Bing: Here’s why ChatGPT-powered Microsoft Bing’s inaccurate’ answers are not a big deal – Times of India

ChatGPT took the world by surprise and rose in popularity because it provided strikingly human-like responses to user queries. Soon came Microsoft with a multi-billion deal and launched ‘new’ Bing search engine with AI capabilities. The company recently started providing early access to those in the waiting list. While people were hoping that the ChatGPT-powered Bing will deliver accurate results, reports point out that the tool made mistakes and showed inaccurate results during the company’s public demo last week.
The development gains significance because inaccurate answers by Google’s AI-powered chatbot Bard tanked its shares’ prices. But here’s why inaccurate answers by these chatbots isn’t a big deal at this point of time.
Bing’s ‘inaccurate’ answers
As per the details highlighted by engineer and writer Dmitri Brereton, the AI-backed tool by Microsoft quoted the wrong operating margin of Gap during the demo event. It said that the clothing brand’s operating margin was 5.9%, however, Gap’s earning report reveals that the margin was 4.6%. The tool also inaccurately answered the company’s net sales growth.

Other mistakes included wrong advice on making a reservation in Cecconi’s Bar via a website. The restaurant does not have a website and one needs to make a call to book a table. Similarly, it made an error while saying that one of the disadvantages of Bissell Pet Hair Eraser Handheld Vacuum is a “short cord length of 16 feet.” The vacuum is a cordless offering.
How Bing compiles its responses
At the event, Microsoft said that Bing has OpenAI’s new AI language model that “takes key learnings and advancements from ChatGPT and GPT-3.5 – and it is even faster, more accurate and more capable.” However, it clearly mentioned in the FAQs for the new Bing that the tool searches for answers from reliable sources, annotates the source but it can make mistakes.
“Bing aims to base all its responses on reliable sources – but AI can make mistakes, and third party content on the internet may not always be accurate or reliable. Bing will sometimes misrepresent the information it finds, and you may see responses that sound convincing but are incomplete, inaccurate, or inappropriate,” Microsoft said.

“Use your own judgement and double check the facts before making decisions or taking action based on Bing’s responses,” the company advises.
Furthermore, the tool is still in its nascent stage and has limitations, which means there is a lot of work that needs to be done to make it a reliable source of information.
In one of tests performed by Times of India-Gadgets Now team, Bing responded with inaccurate specifications. We asked it to compare the Samsung Galaxy S23 Ultra with the Galaxy S22 Ultra. In our first attempt, Bing offered results with a few ‘inaccurate’ specifications. However, when we questioned the ‘accuracy’ of the results, it responded that it took the data from ‘reliable sources’. It asked for a source from our team and when we provided it with the official Samsung website as the source, the tool said it will update the information.
Responsible AI
Recently, Google parent Alphabet chairman John Hennessy spoke about Google’s rival to ChatGPT and highlighted why the company was ‘late’ in bringing the tool to market.

Hennessy said it is important for these AI-powered products to give the right information. “You don’t want to put a system out that either says wrong things or sometimes says toxic things,” Hennessy said. He further said that generative artificial intelligence is still “one to two years away from being a truly useful tool for the broader public.”
Too early to judge AI-generated responses
The goal of both Microsoft and Google is to provide accurate, factual information to users. However, it will take time for this technology to evolve and given that the launch is just a few days old, it will be unfair to judge or even call out the responses from these tools at this early stage.