Chatgpt's full name "Chat Generation Pre-Trained Transformer" (that is, "Pre-trained chat generation converter")，It is an artificial intelligence chat robot program。ChatGPT currently mainly interacts with users in text，Can complete the generating automatic text generation、Automatic Q&A、Automatic summary and other tasks。Current，Some students in some colleges and universities in the United States have completed the academic papers with the help of the software，I can even get good results。This makes some schools introduce new regulations，It is forbidden to use the software to complete the job。These phenomena have caused some philosophical teachers to worry: Does ChatGPT make philosophy teachers 'professional assessment of students' curriculum papers and degree thesis more difficult? Will some speculative students use these technologies to obtain high scores that are not consistent with their real level?

Understand the deep learning technology that ChatGPT relies on ChatGPT over the years，The author will first verify whether ChatGPT can in -depth participation in the generation of high -quality philosophical texts.。Based on the recent measurement of chat software，The author found that this suspicion has been verified。At the same time，The author tends to think，ChatGPT's possible technical progress in the future cannot automatically complete the processing of high -quality philosophical text。

Let's talk about why the author does not trust Bet365 app download ChatGPT。Like traditional deep learning technology，ChatGPT system presets the knowledge formation mode of empirical theory in philosophy。That is to say，According to this philosophy preset，As long as a cognitive system obtains a lot of data on the external environment，You can automatically obtain the law through statistical similarity inside these data，and predict the future。Specifically，Deep Learning Technology simulation human neuron network structure，Statistical processing of existing data，Converted into a "training" process for a complex artificial neuron network。Chatgpt innovation based on deep learning technology，is the introduction of the concept of "pre -training"。The so -called "pre -training"，That is to "feed" the massive corpus system first，But what tasks do you want to complete，Let the system think about the front and back relationships of different words。For example，If the system finds that when the phrase "Lu Yao knows horsepower" appears，The probability of the phrase "Seeing the Heart" is very high afterwards，I will look forward to the appearance of the second phrase after discovering the first phrase。But what you need to pay attention to is，The system doesn’t know what this idiom it means。After the complex "pre -training" stage，Let the system complete some clear tasks (such as completing a job report)。What needs attention is，Make this good performance possible "pre -training"，The amount of corpus and computing power required for the required consumption is amazing。

The principle of the author's disbelief is the principle of this technical path. This is bet365 Play online games here: humans are not learning language like this。Almost no one will pass the statistical processing of massive sentences，Obtaining a postponent probability of "Jiu Jiu Seeing the Heart" and "Lu Yaozhi Himeli" -s instead，Any normal human child understands the idiom through an adult's interpretation of the idiom。Current，Chatgpt's thirst for massive training corpus，Obviously it will cause a problem in principle: as long as the corpus is insufficient，The output quality will fall quickly。Compared with it，Human philosophy experts can study the thoughts of another philosopher through a few philosopher's corpus (we might as well think about，How rare is the philosophical text left to future generations for Hillary Lite and Bamnid)。

It is also based on the above understanding of the deep learning mechanism，I only think that the problem of philosophy is exactly the "Achilles Heel" that exposes the defects of ChatGPT。The specific reasons are as follows three points。First，Philosophy text occupied on the human Internet is very low，Congenital problems with insufficient training text。2，There are a lot of errors in the existing electronic text on the Internet，Therefore, the quality of the original philosophical training data itself is also worrying。third，Human bids specially used in philosophical text processing are also very difficult to find，Professional philosophy teachers who are competent for this work may be more difficult to let go of the scientific research work at hand to order Chatgpt Corghematic Error (need to note that，Almost all of the labels that are labeling for ChatGPT are almost cheap workers Bet365 app download from underdeveloped countries); or the other way is the same: related artificial intelligence companies are unlikely to spend high salary vocational philosophy teachers to order such categories Error (because the business return of doing this is very uncertain)。

Based on the above reasoning，I put forward a hypothesis: ChatGPT cannot respond to the questions related to philosophy raised by the user。To verify the above assumptions，The author has tested ChatGPT with Chinese and English，And this also verified the above assumptions。I found，Chatgpt's performance can be basically classified as the following three problems。

First，Chatgpt's philosophical knowledge base is very small。For example，The author requires ChatGPT to list the main works of the Japanese Marxist Takamura (1900-1945)，ChatGPT means that there is no power for this。If Tomaku is indeed a bit unpopular，Then even if Han Bingzhe didn't know, it shouldn't be a bit (he is currently a hot Korean German philosopher in the world)。Actually，Any human user can easily obtain some basic information of these two philosophers through the Internet。

Second，Chatgpt's "virtue" is very poor。Chatgpt although it is an utensil，But the utensils also have "virtue"。For example，If a dictionary deliberately collects some wrong words，Leading the user's language use level decrease，Then the "virtue" of this dictionary can be said to be poor。ChatGPT's specific performance of "virtue" is "not understanding pretending"。For example，When I asked ChatGPT, "How much is the influence of Marx's" German Ideology "by Marx，This system actually summarized several points。For example，"Tsukaka Run uses Marx's ideological bet365 live casino games theory to analyze the economy and culture of Japan at that time"; "Like Marx，Takakan Run also believes that ideology is a power to deceive the people "。And，It's hard to say that these answers are wrong。But the reason why the system "guess" the answer when facing this question，It is precisely because it has been informed，Takakan Run is a Marxist。Therefore，The system can use the existing text knowledge about "German Ideology"，Guess what to say about Japan's ideology。but obviously，A professional Japanese philosophy researcher can see at a glance that the machine is not understanding pretending，Because of the factors in the original text of the original text in the original text of the original text given in the answer given by the answer given by the machine: a comprehensive attack on the retro tide of Japan at that time。Therefore, this probability means，There is no information about "Japanese Ideological Theory" in the system。To verify this guess，I asked the system a new question: "What is the relationship between the space theory and Kant's spatial doctrine?" The result，The system indicates that there is no information in this area。Danjun's "Space Theory" is his knocking on the Japanese philosophical world，This book that I don’t know about Takakan Run，But the robot or real person who dares to talk about its thoughts，It is a very poor "virtue"。

third，Chatgpt's knowledge migration ability of cross -knowledge modules is very poor。For example，When I ask ChatGPT in English to ask the main philosophical viewpoint of Japanese philosopher Nishida Kataro，The system does give a more suitable answer; but when I ask Bet365 app download this question in Chinese，The system has irresponsiblely fabricated a few philosophical works of Nishida Karo。This shows that ChatGPT is unable to build a bridge between the English knowledge module and the Chinese knowledge module。When the author asked the Roman sound writing method "Nishida Kitarō" with the name "Nishida Kataro"，ChatGPT's answer turned out to be "Sakamoto Ryoma" (political activist in the Meiji Restoration period)。This error is very ridiculous。

Some people may say: These defects of ChatGPT will be resolved with the expansion of its language model。I don’t recognize this optimistic view。Because in a large number of philosophical papers，Only a small amount is worth reading。Therefore，Human philosophers will choose those texts worth seeing according to their own taste。and ChatGPT is essentially a digital "饕" that devour massive data -it will not choose、Chewing and reflection。Even if ChatGPT uses philosophical texts as the main training database (although the author does not believe that doing this can bring business interests)，It is impossible to distinguish between the differences between the spam text and the essence，As a result, the absorption of a minority essence must be at the cost of swallowing a lot of text garbage。Another，ChatGPT cannot handle a phenomenon that often appears in philosophical literature，What should I think about a philosophical problem。There are often amazing differences between different philosophers，This makes each unique point of view a "statistical abnormal number"。Can ChatGPT based on statistical thinking be digested with the hard pine cones? The answer Bet365 app download should be negative。Except for some existing facts，I can't see how practical help can the future ChatGPT bring to the work of philosophers (except machine translation)。

　　bet365 Play online games