As anyone with a Twitter account knows, online arguments aren’t debates. They’re reading comprehension tests – tests which most people are failing. They’re also the grand validation of an old rule of communication: it doesn’t matter what you say – it matters what people hear. What you said doesn’t matter. What they think you said, often wrong and commonly idiotic, is what they respond to. But that, unbelievably frustrating, gap between what you said and what they heard isn’t random. It’s structural. And, frankly, expecting it to be otherwise in that environment isn’t optimism – it’s category error.
But that kind of miscommunication isn’t just online dysfunction. It’s a symptom of something measurable – the fact that the majority of the population of Ireland, and many other countries, are functionally illiterate. How can this be, you might ask, when Ireland is said to have “one of the highest literacy rates in the world”? Well the answer there lies in how we define literacy—and how we measure it.
Every so often, someone cites Ireland’s high literacy rate – often reported as 99% – as proof of the success of the Irish educational system, government policy, etc. But functional literacy – being able to extract meaning, compare information, and navigate real-world documents – is another matter entirely.
The most honest look at this comes from the Programme for the International Assessment of Adult Competencies (PIAAC), an OECD survey run in Ireland by the CSO. PIAAC measures what adults can actually do with text, numbers, and unfamiliar problems.
This isn’t about whether you’ve read Ulysses – a book often lauded as the greatest piece of literature ever written, but which in truth isn’t even the best thing Joyce wrote – or the level of qualification you’ve received. This is about whether adults can function in a society governed by forms, instructions, portals, and interfaces.
PIAAC categorises adult literacy into five levels, each measuring progressively more complex and independent reading skills:
In 2023, the breakdown in Ireland looked like this:
| Literacy Level | % of Irish Adults |
| Below Level 1 | 5% |
| Level 1 | 16% |
| Level 2 | 38% |
| Level 3 | 32% |
| Level 4 | 8% |
| Level 5 | 1% |
Only 1% of Irish adults reach Level 5 – the top tier of literacy. At this level, individuals can handle deeply complex materials, synthesise and evaluate information from multiple sources, and solve unfamiliar problems using abstract reasoning. These are the people best equipped to operate in high-level bureaucratic, academic, policy, or technical environments. They are extremely rare – roughly one in a hundred.
Another 8% reach Level 4. These individuals can interpret conflicting information, follow complex arguments, and work across multiple documents. They’re capable of managing complexity, but not necessarily at the highest level of abstraction or novelty.
Only 9% of Irish adults operate at Levels 4 or 5 of literacy – the threshold where advanced reasoning, abstract thinking, and deep comprehension come into play. And yet, that narrow cognitive tier is the one our world is increasingly designed for.
Over half the country is below Level 3. That means they may struggle to follow instructions, interpret a document, or fill in an online form.
We’ve built a society for the top decile and left the rest to improvise. Things that used to happen in person now happen through portals. Signatures are replaced by multi-step verification. Conversations by interfaces. The systems look modern and efficient on paper, but they weren’t built for real people.
These systems are often sleek, optimised, and, in theory, well designed, but a shocking number of them fail when they eventually meet actual people. Or worse: they succeed, but quietly make life just that little bit harder for a sizable chunk of the population who either struggle to interact with them or simply don’t interact with them because they find it either trying or embarrassing.
PIAAC research makes it clear: Level 3 is the minimum literacy threshold for effective, independent functioning in modern society. Below this point, individuals are increasingly limited in how they can navigate bureaucracy, employment, healthcare, education, and digital systems, often needing support from families members or other people known to them.
And that’s where 59% of the Irish population is – level 2 or below.
That reality sits uneasily beside a cultural reflex we all recognise – the impulse to insist that we’re all equally capable. It brings to mind Garrison Keillor’s old line about Lake Wobegon: “That’s the news from Lake Wobegon, where all the women are strong, all the men are good-looking, and all the children are above average.”
No, not really.
There have been two PIAAC surveys carried out in Ireland in the last decade, and technically the mean score for adults in Ireland slightly improved between the 2012 and 2023 surveys – albeit not in any statistically significant way – but the actual trends we can see do not look terribly positive.
In 2012, 17% of Irish adults were at or below Level 1 in literacy. By 2023, that figure was 21%. The amount of adults at level 3 fell by 4%, a fall only partially counterbalanced by a 1% growth at level 5.
| Literacy Band | 2012 (%) | 2023 (%) |
| At or Below Level 1 | 17 | 21 |
| Level 2 | 38 | 38 |
| Level 3 | 36 | 32 |
| Level 4–5 Combined | 8 | 9 |
If you’re below Level 2, reading anything more complex than a cereal box is going to be a challenge. Not an impossible challenge, but it’s going to take an active effort. And so things like medication labels, bank letters, official forms – these become points of failure.
Some chalk this up to education. And yes, education matters. But PIAAC also functions – quite accurately – as a proxy for IQ, especially in the lower-to-moderate range. It measures how people deal with ambiguity, extract patterns, and apply reasoning. These are cognitive functions, not just academic skills.
Cognitive competence is now the foundation of independence. And if you lack it, the world isn’t designed with you in mind.
And so people who would have historically lived quite successful lives fail.
In the past, this was the problem of the individual concerned, and society made some effort to push people, or at least men, towards roles that suited their capabilities. With the rise of digital government, automation, and the welfare state, it’s rapidly become everyone’s problem.
This is most obvious in the now-ridiculed slogan “learn to code.” The idea was that anyone could upskill. But that’s not how cognition works. IQ strongly correlates with how quickly and effectively people learn new domains.
A significant amount of the population simply do not have the ability to easily retrain into complex roles. That’s not a moral failing – it’s a human limitation. A welfare or jobseeker system that ignores this will produce failure, despondency, and dependency.
Most policy work in this area focuses on outcomes: school results, literacy targets, labour force participation. Almost none of it looks at inputs: the actual cognitive distribution of the population. And why would it? Conversations about IQ – or even basic biological differences like strength between men and women – are treated as unmentionable. They make respectable people uncomfortable.
But if we want systems that “help people flourish,” as the NGO class loves to say, we have to build for who people are, not who we wish they were.
One of the quiet tragedies of modern discourse is that we obsess over particular, often relatively minor, differences – race, sex, sexual orientation – while ignoring one of the most fundamental differences that shapes every interaction, decision, and opportunity: intelligence. If we’re serious about building systems that work for everyone, we might start by doing something radical: asking average – or perhaps even slightly thick – people what they think of those systems. And then listening.
As I said above, a lower IQ isn’t a moral flaw. But we’ve built a society that makes it a liability. We’ve turned independence into a cognitive test. And we’ve made participation contingent on interpreting systems that are, for many, indecipherable.
In that kind of world, where your success depends on grasping macroeconomics, platform interfaces, and abstraction layered upon abstraction, a quite large amount of people will start to see things one of two ways:
Both paths lead to alienation and disempowerment.
We can keep pretending the system works. That everyone is equally capable. That miscommunication is a problem of tone, not comprehension. But the clear and apparent truth is that our assumptions about equality have become so delusional that they’re causing real world harm.
Literacy is not a certificate. Intelligence is not evenly distributed. And building a society that ignores both is not compassion – it’s negligence.
There’s been plenty of talk about AI displacing jobs. But for many people on the lower end of the cognitive scale, that ship already sailed. Their jobs weren’t taken by AI. They were automated away years ago – or offshored, outsourced, and replaced with online portals, QR codes, and corporate helpdesks. Those that remain seem to be relatively secure from displacement by automation or AI, at least in the short to medium term.
The fear now spreading among white-collar workers – panic over being rendered obsolete by a machine – is a taste of what other people have experienced, quietly, and without recourse, for decades. No retraining. No safety net. Just confusion, frustration, and exclusion.
Ironically, AI may be one of the few developments that could actually improve their situation. Because the core issue wasn’t that humans were replaced by machines. The core issue was that human interaction was replaced by bureaucratic process – cold, inflexible systems that made no allowance for individual comprehension, emotional nuance, or community support.
But if AI can replicate a more human interaction – adaptive, conversational, infinitely patient—it could become a bridge, not a barrier. AI could explain things more clearly than a bureaucrat. It could repeat itself without irritation. It could tailor its language to the user’s ability. It could offer 24/7 help, without shame or judgment.
For the first time in decades, it might actually be possible to build systems that meet people where they are – not where policy designers wish they were.
This same logic applies to education. One of the most promising applications of AI isn’t in replacing teachers – although there are a number of people in tech currently throwing their eye in that direction – but in supplementing them. With adaptive tutoring systems, students who struggle with traditional methods could receive personalised support, paced to their ability, endlessly patient, and capable of presenting information in multiple formats. For students who often fall behind – not due to laziness, but due to cognitive limitations – this could mean the difference between disengagement and genuine comprehension. AI could help close the gap between expectation and ability in the classroom, just as it could in public services.
Of course, the above is a – shall we say – best case scenario. Because the truth is, the other solutions proposed for these problems, to the extent they exist at all, are well-meaning but insufficient. We have plain English campaigns, simplified forms, accessibility checklists. They are laudable, but they are piecemeal -and often ineffective.
They assume that small tweaks to a system built for high-functioning users will make that system usable for everyone. But if the foundation is exclusionary, the patch is just decoration. The problem is not cosmetic – it’s structural.