The Question Everyone Is Afraid to Answer — Has AI Become More Creative Than Us?
The AI vs human creativity 2026 debate just stopped being a philosophical argument. It is now a scientific one — with data from 100,000 people to back it up.
A landmark study published in Scientific Reports — the largest head-to-head comparison of human and AI creativity ever conducted — has delivered its verdict: generative AI now outperforms the average human on standard creativity tests.
The research was led by Professor Karim Jerbi at Université de Montréal, with contributions from Google DeepMind scientists and Yoshua Bengio — one of the founding fathers of modern deep learning, and a Nobel Prize-level figure in AI research. When someone of Bengio’s stature puts his name on a study saying AI beats humans at creativity, the world pays attention.
But the findings are more nuanced — and more important — than the headline suggests. Here is the full story.
Get AI research stories like this every week: Subscribe to InternetBuzzFeed free → Plain English. No jargon. The AI stories that actually matter.
What Is the AI vs Human Creativity 2026 Study — Full Breakdown
The AI vs human creativity 2026 study titled “Divergent Creativity in Humans and Large Language Models” was published on January 21, 2026 in Scientific Reports, one of the world’s most respected peer-reviewed journals.
The research team tested more than 100,000 people across multiple countries against several of the world’s most advanced AI models — including GPT-4 — using a standardised psychological test called the Divergent Association Task (DAT).
How the test works
The Divergent Association Task is simple in concept but revealing in practice. Participants are given one instruction: produce ten words that are as semantically unrelated to each other as possible.
For example, a low-creativity response might be: dog, cat, fish, bird, rabbit, hamster, turtle, horse, mouse, snake. All animals. Closely related. Low score.
A high-creativity response might be: democracy, calcium, tambourine, eclipse, velvet, logarithm, archipelago, grief, nitrogen, chandelier. Wildly unrelated concepts drawn from completely different domains. High score.
The AI models were given the same task. Their scores were compared against 100,000 human responses. The AI vs human creativity 2026 results were clear and surprising.
Beats ~90% of human participants on the DAT divergent thinking test
The top 10% of human performers significantly outscored every AI model tested
What AI vs Human Creativity 2026 Results Actually Found
The AI vs human creativity 2026 findings break down into three clear findings that every person should understand:
Finding 1: AI beats the average human — comfortably
GPT-4 and similar large language models scored significantly higher than the average human participant on the DAT. Not slightly higher. Comfortably higher. The AI models demonstrated an ability to draw connections between concepts across wildly different knowledge domains — physics, music, geography, emotion, mathematics — that most humans simply don’t do instinctively under test conditions.
Finding 2: The top 10% of humans still beat every AI
Here is where the story gets interesting. The most creative humans — people who scored in the top 10% on the DAT — significantly outperformed every AI model tested. Their responses showed a depth of unexpected, emotionally resonant, culturally specific creativity that the AI models could not match.
Poetry. Metaphor. Personal meaning embedded in language. The kind of originality that comes from a specific human life, not a training dataset. The AI vs human creativity 2026 study shows clearly that AI has surpassed the average — but the creative ceiling of the best humans remains above AI.
Finding 3: Most people are less creative than they think
Perhaps the most uncomfortable finding of the AI vs human creativity 2026 study is this: the majority of humans, when tested formally, perform at or below the level of AI on divergent thinking. Not because humans lack creativity — but because most people self-censor, follow patterns, and stay within familiar semantic territory without realising it.
In other words: AI doesn’t just beat average humans at creativity. It beats the version of average humans who aren’t actively trying to be creative.
Who Conducted This Study — And Why Their Names Matter
The credibility of the AI vs human creativity 2026 study comes from who built it. This was not a small university experiment or a tech company self-report.
- Professor Karim Jerbi — Department of Psychology, Université de Montréal. One of the world’s leading cognitive neuroscientists.
- Yoshua Bengio — Founder of Mila (Quebec AI Institute), pioneered the deep learning techniques that power every major AI model including ChatGPT. He is widely considered one of the three “Godfathers of AI” alongside Geoffrey Hinton and Yann LeCun.
- Google DeepMind scientists — Multiple co-authors from DeepMind contributed, lending the study extraordinary technical credibility.
- Université Concordia, University of Toronto Mississauga — Additional institutional weight across the author team.
When Yoshua Bengio — who has publicly warned about the existential dangers of AI — contributes to a study showing AI has surpassed average human creativity, he is not celebrating. He is documenting a milestone we all need to take seriously.
What This Means for Creative Jobs in 2026
The AI vs human creativity 2026 study lands in the same week that Meta fired 8,000 employees and 113,000 tech jobs were eliminated industry-wide. The timing is not coincidental — it reflects an accelerating trend.
If AI now surpasses average human creativity on measurable tests, what does that mean for jobs in creative industries?
Jobs most at risk: Content writing at volume, basic graphic design, social media copy, stock photography concepts, standard marketing copy, generic blog posts, and any creative work where “good enough” is the brief.
Jobs that remain safe: Original storytelling, brand identity with deep cultural context, emotional advertising, music composition with personal meaning, film direction, live performance, and creative work where the human story behind it is the product.
The line between safe and at-risk is the same line the study found: average creativity loses. Exceptional, specific, personally meaningful creativity wins.
The uncomfortable implication: most creative professionals are not in the top 10%. And the gap between the AI floor and the human average is now significant — and widening every six months as models improve.
The Deeper Question — Can AI Actually Create, or Just Recombine?
he AI vs human creativity 2026 study measures divergent thinking — the ability to connect unrelated concepts. But many researchers argue this is only one small dimension of creativity.
True creativity, they argue, involves:
- Emotional motivation — creating because you feel compelled to, not because you were prompted
- Cultural meaning — producing work that resonates specifically because of who made it and when
- Failure and iteration — the messy human process of getting it wrong repeatedly before getting it right
- Physical experience — art that comes from embodied sensation, not text prediction
On these dimensions, AI has no score. It cannot be measured on a test because the test would need to include the human experience of being alive — something no dataset can contain.
So who wins the AI vs human creativity 2026 debate? The honest answer: AI wins the benchmark. Humans still own the meaning.
Final Verdict — Should You Be Worried?
If your work depends on average-quality creative output produced at volume — yes, you should take this seriously and start building AI-augmented workflows now.
If your work depends on the kind of creativity that comes from your specific life, your specific perspective, and your specific relationship with an audience — no. The study confirms you are still ahead of every model tested. But the gap is smaller than it was two years ago, and it will narrow further.
The most empowering takeaway from the AI vs human creativity 2026 study is also the most honest one: the people in the top 10% are not there by accident. They got there by pushing past the comfortable semantic territory that most people — and all AI models — default to. That is a learnable skill. And it is the only one the study confirms AI cannot yet replace.
Keep reading InternetBuzzFeed for the AI stories that shape your career and business: Subscribe free → | Read: Meta AI layoffs — what they mean for your job → | Read: Claude AI enterprise deployment 2026 →
Frequently Asked Questions
Is AI more creative than humans in 2026?
According to the largest study ever conducted on AI vs human creativity — testing over 100,000 people — AI models like GPT-4 now outperform the average human on divergent thinking creativity tests. However, the top 10% of the most creative humans still significantly outperform every AI model tested. AI beats the average, but not the best.
What study proved AI beats humans at creativity?
The study “Divergent Creativity in Humans and Large Language Models” was published in Scientific Reports on January 21, 2026. It was led by Professor Karim Jerbi at Université de Montréal with contributions from Yoshua Bengio (Google DeepMind co-author) and researchers from multiple universities including University of Toronto and Concordia.
What is the Divergent Association Task used in the study?
The Divergent Association Task (DAT) is a psychological test where participants are asked to produce ten words that are as semantically unrelated to each other as possible. Responses are scored by measuring the semantic distance between words — the further apart the concepts, the higher the creativity score. Both humans and AI models were tested identically.
Will AI replace creative jobs in 2026?
AI is already replacing creative work at average quality and high volume — content writing, generic marketing copy, basic design concepts. Jobs requiring exceptional originality, emotional depth, cultural specificity, and personal storytelling remain safer. The study suggests the risk is highest for creators whose work falls in the average-quality range.
Who is Yoshua Bengio and why does his involvement matter?
Yoshua Bengio is the founder of Mila (Quebec AI Institute) and one of the three “Godfathers of AI” — along with Geoffrey Hinton and Yann LeCun — who pioneered the deep learning techniques powering all modern AI. He has also been one of the most prominent voices warning about AI risks. His co-authorship on a study showing AI surpassing human creativity carries extraordinary credibility.
What types of creativity can AI NOT do better than humans?
The study found AI cannot match the top 10% of human creatives on richer creative tasks involving personal meaning, emotional depth, and cultural specificity — poetry, storytelling, and work where the identity of the creator is part of the value. AI also has no emotional motivation to create, no embodied physical experience, and no personal history — all of which inform the deepest human creative work.
Visit for more info: internetbuzzfeed.com