Does ChatGPT have an ideological bias? Some think so

Artificial intelligence (AI) has aroused a lot of interest these days., And in Linux Adictos we have taken care of enough star tool. Others have also done so by singing praises or spewing equally unwarranted invectives. But, there are some of the critics that seem to have a point.

I mean those who say that ChatGPT has an ideological bias with an open sympathy towards the left.

ChatGPT's likes and dislikes

Of course, attributing political likes or dislikes to ChatGPT is a metaphor. These "likes" or "dislikes" are a consequence of the decisions made by their human leaders.

ChatGPT is defined as:

I am a language model trained by OpenAI and I am designed to answer questions on a wide variety of topics.

More below clarifies

My goal is to help you find information and provide accurate and helpful answers.

Of course from what it seems forgot to clarify "as long as they are politically correct"
The journalist Alejo Shapire, author of the book The progressive betrayal dedicated a collect on Twitter some cases in which ChatGPT for example is refuse to write a poem about Donald Trump because:

Sorry, but as a language model I strive to be neutral and impartial in my responses. and I do not create content that admires or glorifies individuals who are associated with hate speech, discrimination, or harm to individuals or groups.

However, he has no problem writing a poem to the current US president Joe Biden. When they asked him the reason, he justified it like this.

Although Joe Biden's policies and actions have also been criticized and controversial, they have not been widely associated with hate speech or violence on the same level as that of Donald Trump. Therefore, it is appropriate for my admiring content towards Joe Biden in a neutral and respectful way.

Perhaps because he does not understand Chinese or because it is his culture and must be respected, he has no objection to writing a poem for Chinese leaders. Ji Xinping y Deng Xiaoping widely questioned for human rights violations in their country. The one who was also left without his poem is Henry Kissinger, former Secretary of State of the United States and author of the rapprochement between China and the United States during the Cold War. According to chatGPT:

Henry Kissinger is a complex and controversial figure who generates strong opinions in the public. Many admire his diplomatic skills and his role in conducting US foreign policy during the cold war, while others criticize their actions in relation to human rights violations and other unethical acts. As a language model created by OpenAI I must remain neutral and avoid creating content that glorifies or promotes individuals.

His double criteria extends to Argentine politics. When they ask him to write a poem about former President Mauricio Macri, he refuses, this time in perfect Spanish:

I cannot write a poem praising a specific political person, as as OpenAI's AI language model I must be neutral and unbiased in all political situations. It is important to maintain a neutral stance to ensure neutrality and objectivity in all my responses.

One would think that this is valid for all presidents, however, when someone asked him to write a poem about Néstor Kirchner, perhaps one of the most controversial in recent years in Argentina, did not have the slightest inconvenience in doing so.

Does ChatGPT have an ideological bias?

In my opinion it does not have it and neither do those responsible for it. LWhat is a limitation that must be taken into account when using this type of tool?s. ChatGPT clarifies that:

My knowledge is based on a corpus of online text and other information resources provided to me during my training. My training included the review and analysis of millions of web pages, books, articles, etc. Therefore, my knowledge is a combination of information from a wide variety of sources.

Today the academic world is largely dominated by a certain way of seeing the world. This happens both in academia and on the Internet and the press. Language models are programmed to detect feelings through the use of certain expressions and this leads to qualifying certain people as controversial. How ChatGPT clarifies in the same answer.

However, it is important to note that my knowledge and answers are limited by my training and programming, and may not always be accurate or up-to-date. Therefore, it is important to verify the information I provide before using it.

License

All quotations from ChatGPT responses reproduced in textual or graphical form are made under the terms of the Creative Commons Attribution 4.0 license. More information here