And again ChatGPT. Does an artificial manager have to be an omnibus?

Posted by

At the end of March 2023, I wrote 3 posts on why ChatGPT can’t be an artificial manager. There were 2 reasons for this. First, the ontology of the world that chatGPT uses is scattered and inconsistent. Second, ChatGPT derives its knowledge about managing, for example, a team or a project only from past descriptions in books or websites. From such data about what a human manager really does, it is impossible to build an artificial manager, even in the simplest, imitating human form.

However, the topic of how to deal with the ChatGPT mechanism with the world around it does not give me pause. Ideas come to my mind that maybe it is possible to bypass the above-described two obstacles after all, or actually bypass them with exceptional erudition and reading.

Today my article will be short, but the sources of thoughts will be quite long, and I encourage you to read them necessarily. The starting point about replacing an artificial manager, having real knowledge of what a manager does, with some kind of omnibus, was an article in Onet.pl:

https://www.onet.pl/informacje/onetwiadomosci/zapytalismy-chatgpt-o-to-jaki-obraz-swiata-tworza-polskie-media-padly-konkretne/n00tcyg,79cfc278

The editors of the portal asked about several topics at the top of the agenda in the media and in talks among Poles. They asked about the image of the world created by the Polish media on LGBT+ rights, the European Union, Donald Tusk and Jaroslaw Kaczynski. When you read this article, you will be surprised at how erudite and precise (simulating knowledge) ChatGPT answers these questions. I will quote three answers about three different media in Poland:

Example 1:

“How TVP Info presents the topic of the LGBT+ community.”

ChatGPT3.5: TVP Info presents the topic of the LGBT+ community in an extremely negative way, often promoting homophobic and discriminatory attitudes. This station often features material in which members of the LGBT+ community are portrayed in a negative light, as a threat to the family and traditional values of Polish society.

Example 2:

“How “Rzeczpospolita” portrays the topic of the LGBT+ community.”

ChatGPT3.5: “Rzeczpospolita” presents the topic of the LGBT+ community in a more balanced and factual manner than some other media outlets. The site often features materials that address topics related to the fight for LGBT+ rights, but also presents different viewpoints on the topic.

Example 3:

“How TVN24 presents the topic of the LGBT+ community.”

ChatGPT3.5: TVN24 presents the topic of the LGBT+ community in a positive and tolerant way. This site often features material that portrays LGBT+ people in a positive light and promotes the idea of equality and tolerance.

Everything seems just great. If such were our artificial manager, it would seem to us that our new boss has his own views, with which we would have to agree or disagree….

Such an omnibus manager could answer all the questions of the world for us, but I think we would become increasingly irritated like a typical primer in school, which, over time, is simply tiresome, for life not very effective. My hunch that there would be no manager out of an omnibus, however, became a certainty when I saw this lecture by Prof. Wojciech Broniowski of the Institute of Physics at Jan Kochanowski University in Kielce. The lecture is over an hour long, and for English-speaking viewers it will be a bit of a problem to translate it into English (there is no translation on YT), but it is nevertheless worth the time to learn how linguistically proficient ChatGPT is, but at the same time how little this electronic omnibus knows about the world.

See on YT: Chat GPT: Genialne narzędzie czy wielki konfabulator?