Bing with ChatGPT also makes ‘dumb mistakes’ on first tests

Some users are already using Microsoft’s Bing search engine with OpenAI ChatGPT. But researcher Dmitri Brereton realized that Bing AI was making “stupid mistakes” in financial data and other matters.

For example, the platform helped plan a complete itinerary for a five-day trip to Mexico City. However, information about bars and restaurants, such as prices and location, was not included in the survey.

Brereton states: Bing AI’s biggest mistake was in summarizing clothing brand GAP’s third-quarter financial report.. The original document shows a gross margin of 37.4%, adjusted for 38.7%, excluding the depreciation rate.

Chatbot mistakenly points to a 34.4% gross margin, including fees and depreciation adjustments. The AI ​​also states that the operating margin is 5.9%, given that it didn’t appear in the report that the correct value would be 4.6% (or 3.9% with adjustments for fees).

Reddit users also shared other ridiculous Bing AI bugs. For example, The AI ​​said it’s 2022 and the person should look at their PC settings and check for possible presence of bugs or viruses that changed the date..

Microsoft tries to improve Bing AI

According to Border, Microsoft is aware of these potential bugs during Bing AI testing. Big tech looks forward to getting feedback from users to help fix problems and improve the chatbot.

“We’re looking at the improvements we can make as we continue to learn from the early stages of launch. “We are committed to improving the quality of this experience over time and making it a useful and inclusive tool for everyone,” said Caitlin Roulston, Microsoft’s director of communications.

Experts believe that the Windows owner will still need to make some adjustments until the platform correctly responds to surveys about real data. Remembering that these inaccuracies were also common in ChatGPT’s debut.