About Baidu Wenxin Yiyin's opening to the whole society as well as the new reconstructed AI native apps, there are a few noteworthy information:
1. Baidu Wenxin Yiyin's opening: Baidu Wenxin Yiyin is a semantic comprehension model developed by Baidu, which comprehends the user's inputs and generates text responses automatically. This opening means that Baidu will allow a wide range of users and developers to use the functionality of Wenxin Yiyan, thus applying semantic understanding technology to more scenarios.
2. Newly reorganized AI native apps: This means that Baidu has made comprehensive improvements and upgrades to its AI apps to enhance user experience and functional performance. Specifically, the app may undergo improvements in interface design, functionality expansion, and performance optimization to better meet user needs.
I'll talk about some of the experiences of actually using it:
Answer by Recognizing Pictures
Just after I opened Wenxin Yiyin, the first feature that attracted me was the uploading picture icon on the chat bar, and I immediately chose a picture to upload. After that, some function tips will pop up, including: look at the picture to write a poem, V50 copywriting, friend circle matching assistant and so on. I chose a matching text assistant, and the style was entered as light-hearted and humorous.
Well, I was blown away by the pompous copywriting, and the accompanying text simply let itself go. I can't imagine what the people around me would think of me if I sent this copy to my circle of friends, and I was so scared that I had to tell it to please keep it: high, concise, and few words. It also read my embarrassment and came back to me with eight words: enjoy the food, feel good. Wonderful!
Discovery: more built-in answer templates
The mobile discovery feature should be considered the biggest highlight of Wenxin Yiyin, and it can be said that it lowers the threshold of AI usage a lot. Before I share the website built with the interface to my friends, the biggest problem is that people don't know what they can ask, how to ask, and Wenxin Yiyan is a good solution to this problem, the ability to put a lot of models scenarios.
From the point of view of the modal ability, then, mainly including the text born of the text, the text born of the map and the map born of the text of three kinds. First of all, let's try the effect of a Little Red Book store visit text.
This result is still a surprise overall, you can see that the heart of the text in the Prompt project is still hard work. Basically meet the basic elements of the Little Red Book store visit copy: format, expression, label, the ability to extract and analyze the keywords is not bad. But at the same time also tried to use this Prompt to write other types of small red book copywriting, such as picnic, came out completely ineffective. It shows that this Prompt is specifically optimized for the scene of the store, does not have the ability to migrate. Overall the quality of the Prompt is not bad, and the AI answering ability is quite good.
AI painting
Next test the ability of the AI painting, first of all, two painting words that have been criticized for a long time before: car and red lion head. Previously it was said to draw car, water, horse, dragon, and the real lion's head, so let's see if the ability has grown now.
Surprisingly, I got both of them right. I don't know if it's because the translation function has been upgraded or the drawing ability has been upgraded or if it's been optimized for the target, but in any case, the slots are a lot fewer (and how it feels like the fun has been reduced as well haha).
From the drawing comprehension and drawing out of the effect, I should say that I can make a pass mark now, and will not expect it to be able to draw much better, if there are any articles need to match the words, with this to draw is also a simple and easy to start the choice. However, it is worth complaining that the bottom right corner of each painting will be labeled: AI painting, this is a little difficult to use, but also to P off.
Additionally, the paintings do not support adjustments through continuous dialog, and can only be repainted by entering a full description.
Code Capabilities
Finally, let's compare the code capabilities of Heart of One Word below, which belongs to the category of being an important manifestation of AI capabilities and can be used as an important part of productivity.
There is a python code requirement for merging Excel files, and the overall writing is quite simple and elegant, and the quality of the answer is not too far from ChatGPT, which is a level that can be used. (I'm not going to compare it to GPT4 here, there are still some gaps)
Summary
Before I wrote this article, I read a lot of articles about the use of Wenxin Yiyin, and there were a lot of laughable stories. I thought it would be full of slogans when used, but it wasn't. Wenxin Yiyan's ability and foreign big models certainly still have some gaps, but this mobile app as a whole is still relatively sincere, the overall use of very smooth, very fast answers, with enough high-quality Prompt templates, there will be a lot of thoughtful tips between the question and answer, the whole brings a good experience.