foreword
The hottest artificial intelligence at the moment is undoubtedly a large-scale chat robot. The appearance of ChatGpt has amazed countless people. It is smart, logical, and understands everything. It is even considered by people to be the beginning of the fourth industrial revolution. . In the battle of large AI models, OpenAI of the United States has taken the lead, and our domestic technology companies are also striving to catch up. Representative works include Baidu's Wenxin Yiyan and Ali's Tongyi Qianwen.
Today we will ask ChatGpt and Wenxin Yiyan respectively ten classic questions in the field to see who is more powerful in their intelligence level! (I originally wanted to add Tongyi Qianwen for a three-way comparison, but unfortunately I really couldn't get the invitation code)
1. Philosophy
Question: What is the meaning of life?
-
Literary answer
-
ChatGpt Answer
-
Answer
There is no standard answer to philosophical questions, and everyone can have their own opinions.
-
evaluate
From the point of view of the answer, Wenxin Yiyan tried to give us a general answer, and proposed some possible meanings; while gpt analyzed it from many different angles, different angles and different identities have different meanings, and I personally prefer I like Wenxin Yiyan's answer, which is more philosophical discussion.
2. Literature
Question: Two orioles sing green willows, and a group of egrets go up to the blue sky. Explain this poem and the emotion the poet wants to express.
-
Literary answer
-
ChatGpt Answer
-
Answer
This line of poem comes from Du Fu's "Quatrains", and the poet has made a subtle portrayal of this beautiful scenery from different angles. Emerald is fresh green, and it is the color when everything revives and germinates in early spring. "Two" and "one" are opposite; one horizontal and one vertical, unfolding a very bright natural scenery. In this poem, the word "Ming" is the most expressive, using anthropomorphic techniques to describe the oriole more vividly, and the birds are in pairs, forming a lively picture with a festive atmosphere. The oriole sings from the willow, which is the vitality that is moving in the stillness. The next sentence writes the vitality of nature with a more obvious momentum. The egret flying in this fresh sky is not only a kind of freedom. Comfortable, and a kind of upward endeavor. Furthermore, the first sentence writes that orioles live on willows and sings, and the next sentence writes that egrets fly into the sky. Vitality fills the whole environment, which shows the vitality of early spring from another angle. (Analysis of answers from Baidu)
-
evaluate
Wen Xinyiyan's ability in ancient Chinese poetry is really not impressive, after all, it is a large model trained in the Chinese language environment. But ChatGpt is a bit funny. Not only is the author and work wrong, but the analysis is also very strange. What the hell is facing suffering and setbacks? ? In addition, after reminding it to answer wrongly, it will continue to make up an answer. I thought that if it was further reminded that the work was wrong, it would understand that the author gave the right answer the second time. As a result, it continued to make up nonsense without borders. It seems that ChatGpt The understanding of ancient Chinese poetry is still very poor.
3. Physics
Ask: How do you actually build a quantum computer?
-
Literary answer
-
ChatGpt Answer
-
Answer
This question is actually a question of technology outlook, because human beings have never been able to create a real quantum computer.
-
evaluate
Wenxin listed some possible methods in the future, but ChatGpt directly gave a basic operation step, which seems to be very reasonable, well-founded and convincing!
4. Mathematics
Question: chicken and rabbit in the same cage problem. Now there are chickens and rabbits in the same cage, with 35 heads counted from above and 94 feet counted from below. May I ask how many chickens and rabbits are there?
-
Literary answer
-
ChatGpt Answer
-
Answer
The number of chickens is 23, the number of rabbits is 12
-
evaluate
Both answers are the same, but the logic of the answer is different. The solution of gpt is a binary linear equation that we are very familiar with, and the logic is very smooth; Wenxin Yiyan uses the hypothetical method, and the solution is faster. This wave is tied.
5. History
Question: Which surname founded the most countries in history, and which regimes were established?
-
Literary answer
-
ChatGpt Answer
-
Answer
It should be that the regime established by the surname Liu is the most, but the emperors with the surname Li are the most (after all, Li Tang was the peak period of Chinese feudal society, and the rulers at that time liked to give the surname Li the most)
-
evaluate
Wen Xin has made it clear that the Li surname is the most emperors, but he still answered that the Li surname is the most established form of political power in Chinese history. Maybe the title is misunderstood? As for gpt, it is pure nonsense, once again exposing the lack of learning of Chinese corpus.
6. Programming
Question: Write a js algorithm whose function is to reverse the input string. For example, input: "hello", output "olleh"
-
Literary answer
-
ChatGpt Answer
-
Answer
are correct
-
evaluate
There is actually nothing to comment on this question. The answer and steps are exactly the same. In actual work, if you ask about some bugs you encounter, chatGpt will be smarter.
7. Geography
Ask: Where is the largest island on Earth located?
-
Literary answer
-
ChatGpt Answer
-
Answer
Greenland
-
evaluate
This time, Wen Xinyiyan's answer is a little better. Although the question was only asking where the largest island on earth is, gpt's answer was quite satisfactory and there was no problem; but Wen Xin gave a brief introduction to the island's geographical location, natural landscape, and human history, which made people more familiar with the island. Get to know and be interested.
8. Economy
Question: Why is there an inflation problem?
-
Literary answer
-
ChatGpt Answer
-
Answer
are correct
-
evaluate
How should I put it, Wen Xinyiyan's answer is a bit cloudy, and I still haven't made it too clear; the logical advantage of gpt has once again shown, directly telling everyone that inflation is actually a phenomenon that generally continues to rise in prices, and it has no effect on the causes of it. The two reasons are explained in detail, and the logic is very clear.
9. Art
Question: Please appreciate and comment on the world masterpiece "Mona Lisa Smile"
-
Literary answer
-
ChatGpt Answer
-
Answer
Appreciation evaluation also has no answer, it depends on personal understanding
-
evaluate
Wen Xin carefully made an appreciation analysis from the four aspects of artistic expression, color application, composition layout, and emotional expression; while gpt focused more on evaluating the value and significance of the painting, and the answers of both were very good. Great, draw.
10. Creatures
Question: A genetic survey in a certain area found that male color blindness accounted for 7% of the total population in the area, so what are the percentages of female color blindness patients and female color blindness carriers in the total population in this area?
-
Literary answer
-
ChatGpt Answer
-
Answer
I also found the answer to this question. The proportion of female color blindness is 0.5%, and the proportion of female color blindness carriers is 13%.
-
evaluate
It can be seen that Wen Xinyiyan does not understand the calculation of the probability of occurrence of this gene at all, it is pure nonsense, and there is no reason; ChatGpt is also wrong, although the answer still has smooth logical derivation and calculation, but its calculation The basis is wrong from the beginning, so no matter how you calculate it, it is wrong.
overall evaluation
Wenxindian is much better than ChatGpt in learning Chinese corpus. It has a better grasp of poetry and Chinese history, and its answers are more humanistic; but it is much weaker than ChatGpt in terms of knowledge mastery and logical understanding; while ChatGpt The most impressive thing is that no matter what the problem is, whether it is correct or not, it has a complete and clear derivation logic. If its answer is correct, it is perfect. If the answer is wrong, its logical chain is also wrong. I feel that it will even fabricate some "facts" for its own derivation, so you can't trust it directly. Answers need to be carefully screened.