The Rise of Systemised Stupidity: How Social-media Algorithms Trained Us to Stop Thinking—and AI Could Help Us Start Again

AI Is Being Adopted at Speed. Judgement Is Not.

The world has already spent 2.5+ decades rehearsing and intergenerationally surrendering attention and life.

By late 2025, global social-media user identities had reached 5.66 billion—almost 69% of humanity—and had grown by another 259 million in a single year. Among internet users aged 16+, 96.9% now use at least one social platform monthly. DataReportal, Digital 2026. This is not a minor lifestyle trend. It is a civilisational operating environment.

Every day, billions of people, get out of bed, reach for the phone, plug themselves in, enter systems designed not to improve themselves or judgement, deepen understanding or serve the public good, but to maximise attention, render them submissive, provoke short decreasingly lower-cognitive level reactions and acquite return visits. The algorithm does not ask, “What most deserves your thought and how do I develop you?” It asks, “What is most likely way I can reduce you to a submissive state, keep you captive and here?”

Breakfast habits and photographs of meals appear harmless. Holiday updates are harmless. The occasional unusually detailed account of domestic life seems harmless. But the wider pattern is not. A population accustomed to allowing algorithms to select its attention can become accustomed to allowing systems, crowds and confident strangers to select its priorities, beliefs and emotional responses.

The lust for self-promotion—the craving to be noticed, affirmed and made visible—can become confused with value and reality itself.

The lust for self-promotion—the craving to be noticed, affirmed and made visible—can become confused with value. In an environment used by 5.66 billion social-media identities, where the typical connected adult now spends more than 18½ hours each week across social and video feeds, attention is not merely available: it is engineered, measured and rewarded. DataReportal, Digital 2026

Likes, thumbs-up, smiles and follower counts become confused with truth; repetition becomes confused with evidence; promotion becomes confused with identity. The pathology is not that everyone who posts is narcissistic—clinical labels should not be casually applied—but that platforms can reward narcissistic traits, external validation-seeking, performative certainty and a fragile dependence on audience response.

Research identifies an association between narcissism and problematic social-media use, while cautioning that the relationship is complex and not diagnostic of any individual. Systematic review

Narratives begin to feel real because they are visible, repeated, amplified, manipulated, applauded, victimised and socially reinforced. But visibility is not value. Engagement is not truth or reality. And a widely circulated performance of identity and self-aggrandising puffery is not evidence of substance, competence or reality.

https://video.wixstatic.com/video/abf79d_7561ee0038ed46659a31563b91800e1b/1080p/mp4/file.mp4

This is not an accusation against every social-media user. It is a warning about the architecture around them. Social platforms easily pray on, seduce the susceptible and have industrialised cognitive convenience: the capacity to feel informed, involved and morally active without doing the slower work of testing claims, understanding rerality, causes, seeing consequences or changing one’s mind. Now AI has entered that same environment. And this is where the decision becomes profound.

AI can either become the next layer of total cognitive outsourcing—Across the workforce, AI can draft communications, summarise documents, triage enquiries, schedule work, analyse data, support customer interactions and automate routine tasks—often with a speed and fluency that can obscure thinking or whether the work, decision or outcome has been properly understood—or it can become the greatest systems-thinking instrument humanity has yet built.

At present, the evidence says adoption is running far ahead of transformation.

McKinsey’s 2025 global survey found that 88% of respondents said their organisations were using AI regularly in at least one business function, up from 78% a year earlier. Yet nearly two-thirds had not begun scaling use-case driven AI across the enterprise, and only 39% reported any enterprise-level EBIT impact. AI is everywhere in presentation; it is still comparatively rare in operating-model redesign. McKinsey, The State of AI in 2025

That is the flapping-outcomes problem.

Organisations are buying licences, crafting frameworks, running (or ruining) pilots, announcing copilots, creating AI policies, conducting workshops and producing a growing volume of AI-assisted material. But much of this remains single-loop activity: faster production inside the same inherited machine.

The report is written faster. The meeting is summarised faster. The marketing copy is generated faster. The spreadsheet is analysed faster.

• But why was the report required in the first place?

• What decision does the meeting actually improve? What behaviours does the metric create?

• Where has accountability been lost? What human outcome has the process ceased to serve?

Those are different questions. They demand systems thinking.

The World Economic Forum now identifies systems thinking as a capability expected to solidify in importance by 2030. It also reports that analytical thinking remains the leading core skill identified by employers, regarded as essential by seven in ten organisations; 39% of workers’ existing skills are expected to change by 2030. World Economic Forum, Future of Jobs Report 2025

The market has noticed the need. Education has not yet responded at the required depth.

We still educate most children through a broadly Victorian-era or industrial model- subjects divided into silos; kids controlled in rows, teacher at front, knowledge delivered in batches; progress measured through discrete answers; compliance rewarded; critical thinking, denied, immersion-ignored, complexity treated as an inconvenience; and success too often defined by moving a cohort through a standardised sequence where – madly – year on year results are compared with the last – all that ignore context and societal demise.

There are excellent educators, highly emergent and innovative schools and valuable cross-curricular programmes that deliver real world learning and incredible transformative outcomes - that bind new communities together. Absolutely. But systems thinking is not yet treated as a universal, continuously enabled, taught and rigorously assessed foundational literacy in the way that reading, writing and mathematics are. The dominant architecture remains mass delivery and assessment churn. A few crawl through the cracks. That mismatch is becoming dangerous.

Children are entering an age in which information is abundant, synthetic content is cheap, false confidence is scalable and every individual can access persuasive answers at negligible cost. Yet many are not being systematically equipped to ask the most important questions-

• What system produced this outcome?

• What incentives are driving the behaviour?

• Who benefits, who bears the cost, and what is being displaced?

• What evidence would change our view?

• What consequence will appear elsewhere if we “solve” this problem here?

Without human-use case alignment and those capabilities, AI does not create a more intelligent society. It creates a more articulate one—more fluent, more productive, more confident and potentially less capable of recognising when it is utterly wrong.

That is the risk of systemised stupidity: not a lack of intelligence, but intelligence fragmented from context, consequence, ethics and whole-system understanding.

A person can be highly qualified, professionally successful and technically fluent while still being unable—or unwilling—to see the incentives and feedback loops governing the world around them. An organisation can digitise every workflow and still be blind to the fact that its operating model is creating the very failure it then spends money trying to manage. AI will amplify whichever condition it is given.

Give untrained retail Ai fragmented data, shallow goals, poor incentives and untested assumptions, and it will help a person and any institution perform its total dysfunction at extraordinary speed. Give it a clearer purpose, reliable evidence, human challenge and whole-system visibility, and it can expose relationships and consequences that no individual team could see alone. That is the personal and strategic choice.

In an organisational sense -

Single-loop AI adoption asks: How do we do the present task faster?

Double-loop AI adoption asks: Why do we do this task at all, and what assumptions govern it?

Triple-loop AI adoption asks: What should this system exist to achieve—for people, society and the future—and how must we redesign its measures, incentives and decisions to serve that purpose?

The first produces efficiency.

The second produces redesign.

The third produces maturity.

This is the real opportunity of the AI-enabled exponential age- a systems-thinking renaissance. Not a return to slower work, but a move to deeper work. Not anti-technology nostalgia, but the disciplined refusal to let technology replace human judgement.

It means designing deliberate friction into decisions. Form a view before consulting the model. Test the model against lived experience and contrary evidence. Make dissent safe. Trace unintended consequences. Require leaders and boards to distinguish output from outcome, activity from progress, and confidence from truth.

The organisations that win will not simply have the most AI tools. They will be the ones that use AI to see the whole system: its risks, feedback loops, blind spots, dependencies, incentives, human effects and emerging opportunities.

The same is true for society.

We do not need more people merely able to consume a tool, generate an answer, post gibberish, or an opinion or repeat an algorithmically supplied certainty of submissiveness.

We need people who can see beyond the feed, beyond the immediate metric and beyond the inherited process.

AI can help build that capacity. But it cannot choose to do so.

That remains the real human task- to use unprecedented computational power not to automate mass delivery of thoughtlessness, but to build better cognitive and applied judgement, better systems and better outcomes.

Not faster churn. A deeper civilisation.