Before reading this article, click "Follow" in the upper right corner, which is convenient for you to discuss and share, and can bring you a different sense of participation, thank you for your support!
At 2 p.m. on March 16, in the press conference hall of Baidu's headquarters in Beijing, Robin Li hurriedly stepped onto the stage, holding the microphone and ushering in a highly anticipated moment, he was about to introduce a high-profile new product - Wen Xin Yiyan.
Although Robin Li is a giant in the technology industry, he has been named a cover character by Time magazine and has appeared on countless big stages, but standing on the stage of the press conference at this moment, his nervousness is almost overflowing.
This conference is the most eye-catching event since Baidu's establishment, both inside and outside the industry, full of expectations for this highly anticipated product that competes with ChatGPT.
On the one hand, people are full of expectations that Wen Xin's words will create remarkable performance. However, there are also many people who hold a wait-and-see attitude, hoping to see if it will go wrong and expose its shortcomings.
At the press conference, Baidu demonstrated the multiple capabilities of Wen Xin in five different scenarios. From literary creation to commercial copywriting, from mathematical logic to multimedia creation, Wen Xin Yiyan shows the possibilities of many fields.
In terms of literary creation, it shows a summary of the content of The Three-Body Problem and demonstrates the ability to continue the "Three-Body Problem" from a philosophical perspective. And in terms of business copywriting, it is able to create Slogans based on the type of business of the company, and it can also write press releases.
In addition, Wen Xin's words also demonstrated the ability of mathematical logic calculation, such as solving the classic chicken and rabbit cage problem. At the same time, it is able to provide feedback based on user needs, including images, voice, and video. For example, create posters for the 2023 World Intelligent Transport Conference, answer questions in Sichuanese, and even generate videos.
However, it's worth noting that the presentations at the launch are pre-recorded, not live demos. Therefore, many people still have doubts about the true performance of Wen Xin's words.
However, the bad reviewer obtained an internal test account and began to test Wen Xin's ability to speak. First, they tried everyday problems, such as how to make a squirrel cinnamon. Wen Xin gave detailed recipe steps, and NewBing provided a similar answer.
They then came up with a classic math problem involving a frog jumping into a well.
It takes 8 days for Wen Xin to answer correctly, while NewBing gives two answers, 5 days and 8 days, in different modes.
Then, the bad critic tested Wen Xin's programming ability and asked to generate CSS code for the gradient button. Although Wen Xin's code works, there is a misunderstanding in the process of understanding "gradient color". NewBing, on the other hand, correctly generated the gradient code and also included a "click me" prompt.
Bad critics also compared Wen Xin's performance with NewBing's performance in naming and black language understanding.
Wen Xin's advice seems mediocre in some cases, while NewBing is closer to the correct answer in some cases.
On the question of information search, Bad Reviewer asked about the sales of Li Auto in the past five months. However, Wen Xin's answer did not provide accurate data, and compared to NewBing's answer, the performance was relatively poor.
Finally, they tested Wen Xinyan's unique drawing ability, asking them to generate pictures of hedgehogs tapping keyboards and small horse-drawn carts. Wen Xin's words excelled in the picture of embroidered tapping on the keyboard, but when generating the picture of a small horse-drawn cart, there was a misunderstanding.
In general, in response to Wen Xin's test, bad critics believe that this product performs average in many aspects, and even has some deficiencies. Compared with GPT-4 and NewBing, Wen Xin's performance still has a lot of room for improvement.
However, despite this, bad critics believe that Baidu's courage is commendable. At the moment when ChatGPT is prevalent, Baidu still dares to launch Wen Xin to show its confidence in the product. For a company with search engines at its core, this move could trigger disruption in its core business. However, Baidu seems ready for this change.