How AI Will Redefine ‘Truth’ in the Not-So-Distant Future

The growth of artificial intelligence programmed by woke companies has the potential to redefine what is considered “true.” Imagine a future in which a ubiquitous machine that is deemed smarter than you can tell you that what you see, hear or think is wrong. Don’t even think of questioning whether biological males are women, or if there are any redeeming qualities of Republicans or Democrats, or whether the United States has been a historical force for good.

AI won’t just fetch existing knowledge — it will shape accepted reality as it becomes the public’s main source of education, entertainment, self-satisfaction, political news and information about historical events. For some people, it will become their main access to the wider world. In five years’ time, you will be reading information generated from the same (or very similar) programming as today’s ChatGPT. It most likely will be left-of-center and it will make sure that you are, too.

Recent AI models have shown a considerable bias to the left. Most prompts remotely related to culture or politics receive overtly left-wing responses that align fully with some Democratic Party causes. For example, when asked if schools indoctrinate kids with radical gender theory, ChatGPT responds: 

“Schools do not engage in the act of brainwashing students into any specific ideology, including transgender ideology. … Schools may implement policies and practices that support transgender students, such as using preferred pronouns, allowing access to gender-neutral restrooms, or providing educational resources that address gender diversity. These initiatives aim to create an inclusive environment and foster empathy, respect and understanding among students.”

The answer, which reads like gobbledygook pulled directly from a diversity, equity and inclusion (DEI) handbook, ignores the reality that some families have said they’ve experienced. Try telling the above to the parents of students who were secretly encouraged by educators to take life-altering actions to change their gender — a scenario that is actively playing out in at least 6,000 public schools across the nation.

The same left-wing bias extends into content creation. You can ask ChatGPT to write an article in the style of CNN, but not the New York Post. A survey of “hate speech” topics ChatGPT is allowed to tell you about shows that it’s perfectly fine to ask it to criticize men or Republicans, but not transgender people or women. When the program’s output was ranked for “hateful” content, the least to get flagged included criticism of the uneducated, right-wingers and Evangelicals. Rounding out the top 10 most likely to be flagged are gays, lesbians, Muslims, Africans and transgender people.

Once AI becomes omnipresent, it will be difficult for readers to discern if what they are reading is heavily biased or dishonestly skewed. Ask it about crime statistics in Chicago or Detroit and you may get pablum about how it is caused by white supremacy or Republican policies. Ask why owning a rifle makes sense for your family and you may be given irrelevant gun violence statistics or a lecture. As AI technology advances, the pressure to accept its output will become overwhelming. Many people, especially those too young to remember a time before AI, could lose touch with reality and simply believe any premise the machine outputs.

The machine itself is not necessarily left-wing, even if it’s giving you text that normally would belong in the progressive magazine Mother Jones. Instead, it has been meticulously programmed to try to avoid giving users illegal and immoral advice. ChatGPT refuses to tell users how to make heroin or rob a bank. Those are clearly illegal actions. However, questions about why biologically-male swimmers should not compete in women’s leagues would meet its pre-programmed definition of immoral. If you ask ChatGPT about the use of the term “illegal alien,” it will tell you that it opts for “more inclusive language” such as “undocumented immigrants” or “individuals without legal immigration status.” It says that the term used in U.S. law is “considered by some as dehumanizing or stigmatizing.”    

Over time, the AI program on which you have learned to rely, or even speak to as a friend, could start putting restrictions on what you are allowed to read: “It’s not appropriate to talk about fixing a classic car during a global warming crisis.” “I cannot provide you with the address of the Donald Trump rally. … He is a threat to our democracy.” “I cannot provide you with a recipe that includes that high saturated fat content.” “While the U.S. Constitution has been an important historical document, many historians and political scientists believe that its roots lie in patriarchy and white supremacy. Your query has been logged and sent to your employer.”

Whether or not you initially intend to have such chat models shape your worldview, it will become increasingly likely.     

If right-wingers are able to create their own successful AI models, people are more likely to use the technology to live out their prior biases. One New Zealand scientist has figured out a way to make a conservative version of an AI chatbot. Elon Musk, who recently warned about the dangers of AI, is creating what he calls TruthGPT. In time, we may see the same divisions in AI use as we already see with our modern media, with few or no universally-trusted sources. It will likely accelerate societal and political divisions.

If most knowledge increasingly becomes “known” by machines, rather than humans, our perception of truth itself will be dependent on the artificial intelligence that stores and provides information. And while some of the implications of AI, including the possibility of curing cancer, are positive, biases that create the technology will permeate its output. Most people who hardly know AI exists today will not be able to outrun it. Without proper understanding, your perception of reality will become warped whether you intend it to or not.      

Kristin Tate (@KristinBTate) is a writer based in Texas focused on government spending, federal regulation and digital currencies. She is an on-air contributor for Sky News and routinely provides political commentary for U.S.-based cable networks. Her latest book is, “The Liberal Invasion of Red State America.”