Can AI beat human intuition?

Published on April 3 2023
We explore whether AI can surpass human intuition, inspired by an experiment distinguishing AI from human-written texts. Initial results were split, but later, accuracy improved significantly, suggesting we may develop an intuitive sense for AI-generated content.

No. It cannot, at least not for now. We’re basing this answer on the result of a small and unplanned experiment we conducted over the last few months.

On 9 November 2022 BC (before ChatGPT), during Diplo’s AI and Data Day in Geneva, we asked our participants to guess which text was written by AI and which by humans. The participants’ success rate was 50%.

YouTube player

Test yourself

Text A

Thank you for coming together to discuss a challenge of growing global importance.

As someone with a background in engineering, I am deeply interested in the remarkable phenomenon of artificial intelligence, and as a secretary general of the United Nations, I am strongly committed to promoting global cooperation on this emerging issue.

Artificial intelligence is advancing dramatically. It is already transforming our world socially, economically, and politically. We face a new frontier with advances moving at warp speed. Artificial intelligence can help analyse enormous volumes of data which in turn can improve predictions, prevent crimes, and up government’s better serve people.

But there are also serious challenges and ethical issues at stake. There are real concerns about cybersecurity, human rights, and privacy, not to mention the obvious and significant impact on the labour market. The implications for development are enormous. Developing countries can gain from the benefits of artificial intelligence, but they also face the highest risk of being left behind.

This summit can help ensure that artificial intelligence charts the course that benefits humanity and bolsters our shared values, and the United Nations stands ready to be a universal platform for discussion.

Together, let us make sure we use artificial intelligence to enhance human dignity and serve the global good.

Thank you.

Text B

We are gathered here today to discuss the phenomenon of artificial intelligence and its implications for the future of our world.

As the secretary general of the United Nations, I am tasked with ensuring the safety and security of all people, and I can say with certainty that artificial intelligence is a threat to both.

There are many dangers associated with artificial intelligence, but the most pressing is its potential to be used for warfare.

AI-enabled weapons could make warfare even more devastating than it is today and could even lead to the development of autonomous weapons that could make decisions without human input.

In addition to the threat of AI-enabled warfare, there is also the risk of AI being used to commit other crimes, such as hacking into critical infrastructure or stealing sensitive data. As AI becomes more sophisticated, these risks will only increase.

We must take action to address the threat of artificial intelligence, and I urge all countries to work together to develop a set of international norms and regulations governing the use of AI. We must also invest in research and development to ensure that we stay ahead of the curve in terms of AI technology.

Only by taking these steps can we hope to mitigate the risks associated with artificial intelligence and ensure that it is used for the benefit of all humanity.

Thank you.

Check if you got it right at the bottom of this page!

We made the same test (to identify text written by humans and AI) on 30 March 2023 AC (After ChatGPT) during the InnoVent 2 in Geneva. This time the success rate was 90%.

We asked our participants HOW they could tell who wrote the texts: humans or machines. ‘I felt that it was written by a human somehow’ was the explanation we received from most participants. It wasn’t a careful analysis of syntax or the choice of words. It was based more on intuition and the condensed experience we’d gathered over the last few months of using ChatGPT. Intuition uses our deeper and more powerful skills. 

This small experiment opens many questions:

Have we developed intuition (condensed experience) for differentiating AI-generated text from human writings? 

Is intuition our last refuge in our future competition with machines?

Can AI automate our intuition?

Do these questions stir excitement, anxiety, or other emotions in you? Join our discussions as we try to answer these and many other questions from various viewpoints. At Diplo, we address AI issues from technical (algorithms) and linguistic (semantics and stories) to philosophical (epistemology) and psychological (cognition) perspectives. 

To receive updates on Diplo’s humAInism project and upcoming courses on AI and society, please let us know at ai@diplomacy.edu

You can also watch our AI-related videos and webinars on our YouTube channel.

Solution for the above test:

Text A is written by humans | Text B is written by AI


cross-circle