Artificial Intelligence (AI) can now easily create images, music and text that can be done by a skilled human. We see that the world of online content has undergone a major change towards the mid-2020s. This is further accelerated by ChatGPT, which is growing rapidly, reaching the first one million users in just five days, and 100 million users in a few weeks.While those interested in technology know a lot about artificial intelligence and chatbots like ChatGPT, it’s not for everyone, and we shouldn’t expect it to happen. In fact, some people may not even know how sophisticated the outputs of AI tools are and whether what they read or see online was produced by a human or an AI.
A survey of thousands of people was conducted to further explore this issue. People were asked to guess whether text for health, finance, entertainment, technology, and travel content was created by artificial intelligence or humans, and the results are quite surprising.
People can’t tell the difference
According to the results, more than 53 percent of users cannot accurately describe content created entirely by artificial intelligence chatbots like ChatGPT. This rate rises to 63.5 percent when the GPT-4 model is used. On average, the GPT-4.0 language model performs 16.5 percent better than GPT-3.5, convincing people that AI-generated text was written by a human. AI-generated health content was the most deceiving content, with 56.1 percent of people believing that AI content was written or edited by a human.
Readers most accurately predicted AI-generated content in the tech industry; This is the only industry where more than half (51%) of people correctly identify AI-generated content. Content generated by GPT-4.0 was mostly undetected when it comes to travel, and 66.5 percent of readers thought the content was written by humans. Those more familiar with AI tools like ChatGPT were marginally better at identifying AI content, but their margin of error was as high as 52 percent. This rate rises to over 60 percent in unfamiliar people.The elderly are more successful than the young
The majority of people (80.5%) believe that online publishers who publish blogs and news articles should clearly indicate whether artificial intelligence was used in their creation. More than seven in ten (71.3%) say they would trust a brand less if they were given AI-generated content without being told.
Most people (46.5%) said they would not mind if AI gave them health and financial advice, while 42.9 percent said they would only use such advice if a human edited and reviewed the content. Interestingly, it was the youth who had the most difficulty in predicting the AI content correctly. In general, it is seen that people give more correct answers as they get older.
50573
