Microsoft – Last week, two significant tech companies competed to show off their advancements in AI.
At two separate events, early iterations of Google and Microsoft’s AI-powered search engines were on display.
Microsoft had an advantage because their event occurred the day before Google’s, whose dismal failure caused Alphabet’s shares to decline.
Due to the presentation’s broad publicity, more than a million people made attempts to sign up for and use Microsoft’s new tool in the first 48 hours.
Technology may have been “brought to knowledge work” during the industrial revolution, according to Microsoft CEO Satya Nadella.
Since its AI confirmed worries about accuracy, the accomplishment was not without flaws.
The demo
In the trial, the Bing search engine’s ChatGPT-inspired AI system looked at financial data, including information from Gap and Lululemon.
When compared to the real reports, the chatbot’s results revealed that it had a flaw in that it had ignored certain data.
The viewers also saw that several of the data appeared to be incorrect.
Independent search researcher Dmitri Brereton published the following on Substack on Monday:
“Bing AI got some answers completely wrong during their demo. But no one noticed. Instead, everyone jumped on the Bing hype train.”
Brereton also brought attention to what seemed to be factual inconsistencies with the demo’s abnormalities in the vacuum cleaner’s specs and the trip arrangements to Mexico.
The researcher said he wasn’t deliberately looking for mistakes.
Before he attempted to compare the Microsoft and Google AI revelations in his article, Brereton was unaware of the mistakes.
Meanwhile, the inaccuracies were classified as “hallucinations” by AI experts.
Artificial intelligence refers to tools’ propensity to provide data based on in-depth linguistic models as hallucinations.
When Google staged a comparable event, their AI system similarly generated factual mistakes that were simple to see.
AI and search engines
As a method to demonstrate their advancement, Google and Microsoft are striving to include new varieties of generative AI into their search engines.
After OpenAI introduced ChatGPT in November, the rivalry grew more heated.
Microsoft gave billions of dollars to OpenAI.
Several businesses, like Stability AI and Hugging Face, had significant growth at this period as a result of private financing rounds with billion-dollar values.
Read also: Twitter faces more competition in 2023, ex-employees on the rise
On the other hand, Google was hesitant to include AI-generated solutions in its search engines because it needed to uphold its reputation for offering the best results.
The company was also concerned about safety.
However, Microsoft made a point of emphasizing how quickly some members of the public may be exposed to its technology at its launch.
“I think it’s important not to be in a lab,” Nadella added. “You have to get these things out safely.”
Demo problems
The results for company profitability were problematic when Bing introduced their AI technology.
A Microsoft marketing official named Yusuf Mehdi directed Bing AI to highlight the company’s November third-quarter financial results after visiting the Gap investor relations website.
The AI-generated findings revealed the following inaccuracies in the summary:
- Gap’s declared gross margin was 37.4%, but once Yeezy was removed, it increased to 38.7%.
- The operational margin of the corporation was 4.6% as opposed to 5.9% (this information was removed from Gap’s report).
- Adjusted diluted profits per share were $0.71 as compared to the reported $0.42. According to Gap’s report, there was a gain from adjusted income taxes of over $0.33.
- According to Gap, net sales would decline sequentially by the mid-single digits in the fourth quarter, resulting in reduced revenue for the whole year. But there is no prediction for the operating margin.
Response
Microsoft is aware of the mistakes and anticipates that the Bing AI will continue to make them.
“We’re aware of this report and have analyzed its findings in our efforts to improve this experience,” said a Microsoft spokesperson.
“We recognize that there is still work to be done and are expecting that the system may make mistakes during this preview period, which is why the feedback is critical, so we can learn and help the models get better.”
Image source: Windows Central
