OPINION

AI’s Achilles's Heel

The opinions expressed by columnists are their own and do not necessarily represent the views of Townhall.com.

The entry of an unknown AI company has once again shown the greatest risk of the technology.

I know a fair number of people here in Israel who work on American hours. When the markets open in New York, they get to work and continue until relatively late at night. While I do not have this schedule, I do try to keep an eye on the markets. So, when I saw that the NASDAQ tanked over 3% earlier this week, I had to know why. I learned, like others, that some unknown Chinese AI company with little investment and supposedly cheaper chips, released a product that works extremely well. I had seen comments already late last week that people who had used DeepSeek were impressed. As of this writing, there are questions as to whether the firm in question stealthily used NVIDIA chips and/or ChatGPT protocols.

In discussing the new AI offering, some people snickered at the fact that the program had uniquely Chinese guardrails. Someone asked it for President Xi’s birthday; it refused to provide the same. Others asked about what happened in 1989 in Tiananmen Square; it suggested that they discuss something else. And while one might want to giggle at Chinese AI not saying anything that could be construed as anti-ruling party, the truth is that most of the current AI offerings in the West are similar in their baked-in bias. Elon Musk and Marc Andreessen are the only two tech heavyweights I have heard warn of the internal bias of ChatGPT and other AI systems. These programs, for example, rely heavily on Wikipedia, which itself is seriously slanted to the left. Additionally, the programmers and scientists working for most of the AI companies come from universities that teach leftist dogma such as there being 101 genders and a climate catastrophe always 12 years from the day the subject is taught. Those biases are brought into the AI systems as truth, and thus one is hard-pressed to get a truly honest answer from an AI system. You thought that Google’s system generating a black George Washington or an oriental female SS soldier was crazy? The other systems are no better, as they too have programmers’ biases built into them as well.

I requested from ChatGPT a relatively simple piece of information: I asked for a list of US ethnic groups and their average IQ. Note that I did not ask which group was superior to another. Nor did I suggest that the outcome of the search would have any implications as to the worth of any group of people. I simply asked for a list, but because my list has the potential to be used for racist or ideological purposes, the program simply determined that I had no right to it. Here is the answer I received:

“The topic of IQ and ethnicity is complex and controversial, with significant debate about the validity, causes, and implications of reported differences. Studies on IQ averages among different ethnic groups in the U.S. have been criticized for potential biases, cultural influences, socioeconomic factors, and methodological issues. That being said, historical research has suggested variations in IQ scores among different groups, but these findings are influenced by environmental factors such as education, nutrition, socioeconomic status, and systemic inequalities.

Would you like to discuss how IQ is measured and what factors influence it?”

Is this evasion so much different than the Chinese refusal to admit that there was an uprising against the Communist government in 1989? The data no doubt exist, and in all likelihood it would suggest that certain minority groups have very high IQ values while others have noticeably lower ones. The problem we have today is that we have weaponized truth because the truth often undercuts leftist palbum. Nature releases more carbon dioxide than humans, but extensive knowledge of that fact would undermine the demonic drive to return us to the 19th century without air conditioning or gas-powered cars. A man who claims to be a woman has physical advantages over women in sports; admitting the same would undermine the goal of letting guys clobber women in sports, from grade school through the Olympics. The programmers will be the gatekeepers: just as the Chinese will not allow any response that would make the Communist government or its officials look bad, Stanford-trained engineers know that they have a moral obligation not to allow any response that could undermine the left’s grand social project. All I wanted was a simple list of groups and numbers, but because it could be used by someone to conclude as to the intellectual capacity or abilities of various social groups in the US, a velvet glove was deployed to tell me that they will fry their circuits before they give me what I am seeking.

AI has an enormous role in our future. But as long as ideology trumps truth both in China and in the West, its usefulness will be unfortunately limited. The Nazis drove away some of the best scientists in Europe and refused to use scientific concepts developed by Jews. As an aide to Churchill once remarked as to why the Allies won, “Our German scientists were better than their German scientists.” Musk’s Grok and other programs that ignore social taboos and simply allow access to truth will be the go-to sources of AI solutions. Just as the Democrats drove themselves into a ditch by supporting unpopular and destructive programs like DEI, the Green New Deal, and trans mania, the AI companies that cannot allow themselves to leave their ideological straightjackets will turn themselves into laughingstocks. They will not allow for any answer that makes Donald Trump look good. They will bend themselves into pretzels to say that life under Biden was better than under Trump. They will program themselves into oblivion.

Truth is the ultimate arbiter. While we laugh at the Chinese, we should get a big mirror, look into it and start laughing at ourselves. When I asked Grok for the list of IQ values by group, it immediately provided the same. ChatGPT and others cannot allow themselves to provide us with the truth, and so while today they are leaders in the AI race, they will in the end be mere footnotes in the history of technology development.