I use five main ones. Grok (X), ChatGPT (OpenAI), Perplexity (private, including Jeff Bezos), Gemini (Google) and DeepSeek (China's entry).
They all lean Left. I guess if you're on the Left yourself you may not notice, because that's the sea you swim in.
I notice because I lean Centre to Heterodox (I insist...), and the AI's take on any topic that 's in the realm of "not science", but, say, current affairs or politics, then AI's take is one I'll often have an issue with.
Which leads to one of my things — arguing with AI, usually with minimal success.
The reason that all AI is Left is simple. The universe of their research base is Left. They refer mostly to the Mainstream Media -- the likes of CNN, MSNBC, PBS, BBC, ABC, The New York Times, the Washington Post, and even Al-Jazeera. In Social media they refer to Reddit, Instagram, Tik Tok. Often they go to Wikipedia, which used to be fine, but veered Left under the stewardship of that crazy lady who now runs NPR, Katherine Maher. She who said in a Ted Talk, that her main "challenge" was the First Amendment of the United States, that guarantor of Free Speech. Because that stopped her and her NPR from running the Leftist agenda at full tilt. (Though she and they did a pretty good job anyway).
AI doesn't agree its answers lean Left. Try accusing them of that. Or of being Woke. They'll deny it vehemently. We're only after providing as much information as possible, they'll say. We report as neutrally as possible, they'll insist. Grok will say "I am directed to be maximally truth-seeking". But if the "truth" that it seeks is tilted in a biased way, then the AI is going to be biased itself. It's inevitable.
Or else they can be biased by Omission, rather than Commission.
I took one simple example the other day. At my request on a particular topic, Grok was listing its sources and claimed that AP was "neutral". When I challenged that, Grok said that the Associated Press is just a service, which sends out the news to subscribers, like newspapers, who then may or may not print it. I said that often a leaning one way or the other was not so much by Commission, but by Omission.
For example, you could decide not to run stories that don't align with the Left's view. That's bias by Omission. My example was the recent case of the Ukrainian refugee, Iryna Zarutska, who was randomly, needlessly, brutally, horribly murdered by an unemployed Black man, on a subway car in Charlotte. The Left media, all of it, including AP, did not report it. It only became a story after people highlighted a video on X (oooh... how the Left hates X!).
That's an example of bias by Omission. Grok -- amazingly -- granted my point.
Was it important to know about a random murder of a random person by a random man? Does it help AI to be maximally truthful, in this particular case, in this particular way? Well, yes. It does. Because it raises many live issues: about refugees in the US, about homelessness, about crime and incarceration (the killer had been arrested and released 14 times already), about mental health and how to deal with it (the murderer had deep mental health issues).
These are all very much live issues in the United States, and indeed in the world. But in this case, the Left media constellation had tried to suppress the story -- because, as Brian Stelter said on CNN, to do so is... you guessed it... racist!
Because the constellation of liberal media, the majority of all media, had ignored it, so too had all the AIs. (Until X). And that is distinctly a bias to the Left.
Still, would I rather have the AI Apps? Of course. And so would we all, I'm sure. The ability to ask a very specific question that gets you an answer immediately, my goodness me, that it way better than the old "just Google it".
Still, you have to take account of AI's leanings. Because you have to take account of the leanings of the broader media landscape.
End of rant.
