Artificial intelligence, the most popular technology of our time, continues to be on the agenda with its failures as well as its successes. A special research reveals that the technology is not yet working accurately and functionally enough. Here are the details
AI tool ChatGPT fails to answer programming questions accurately
According to the latest research conducted by Purdue University, it was found that the artificial intelligence tool ChatGPT answered more than half of the software programming questions incorrectly. It was announced that the use of technology in the programming sector, which has only a forty-eight percent success rate, should be reviewed.
As part of the research conducted by the research team, ChatGPT was asked a total of 517 popular Stack Overflow questions and the results were evaluated with a dozen volunteer developers. Surprisingly, ChatGPT was only able to answer forty-eight percent of the questions correctly, and seventy-seven percent of the answers were characterized as “excessively long”.
It was noted that participants mostly preferred to use the answers they received as a result of the questions they asked. It was stated that the ‘political’ approach of artificial intelligence and the fact that it expresses itself effectively are effective in using the answers in projects, even if they are not accurate and functional enough.
Since this situation causes many problems in the medium and long term in large projects, many giant technology companies, including Apple, Google and Meta, have completely banned their developers from using artificial intelligence in their work.
So, what do you think about the issue? Do you think ChatGPT does not meet the expected performance in the field of programming? You can share your opinions with us in the comments section.
{{user}} {{datetime}}
{{text}}