We no longer speculate whether AI could process and interpret most of the visual information that we encounter. It can already do it. As AI becomes embedded in almost everything we do, a future is emerging where we could be engaging primarily with algorithmically distilled summaries of reality: filtered, simplified and curated by machines. This shift has the potential to reshape profoundly how we think, process and engage with the world.

Will this be a step towards greater efficiency and cognitive freedom or a dystopian pathway towards atrophied critical thinking and human agency?

The Age of Cognitive Offloading

Consider how humans have outsourced cognitive tasks to tools. We have offloaded memory to writing, calculations to calculators and navigation to GPS. Now, we are on the verge of offloading visual information processing itself, where we are entrusting AI with the task of scanning, analysing and summarising data into neatly packaged visuals.

Imagine receiving an AI-generated briefing that summarises everything captured by the panopticon of surveillance cameras or being given an executive summary of an academic paper complete with AI-interpreted visual charts. In our field, it could do the same for lay summaries of clinical studies or analyse radiology images where doctors are only provided with the relevant highlights.

This all sounds great, but what happens to our cognition when we no longer engage directly with raw information?

Efficiency versus Atrophy

The Efficiency Argument

Proponents argue that AI systems will:

·       Reduce cognitive overload in information-heavy environments.

·       Accelerate decision-making by distilling the signal from the noise.

·       Free humans to focus on strategy, creativity and empathetic tasks that machines struggle with.

In high-stakes clinical environments, AI can triage medical images rapidly and flag only the cases that require a radiologist’s eye. In other words, the radiologist becomes the ‘second opinion’. This means that AI becomes a cognitive amplifier, enabling humans to work smarter, not harder.

The Atrophy Argument

While we embrace all this convenience, it could encourage:

·       Cognitive atrophy: if humans rarely engage in raw data interpretation, then our analytical skills will suffer.

·       Automation bias: we will trust machine outputs more than any other, even when they are flawed. This bias is already very evident.

·       Loss of critical thinking: when we consume only pre-filtered content, then there is a real risk of accepting AI outputs at face value without questioning any underlying assumptions.

We only need to think about how many people today follow GPS blindly, sometimes leading to navigational errors, such as driving into bodies of water, that they could have avoided with basic situational awareness.

What about Visual Literacy?

One potential casualty is visual literacy. This is our ability to interpret and derive meaning from visual stimuli, such as images, maps and body language,

As AI-generated summaries become the default interface between humans and the world, would we still be able to read complex visual datasets on their own? Would we still be able to notice subtle visual cues or inconsistencies in images or videos? Would be still develop creative insights from serendipitous or non-essential visual details?

We could develop a “shallow cognition economy”, where all we do is skim surface-level interpretations but lack the tools to interrogate or critically engage with the underlying data.

AI as the New Gatekeeper

We should rightly be concerned about who controls the AI algorithms that curate these visual summaries.

Just as the search engines that shape what information we see first, AI-powered summarisation tools could create narrative biases, whether inadvertently or deliberately, where:

·       Inconvenient details are omitted.

·       Certain themes are prioritised that are based on AI’s training data.

·       Reality could be reframed to suit corporate or political interests.

So, let us consider: AI shields us from the unprocessed and undistilled visual world and our perception is mediated by the preferences of algorithms. What effect will this have on freedom of thought and transparency of our thought processes?

Lessons from our Digital Habits Today

This is not a distant, dystopic future. We can already see signs of this happening:

·       Social media feeds prioritise algorithm-curated visual content, such as Instagram, TikTok, where users consume summaries of events, emotions and opinions in bite-sized formats.

·       AI-powered news aggregators and summarisation tools, such as Google News Briefings, reduce articles to bullet-point takeaways, making deep reading less common.

·       In our jobs, we rely more and more on slide decks and dashboards instead of reading full reports.

This implies a shift away from linear reading and visual exploration towards more passive consumptions or pre-processed insights.

Kahneman’s System 1 and System 2 of Thinking

Daniel Kahneman introduced the dual-speed System 1 and System 2 thinking processes. System 1 thinking is fast, instinctive and heuristic-based. System 2 is slow, deliberate and analytical.

AI-generated summaries may push us to lean even more heavily on System 1, promoting rapid but superficial judgments. Without System 2 engagement, complex problems may receive oversimplified solutions, increasing risk of misinterpretation or confirmation bias. Imagine System 1-heavy, knee-jerk, shooting-from-the-hip decision making in an international crisis…

There is also a Utopian Scenario

This shift does not have to be inherently dystopian. There is a potential utopian counterpoint, where:

·       AI amplifies human potential by freeing us from mundane visual tasks.

·       Visual summaries become tools for accessibility, helping individuals with visual impairments or cognitive challenges to engage with the content more easily.

·       We could use the AI-filtered data as the first layer, the first coat of paint. Then we can actively interrogate it through critical thinking and human intuition.

Examples of this could be:

·       Doctors auditing AI-interpreted medical scans for nuance and context.

·       Journalists using AI to filter large video archives while retaining editorial control over final interpretations.

Could the future be Human-AI Co-Perception?

A more balanced future could lie in co-perception. This is a collaborative model where AI handles routine tasks, but humans remain as the interpreters. AI would process the raw visual input and suggest insights, while humans decide when to drill deeper, to question anomalies or re-examine the original source. This would preserve human agency while using AI to manage volumes of visual information that would otherwise be overwhelming.

Whether this trend evolves into a dystopia or an augmentation of human potential, it is a question of autonomy. It depends on how we design, regulate and culturally integrate these technologies. If we allow AI to become a black box filter with no transparency or traceability, then we could easily disengage from critical thinking and visual literacy (if not literacy in general). However, if we treat AI as a partner, we could work with it to navigate complexity without sacrificing cognitive depth.

Thank you for reading,

Mark Gibson

London, United Kingdom, May 2025