Artificial Intelligence is ALREADY Outperforming Humans in Creativity

illustration: bing.com/create

In the study conducted by Montana researchers, ChatGPT ranked in the top 1% for originality. The Torrance Test of Creative Thinking (TTCT) has been used for decades to assess human creativity. It evaluates creativity based on several factors:

• fluency,
• flexibility,
• originality
• elaboration.

According to Newseria, the researchers submitted eight ChatGPT-generated responses and a control group of 24 University of Montana students for evaluation. The results were compared with the scores of 2,700 students nationwide who took the TTCT in 2016.

- We decided this would be an interesting subject to study because ChatGPT and GPT-4 perform very well on other tests, typically displaying traits like memory and extensive knowledge. Along with other researchers, we concluded that GPT-4 can create creative responses to tasks, says Erik Guzik from the University of Montana. - Interestingly, OpenAI claims GPT-4 is more creative than GPT-3. This is fascinating in itself and motivated us to explore GPT-4`s creativity by measuring the fluency, flexibility, and originality of its responses in the Torrance Test of Creative Thinking.

Creativity at an EXCEPTIONAL Level

The evaluation was conducted by Scholastic Testing Service (STS), which was unaware of AI`s participation in the study. The STS analysis revealed that ChatGPT achieved exceptional results, ranking in the upper percentile for fluency and originality but slightly lower in flexibility at the 97th percentile.


- Some people are surprised that artificial intelligence demonstrates creative thinking skills, but if we go back to its beginnings, we find its founders openly stated their goal was to simulate creative thinking abilities. They aimed to create a simulation of intelligence, and creativity is a part of that, says Erik Guzik in an interview with Newseria. - The creators of AI clearly defined their goal and used the word “creativity” in the initial documents outlining potential achievements of artificial intelligence.

The study`s results could have far-reaching implications. They suggest that AI could be used to generate new ideas and solutions in fields such as medicine, business, and science.

AI algorithms are already being used to develop new drugs. The biotechnology startup Insilico Medicine was the first to advance a drug designed by generative AI to Phase II clinical trials involving patients. This drug offers hope to patients suffering from idiopathic pulmonary fibrosis, which leads to respiratory failure.

The results from Montana researchers show AI can be a powerful tool to support human creativity.

Artificial Intelligence and Marketing

Marketers using generative AI also see its potential but have concerns about its safety and accuracy. According to the Generative AI Snapshot Series study conducted by Salesforce and YouGov, the industry anticipates significant changes soon.

Salesforce surveyed over 1,000 marketing professionals. The results, cited by Newseria Biznes, showed:

• 51% already use generative AI
• another 22% plan to adopt it soon.

This means nearly three-quarters of marketers recognize its potential and are already leveraging it. Marketers primarily use generative AI for:

• content creation (76%)
• writing copy (76%).

Many predict significant changes in their work due to this technology. As many as 53% say generative AI is a "game changer," seeing potential in personalizing messages, creating marketing campaigns, and optimizing SEO strategies.


Although the technology seems promising, marketers have concerns. 67% believe their company data is not adequately prepared for AI. Similarly, 63% think trusted customer data is essential. Despite the enthusiasm, many believe effective AI use requires human oversight.

Algorithmic Exclusion

Dr. Kuba Piwowar, a sociologist and cultural expert from SWPS University, highlights another critical aspect related to AI: algorithmic exclusion. This phenomenon, caused by improper data selection for analysis, leads to better outcomes for some groups while disadvantaging others.

This occurs because society and culture are challenging to capture as digital data. Typically, marginalized groups - such as women, non-heteronormative individuals, people of color, or lower-income individuals - fare worse.

Classification systems organize reality. If someone doesn`t fit into an accepted category, the system doesn’t see them, making it impossible to create robust policies for these groups. We can’t determine whether an issue is small or widespread or how often it occurs. We overlook groups of people who want to be recognized.

These models reflect societal phenomena. For example, if you search for “physicist,” you’ll likely see images of men in lab coats rather than Marie Curie or other notable women researchers. This happens because men dominate this professional group in society.

Dr. Piwowar believes education is crucial. Analytical skills should be taught in primary and high schools to familiarize students with these concepts. At the university level, especially in technical fields, ethics courses should be introduced to explain the dilemmas surrounding algorithm operations. In the United States, there are even degree programs on ethical AI and responsible AI use.