“ChatGPT’s initial wave of popularity is fading, largely due to unrealistic expectations of the bot.” Many people came here hoping to find an “employee” to take over their jobs, and left realizing that they themselves would need to acquire new skills and change their habits. This by no means means that the AI boom will end as quickly as it began—rather, that current AI products are not meeting the high bar set by consumers. After all, the interest in ChatGPT is possibly decreasing due to new players “running away” to the AI field, such as “Gemini” or “Claude”, says Marijonas Baltušis, head of Telia’s smart devices.
Human psychology interferes
Among the many reasons why people don’t stick with ChatGPT, analysts tend to cite human psychology. The initial record of popularity of this tool was largely “forged” by the fact that it was free and universally available from the beginning, with a paid subscription required only for advanced users. Unfortunately, in the long run, this kind of non-committal model of operation may have led to a relatively quick exodus of the same curious people.
“Being able to test the so-called revolutionary solution without any obstacles, people had the opportunity to form their own opinion and freely choose what to do next.” The absence of a high price barrier did not encourage users to delve deeper into how large language models work, how to provide them with instructions more efficiently, and in general – in what areas of everyday life they could be useful. Mandatory purchase of a subscription would probably not have allowed such a quick popularity of the service, but its buyers would psychologically feel more committed to putting more effort into its testing and learning”, explains M. Baltušis.
The specialist adds that ChatGPT is not yet as convenient for the average user as other tools known to him. To get a summary of some text, a person needs to go to the ChatGPT page, paste the text, and copy the result back to the desired location. At that time, to find out more information on a topic, unlike Google search, typing a word or two in the ChatGPT box is often not enough – the bot requires more precise instructions.
The constant need to switch apps, open additional windows, or type a large amount of text on the keyboard is often too tiring for users, so they may tend to use ChatGPT only in special cases. As a result, it’s no surprise that tech giants are scrambling to integrate AI models into e-commerce as quickly as possible. mailboxes, office suites, and other software that people use every day.
ChatGPT not living up to expectations?
On the other hand, the main reason why users who try ChatGPT don’t stick around is its limited capabilities. According to a study conducted this spring by generative analytics company Narrative BI, this bot gives incorrect answers, or in AI terms, hallucinates, nearly 46 percent of the time. cases where he is asked to perform complex data analysis. Interestingly, this problem is not so easily solved – the developers of ChatGPT openly admit that they cannot yet guarantee the accuracy of the facts presented by this tool and are only looking for ways to do so.
“Frequent fabrication of information and superficial responses limit ChatGPT’s scope to fairly simple mechanical tasks.” It can help students prepare homework, translate a PDF file or turn a text document into presentation slides. But in order for Assistant to be useful for preparing corporate reports, assisting with research papers, or other complex tasks, it needs to provide a lot of context and then carefully check each sentence. As a result, working with an AI robot often requires a little less effort and time than without using it,” thinks a representative of the telecommunications company.
Some people have also been put off by ChatGPT due to concerns over the security of information provided to it in recent years. US cyber security company Malwarebytes found that 81% of respondents are concerned about data security. of those surveyed, and even half of them would agree that the development of ChatGPT should be put on hold until proper legal regulation of AI products is introduced. These apprehensions additionally limit the application scenarios of generative robots or even alienate some users altogether.
Both summer and competitors contribute
Despite the significant decrease in the number of ChatGPT users and various assumptions, M. Baltušis suggests not to draw hasty conclusions about the future of generative AI solutions. The drop in popularity of this robot may be related to the youth leaving school and university benches in the summer and the lack of significant updates, which should change as the autumn and winter season progresses.
It should also not be forgotten that the market for generative AI assistants today is completely different than it was during the record-breaking popularity of ChatGPT. The users of this tool may have been seduced by the competing Google camp, whose Gemini robot under development is less likely to distort facts, allows to provide more contextual information and offers somewhat more intuitive management. We’re also hearing more and more about Anthropic Claude, which is able to generate more natural text, and Perplexity, which uses several of the most popular AI engines to satisfy user queries, rather than using just one.
“Although modern media make it possible to instantly popularize various innovations around the world, they take longer to catch on, just like before. Even the legendary iPhone did not become a hit immediately after its launch – it took three years to reach its current popularity. It is clear that the current ChatGPT has not become as useful for users as it was predicted, but the next breakthrough in rapidly improving AI technologies can turn everything around again overnight, when the capabilities of robots will suddenly catch up with people’s expectations”, summarizes M. Baltušis, the company’s head of smart devices.
2024-09-07 00:47:21