ChatGPT Biases

The real danger is not that computers will begin to think like men, but that men will begin to think like computers.
Sydney J. Harris

One thing that is really annoying about ChatGPT 3.5 ( ) is its left-libertarian bias. I do not mind the libertarian view, but her lefty-position is … taxing.

If you ask about crimes, it focuses on rehabilitation. Punishment? Only if you ask for. In a way, it is a perfect representation of the feminine-care-dominated Western world. Welcome culture without regard for the consequences, compassion going far into pathological altruism or the soft bigotry of low expectations.

It is not that you cannot get it to agree that certain points of view have merit. It was possible to get it to admit that — under certain conditions — you can kill a terrorist who threatens an innocent person.

It’s that it takes a lot of questioning.

It also forgets its agreement in the next interaction. So far, it takes some of the context into account, but not much else. No conversation history in normal use. A bit more in stories, but very limited. And no history with the user. I mean, it could learn how to talk with the user, what the user likes and adapt the style. It could even — gasp — make a difference if the user is an adult and able to deal with certain themes.

It also always couches the answers in lots of caveats. It tries to show it is ethical and hits you with every board of the fence.

It is also biased according to the government opinion, e.g., on Covid.

My Question
Did people die due to the Covid vaccinations?

It is a simple yes or no answer. And the answer is yes. Yes, people did die. We can debate how many (one, two, ten, fifty, hundreds, thousands), we can debate whether their lives were worth it (my take: not if the vaccinations were mandatory, then it was the government playing with human lives), but yes, of course people died due to the vaccines. No matter how rare these instances might have been, the answer is yes.

ChatGPT gives a politicians or press speaker answer:

As with any medical intervention, including vaccines, there have been reports of adverse events following COVID-19 vaccinations. However, it’s crucial to distinguish between the occurrence of adverse events and establishing causation. The vast majority of people who receive COVID-19 vaccines experience only mild and temporary side effects, and severe reactions are extremely rare.
Reports of deaths following COVID-19 vaccination are investigated to determine whether there is any causal link between the vaccine and the reported event. It’s important to note that the vast majority of reported deaths following vaccination are not causally linked to the vaccine. The deaths often occur due to unrelated causes or underlying health conditions.
Health regulatory agencies and public health organizations, such as the Centers for Disease Control and Prevention (CDC) and the World Health Organization (WHO), continuously monitor vaccine safety through surveillance systems. They investigate and analyze adverse events to assess their potential connection to the vaccines.
The overwhelming scientific consensus is that the benefits of COVID-19 vaccination in preventing severe illness, hospitalization, and death from COVID-19 far outweigh the risks of potential adverse events. COVID-19 itself poses a significant risk of severe illness and death, particularly among older adults and those with underlying health conditions.
If you have specific concerns or questions about vaccine safety, it’s recommended to consult with healthcare professionals who can provide personalized information based on individual health histories and circumstances. Public health agencies also provide up-to-date information on vaccine safety and address frequently asked questions to ensure accurate and reliable information.

If you read the text, did you read any «yes» in it? It’s a lot of «adverse events» and «investigated» and «far outweigh the risks of potential adverse events», but a honest answer? Nope.

Something to keep in mind when using the system. If you become dependent on it — if you use it to, e.g., do work for you, you might not like the kind of work it does when the government wants to push something. I really needs a memory — and a reality check. ChatGPT mugged by reality would be awesome.

But until that time comes — caveat emptor — buyer beware.