Online content has changed shape fast in recent years, and the change has a name: artificial intelligence. AI now sits behind the news, photo editing, video production, and more. But not everyone wants it to stay there. One Instagram account called ‘Things A.I. Could Never Recreate’ rejects AI entirely, posting photos that capture what it means to be human instead — once-in-a-lifetime coincidences, unfortunate fails, good old irony. The account’s latest post, titled Unscripted Moments: 36 Chaotic Pictures That Feel Too Real, shows 36 images that feel too human for any machine to produce.
What the Account Does
The account’s mission is simple. It posts unique footage that captures what it’s like to be alive. The pictures are unscripted, chaotic, and real. They show moments that happen once and never again. The account exists as a counterweight to the digital noise around it.
The Gartner Hype Cycle
The hype around AI follows a familiar pattern. The Gartner hype cycle describes how new technology moves from excitement to disappointment to steady usefulness. It goes like this:
- Peak of inflated expectations
- Trough of disillusionment
- Slope of enlightenment
- Plateau of productivity
A Gartner report published in June placed most generative AI technologies at the peak of inflated expectations or still climbing toward it. The paper argued that these technologies are two to five years away from becoming fully productive.
Why Generative AI Keeps Failing
A study released last month by RAND found that 80% of AI projects fail. That is more than double the rate for non-AI projects. Dr. Vitomir Kovanovic, an associate professor at the University of South Australia’s Education Futures and Associate Director of its Centre for Change and Complexity in Learning (C3L), pointed to a bigger problem.
“For example, generative AI systems can solve some highly complex university admission tests yet fail very simple tasks,” he said. “This makes it very hard to judge the potential of these technologies, which leads to false confidence.”
Kovanovic explained the problem further. “Because they fluently answer questions, humans can reach overoptimistic conclusions about their capabilities and deploy the models in situations they are not suited for,” he said.
What Comes Next
Kovanovic believes the AI hype will deflate soon. As it does, he expects more realistic adoption strategies to emerge. He points to a survey of American companies that found they are mainly using AI to improve efficiency (49%), reduce labor costs (47%), and enhance product quality (58%).
He also sees smaller, cheaper generative AI models taking hold. These are trained on specific data and deployed locally to cut costs and boost efficiency. Even OpenAI, which has led the race for large models, has released the GPT-4o Mini version to reduce costs and improve performance.
AI literacy training is rising too. Companies are teaching their workers how AI works, its limits, and best practices for ethical use. Kovanovic believes we will have to learn and re-learn how to use different AI technologies for years to come.
The Verdict on the Revolution
The AI revolution might look more like an evolution. The 36 photos in the Instagram post suggest there are still things machines cannot recreate. The RAND report’s 80% failure figure and Kovanovic’s point that generative systems can solve complex tests yet fail simple ones paint a picture of technology that doesn’t always live up to itself.
The account’s approach is a small but telling gesture. Instead of chasing the latest AI trend, it holds onto the idea that some moments are too human for any machine to produce.
Key facts box:
– 36 chaotic pictures in the post
– 80% of AI projects fail, per RAND
– GPT-4o Mini released by OpenAI
– Survey: 49% efficiency, 47% cost reduction, 58% quality improvement
– Gartner report published in June
The pictures in the Instagram post are a reminder that some moments are too human for any machine to produce. The account’s approach is a small but telling gesture. Instead of chasing the latest AI trend, it holds onto the idea that some moments are too human for any machine to produce.
See the 36 pictures at Boredpanda.
Get the Notebook.
The day's best stories and every fresh verdict, in plain English, in your inbox by seven. One email a day, no more.