Every generation believes the next one is getting dumber. This time, however, the data is making that accusation much harder to dismiss. For most of the 20th century, humanity was on a cognitive winning streak. Probably thanks to better nutrition, widespread education, and improved healthcare, average IQ scores steadily rose by about three points per decade. Psychologists called this the “Flynn effect,” and it felt like undeniable proof that we were steadily becoming a smarter species.
But around the turn of the millennium, the lift stalled, and then started moving backward.
Recent studies show that over the past few decades, IQ test results have been dropping across many highly developed nations. This isn’t just a pessimistic internet rumour, but a systemic shift in how human brains are operating. So, are we actually getting less intelligent, or are the tests just measuring the wrong things? And what happens to human innovation when we throw artificial intelligence into the mix?
Let’s look at the data.
The long rise in IQ scores began to break with the arrival of TikTok, or at least this is what some boomers might think, but the truth is that it actually began way earlier than most people realise. Extensive data from mandatory military conscription tests in Scandinavia (specifically Norway, Denmark, and Finland) showed that by the mid-1990s, younger cohorts had stopped gaining ground, and their cognitive scores were actually falling.
Independent researchers analysing data across Europe, Australia, and North America confirmed that average IQ scores in several developed nations have fallen by an estimated five to seven points per generation since that 1990s turning point. In the US, a recent analysis of nearly 400,000 Americans showed drops in verbal reasoning, visual problem-solving, and computational maths.1
Does this confirm that we are getting dumber?
The drop in scores is mathematically confirmed, but the debate is still open on what the drop actually means.
If we look at the timeline, the reversal began with Gen X and early Millennials in the 1990s, long before smartphones, social media, or AI tools became part of our daily lives. This suggests the initial decline wasn’t caused by the modern digital dopamine loop. Instead, researchers point to broader environmental shifts in how we live, learn, and work.
It’s also important to notice that scores have not declined evenly in all areas of IQ tests. While Americans performed worse in verbal and mathematical reasoning, their scores in 3D spatial rotation actually increased. It seems our brains are not rotting after all, or at least not fully. They have simply adapted to an ecosystem that rewards visual processing and quick filtering over deep, sustained concentration. The impact of that adaptation is still poorly understood. But one obvious question follows: if intelligence is adapting to different environments rather than simply declining, should we expect every society to adapt in the same way?
While the West has largely seen a plateau or decline, some data suggests the Flynn effect has continued, or at least stabilised without dropping, in highly developed East Asian nations like Japan and South Korea. Researchers often attribute this to differences in educational rigor, cultural emphasis on standardised testing, and different digital consumption habits.
If different environments produce different cognitive outcomes, then perhaps IQ isn't the whole story. Intelligence isn't just about solving abstract puzzles in a testing room, but also about how we process information in everyday life. And today, that environment is increasingly designed by algorithms.
Even if our cognitive skills started shifting when the internet was just starting to become popular, the modern social media ecosystem has weaponised that shift.
We have entered the era of anecdata, where personal anecdotes shared on social media often carry more weight than peer-reviewed science. We see this when people trust TikTok influencers over physicians for medical decisions, like injecting experimental peptides for building muscles. We see it in the political arena, where figures can simply dismiss the overwhelming scientific consensus on climate change, and millions follow blindly.
Ironically, I'm about to use an anecdote myself. Unlike scientific evidence, my experience proves absolutely nothing. At best, it illustrates a pattern that the data might help explain. Still, I can't help noticing this almost every day: people sharing a controversial video or an article without attaching a single comment of their own. Or watching a film, immediately searching for a YouTube review, and then regurgitating that reviewer’s opinion as their own.
It reminds me of a brilliant scene in the film Captain Fantastic. A father, raising his children off-the-grid in the forest, asks his teenage daughter what she thinks of the book Lolita. When she starts passively summarising the plot, he cuts her off and forces her to explain what it actually made her feel. He forces her to use her own brain to build a critical opinion.
My concern isn't that people have become incapable of thinking. It's that fewer situations require them to. The internet has become astonishingly good at supplying ready-made opinions, conveniently packaged explanations, and algorithmically selected conclusions. Thinking for yourself has become optional, and like any muscle, what isn't used eventually weakens.
When we spend hours a day in an algorithm designed solely for emotional engagement and passive scrolling, our brains adapt by losing the endurance required for critical analysis. We become highly susceptible to charismatic authority and echo chambers, not because we lack innate intelligence, but because our environment has stopped demanding it of us.
If that loss of cognitive endurance is real, then it shouldn't only affect political debates or online discussions. One place we might expect to see its effects is in humanity's greatest intellectual output: scientific discovery itself. That is an extraordinarily difficult thing to measure, but one group of researchers tried. A massive 2023 study published in Nature analysed 45 million papers and 3.9 million patents over six decades and found a startling trend: science and technology are becoming significantly less “disruptive” over time.
How did they measure this?
The researchers created a metric called the CD Index (Consolidation-Disruption Index). The logic is brilliant but simple, and it starts by dividing papers into two groups:
Disruptive science: if a paper completely changes the game (like Einstein’s theory of relativity), future scientists will cite that paper, but they will stop citing the older papers that came before it, because the old foundational knowledge is now obsolete.
Consolidating science: if a paper just adds a tiny incremental tweak to existing knowledge, future scientists will cite the new paper alongside all the old foundational papers.
The Nature study found that modern research is overwhelmingly consolidating. We are churning out more volume than ever, but it is largely incremental.
The counterargument:
Not all scientists agree with this grim conclusion. Critics point out that the CD Index might just be measuring a change in how we cite things. Today’s academic culture of publishing at all costs, combined with digital search tools, means researchers often enrich their papers with massive bibliographies to cover their bases. Furthermore, patent attorneys now intentionally cite every prior patent they can find to avoid legal liabilities. Lastly, the Nature data only went up to 2010, meaning it entirely missed the recent explosive, disruptive eras of CRISPR, mRNA vaccines, and generative AI (I’d like to add nuclear fusion to the list, but we know that it’s always 30 years away, and always will be).
All of these threads (declining IQ scores, changing attention, algorithmic thinking, and the apparent slowdown of disruptive innovation) lead to the same uncomfortable question. For centuries, we offloaded physical labour to machines so we could focus on cognitive work. Today, we are offloading the cognitive work itself: memory to search engines, spatial navigation to GPS, and critical analysis to LLMs.
Can we outsource our foundational skills without losing our ability to innovate? Innovation requires connecting disparate dots. If you don’t hold the mathematical formulas, historical context, or scientific principles inside your own head, you cannot accidentally bump them together in the shower to create a paradigm shifting idea. You cannot disrupt a field you do not deeply, internally understand.
If human brains are losing their endurance for deep, abstract reasoning because our digital tools do it for us, are we entering an era of intellectual stagnation? Or are we simply evolving into a new kind of symbiotic relationship with our technology, one where AI handles the heavy computational lifting, and humans act as the agile curators of that intelligence?
I’d love to hear your thoughts in the comments. Are we witnessing the decline of human intellect, or just the messy transition into a new kind of mind?
Intelligence has always adapted to the tools available. The printing press changed memory. Calculators changed arithmetic. Search engines changed recall. AI may prove to be another such transition, or the first technology that quietly erodes the very cognitive foundations on which future breakthroughs depend. The worrying part is that these shifts only become obvious in retrospect. By the time we know which path we're on, we'll already have walked most of it.
