On Tuesday, the 10th of February, Safer Internet Day was marked in more than 160 countries around the world. The central event in Latvia was the educational conference “Dialogue with Artificial Intelligence,” organized by the Safer Internet Centre Drossinternets.lv. As part of the event, a Turing test was held in Latvia for the first time to examine whether participants could distinguish between answers generated by artificial intelligence (AI) and those prepared by humans.
On Safer Internet Day and throughout February, educational events and activities take place in Latvia, across Europe, and worldwide with the aim of educating children, young people, and adults about online safety.
This international initiative is dedicated to the responsible, smart, and safe use of technology, with particular emphasis on the online safety of children and young people.
“An anniversary year allows us to look back at what has been accomplished. Nineteen years ago, on Safer Internet Day in 2007, the main topics were identity theft, risks in online games, the need for content filters, and emotional abuse on the internet. In 2017, the focus was on young people’s habits in social media. In 2021, key issues included safe communication online, fake news, recognizing disinformation, and terminology. Currently, one of the biggest challenges is the rapid emergence of AI, which is why this year’s conference theme is ‘Dialogue with AI,’” explained Maija Katkovska, head of Drossinternets.lv.
The Turing test consisted of four rounds.
In each round, questions were answered both by AI programs and by well-known figures in Latvia:
Minister for Education and Science Dace Melbārde, actor Mārtiņš Vilsons, journalist and TV personality Kristīne Garklāva, and brothers Ralfs and Martins from the TV cooking show Ralfs gatavo.
The Turing test was created in 1949 by computer science pioneer Alan Turing to determine whether a computer could be as intelligent as a human. Even today, more than 70 years later, society is concerned with the same question, and discussions are particularly intense in the field of education. Preserving the core idea of the Turing test, 220 conference participants—teachers and students—attempted to determine which answers belonged to the named personalities and which were generated by AI.
“For the teachers and students who participated, identifying human answers from AI-generated ones was by no means easy, because we included questions related to values, emotions, desires, and experiences. For example: What would you never believe in life? What is the most beautiful song in the world? Which came first—the chicken or the egg?” said Liene Valdmane, Head of Education at Drossinternets.lv.
In the Turing test conducted in Latvia on the 10th of February, participants failed to overcome the 30% threshold set by Turing,
meaning that artificial intelligence is recognized as the winner of the test.
Turing believed that if more than 30% of participants fail to distinguish human answers from AI-generated ones, then AI should be considered the winner.
On Safer Internet Day, 40.8% of participants made mistakes in the Turing test, failing to distinguish between human and AI-generated answers.
Commenting on the results, Liene Valdmane emphasized: “The Turing test reflects key issues of our era—technological development, the role and uniqueness of humans in the world, responsibility, and rapid change. The test is not a scientific measure of the real situation; rather, it should be viewed as a game that encourages each of us to reflect on our own AI literacy, strengthen critical thinking, and cultivate what is uniquely human and cannot be copied or artificially generated. Of course, we would all like the result to be the opposite, reassuring us that we are in control. But reality shows that the rapid development of technology also requires us to change.”
“In the final round, I was convinced that the human would win, but I had forgotten that AI can be assigned a specific role and even an age. Many participants, including myself, were ‘tripped up’ by the inability to distinguish children’s answers from those generated by AI. Those who work with AI daily and can critically evaluate responses based on text structure and small details—such as emojis and punctuation—performed better. Humans often write in incomplete sentences, while AI structures are usually fully formed,” explained Inga Narmonta, Deputy Principal for Education and Latvian language teacher at Ropaži Secondary School.
Namejs Valters, a student at Riga Art and Media Technical School, said this was his first time participating in such a test, admitting it was not easy and that he himself made mistakes in two rounds.
“It’s interesting to see how differently people from various professions think.
The contrast between adult and children’s answers was even greater,” he noted. To better distinguish AI in the future, people need to learn more intensively. “I think schools should have a subject dedicated to AI—either as a separate class or integrated into computer science. The more knowledge there is about AI, the easier it will be to recognize it,” the student emphasized.
Safer Internet Day activities are taking place within the framework of the “Experience AI” project, implemented in Latvia by Drossinternets.lv in cooperation with Raspberry Pi Foundation. So far, 600 teachers have been trained, educational modules have been developed, and AI guidelines have been created to encourage schools to begin dialogue about AI’s role and significance. In March and April, additional teacher training sessions titled “Experience Artificial Intelligence” will be held, providing both enhanced knowledge about AI and extensive methodological support for educating students. Trained teachers have highly praised the program, noting that it fosters understanding of AI’s nature and provides confidence in discussing the ethical and safe use of AI with students.
Read also: Lithuania develops night vision cameras to intercept smuggler balloons