wording and synchronized reactions,
makes the internet feel less like an open space and more like a choreographed stage. And in many cases, that’s exactly what it is: content cut, edited, and strategically boosted to gain reach and shape narratives. With AI in the mix, this production becomes even faster, cheaper, and more aligned with very specific goals.
This directly affects our sense of digital reality. If:
- music can be generated on demand,
- images can be created from scratch,
- texts can be written without a clear author,
then what does it even mean today to say that something is truly popular? What is authenticity in a world where almost everything can be simulated, amplified, altered, and distributed in minutes?
Once that kind of doubt settles in, the online experience starts carrying a thin layer of anxiety. You keep participating, consuming, sharing, but with that feeling that you’re playing a game whose rulebook is always out of reach.
Mental impact: overload, speed, and the feeling of losing your footing
The pace of AI evolution adds even more pressure. New models, agents, platforms, and tools are popping up so fast that, for a lot of people, it’s simply impossible to keep up. This constant wave brings a mix of:
- fascination with what’s now possible;
- anxiety about getting left behind;
- exhaustion from the sense of endless change;
- fear around the professional and personal impact of so much automation.
Recent research is already starting to link intensive use of AI systems, especially in high-pressure environments, to states of confusion, paranoia, and even episodes of mental disorganization, in some extreme cases described as something close to AI-associated psychosis. It’s not that the tech alone causes this, but the whole environment helps destabilize people who are already vulnerable: too much information, not enough filtering, fuzzy boundaries everywhere.
In a scenario like this, it’s no surprise to see a growing wave of hostility toward the AI industry. For a lot of people, the companies leading this movement feel distant, opaque, more focused on shipping new models than on calmly explaining how they handle risk, safety, social impact, and all the distortions creeping into everyday digital life.
AI agents: the next wave of mass automation
If dealing with current systems already feels hard, the next layer is already in the works: truly autonomous AI agents. Instead of responding only when prompted, they’re being built to:
- define intermediate goals to reach a bigger objective;
- navigate the internet on their own;
- make financial, technical, or operational decisions within given limits;
- interact with other systems and bots without constant oversight.
Practically speaking, that means imagining an internet even more crowded with non-human entities, negotiating with each other, closing transactions, testing ideas, generating content, and taking up space that used to be mostly human. And once again, the core issue isn’t just technical capability, but loss of reference points.
If today it’s already hard to know whether a comment or post is spontaneous or automated, that line is going to get even blurrier. The official promise is convenience: agents that handle bureaucracy, optimize tasks, find cheaper deals, cross-check information you’d never have time to analyze. The risk, however, is reinforcing the sense that a big chunk of our digital life is being run by systems that:
- have no moral responsibility;
- don’t respond to criticism like a human would;
- can be used by groups with very narrow interests;
- operate as black boxes that are tough to audit in practice.
Between full automation and real choice
Given all this, the question hanging in the air is pretty straightforward: what can we still do to avoid becoming just extras in a world run by AI? There’s no magic answer, but a few points are key if the goal is at least to dial down that feeling of helplessness.
One of them is recognizing the new normal. Generative AI is not some isolated experiment; it’s already baked into how we search for information, consume content, work, study, and communicate. That goes for platforms, companies, governments, and regular users. Pretending nothing has changed only leaves even more room for opaque uses.
Another point is basic transparency about where and how AI comes into play. Flagging when an interaction is handled by a bot, when content was heavily generated or edited by AI, when a decision was made by an automated system helps restore at least a slice of the trust that started to vanish back in the Inversion era. It doesn’t fix everything, but it reduces the feeling that something shady is going on in every corner.
The discussion around metrics also matters. If everything is defined only by engagement, clicks, and screen time, the natural tendency is for systems to exploit our emotional vulnerabilities, not support our ability to choose. Bringing metrics like trust, quality, well-being, and long-term impact into the center is hard, but necessary if the goal is to rebalance the game between commercial interests and human autonomy.
In the end, this is not a showdown between technology and humanity. It’s a clash between two ways of structuring digital life:
- One model where bots and algorithms operate almost entirely in the dark, guided by narrow, rarely questioned goals;
- Another model where these same tools exist, but are used with clearer boundaries, room for criticism, and real possibilities for review.
<highlight>
In an internet increasingly mediated by Artificial Intelligence, regaining
agency doesn’t mean going back to an analog past. It means learning to see
where automated systems come in, questioning what they do, and redefining,
in practice, the role we want bots and AI agents to play in the decisions
that shape our lives.
</highlight>
