ChatGPT became popular in no time for providing accurate and good answers to any query that people have. Following this, Microsoft also announced its AI-powered Bing tool to catch on the trend and help people offer a better experience using artificial intelligence. However, the ChatGPT-powered Bing tool from Microsoft made quite a lot of mistakes and showed inaccurate results during the company’s public demo last week.
The AI-backed tool by Microsoft quoted the wrong operating margin of Gap during the demo event, according to the details revealed by engineer and writer Dmitri Brereton. The tool reported clothing brands’ operating margin as 5.9 percent; however, Gap’s earning report reveals that the margin was 4.6 percent.
The tool also showed fake answers in terms of net sales growth. It reported that net sales growth in the low double digits, but it is somewhere down to mid-single digits year-over-year in the fourth quarter of fiscal 2022.
This is not it. The ChatGPT-powered Bing tool from Microsoft also made a mistake when asked about nightlife recommendations for Mexico City during the demo. The tool advised people to visit Cecconi’s Bar website to make reservations and check their menu. But, it was discovered that there is no website for this and people need to make a call to make a reservation.
Bing also made an error when the tool was asked to show results on the advantages and disadvantages of Bissell Pet Hair Eraser Handheld Vacuum. It reported that the product has a “short cord length of 16 feet,” which is a disadvantage as per the tool. But the reality is, the vacuum is a cordless to offer portable experience to users. In addition to this, Bing provided details on the basis of top-recommended pet vacuums, instead of the best-selling.
Lastly, Microsoft also asked its AI-powered Bing tool to display a quiz of 90s music. The good thing is that the tool managed to show correct names of the musicians for each question. However, the answers for the questions were “Answer A,” which was a little weird.
Unlike Bing, ChatGPT has been proven to be a very useful AI tool as it is able to provide answers to all the queries properly. It has just been discovered that the platform is capable of passing a major medical licensing exam too. A few researchers have used ChatGPT to analyse whether the AI tool can pass the major United States Medical Licensing Exam (USMLE), which basically includes three sets of exams that the doctors need to take to acquire a medical license. Surprisingly, ChatGPT managed to score between 52.4 percent and 75 percent in all the exams, according to the details revealed in the journal PLOS Digital Health.
