It may be possible that the "conquer the world with bots and a purely unemotional approach" won't work either.
And in fact, the thing the TechBros are planning on to accomplish their goals of dominance and control are the very things that will defeat them.
New chatbot on Trump’s Truth Social platform keeps contradicting him
An “answer engine” on Trump’s social media site says the 2020 election wasn’t stolen, tariffs aren’t boosting the stock market and Barack Obama is seen favorably.
President Donald Trump and the new AI search tool on his social media network, Truth Social, don’t exactly see eye to eye.
Truth Search AI contradicts the president by saying that tariffs are a tax on Americans, the 2020 election wasn’t stolen, and his family’s cryptocurrency investments pose a potential conflict of interest. Asked about Jan. 6, 2021, it said the “insurrection” at the U.S. Capitol was violent and linked to Trump’s “baseless claims of widespread election fraud.”
Trump and his allies, after years of criticizing tech companies and news organizations as biased and untruthful, have worked to develop Truth Social as part of an alternative social media ecosystem where they’ve said their viewpoints will not be suppressed.
But as companies expand their use of chatbots and “answer engines,” the rollout of Truth Search AI highlights one challenge for that approach: Artificial intelligence tools don’t always give the answers their owners might want or expect.
“Their own AI is now being too ‘woke’ for them,” said David Karpf, a professor at George Washington University who studies political communication, using a term commonly employed on the right to describe liberal viewpoints.
Trump Media and Technology Group, Truth Social’s owner and parent company, unveiled the tool Wednesday, calling it a “public beta test.” The company cited an executive at the search engine’s developer, Perplexity, saying the tool offered “direct, reliable answers” and would “bring powerful AI to an audience with important questions.” The chatbot is free for all Truth Social users.
AI search engines and chatbots are trained by being fed large amounts of information from across the internet. But the “black box” nature of AI tools makes it challenging for developers to fully control what they say.
Many of Truth Search AI’s answers to questions posed by The Washington Post linked to sources based on conservative news outlets, such as Fox News, Newsmax and the Washington Times. But the tool did not say specifically which sources it had drawn from.
Jesse Dwyer, a Perplexity spokesman, said that Truth Social had used a “source selection” feature to limit the websites the AI tool relied on but that Perplexity did not know which websites those were. “This is their choice for their audience, and we are committed to developer and consumer choice. Our focus is simply building accurate AI,” he said. (After this article was published, Dwyer clarified that Truth Social probably had used source selection but that he could not be sure because Perplexity does not see or control what any developer is doing with the company’s application programming interface, or API.)
The White House declined to comment. Trump Media spokeswoman Shannon Devine said: “With transparently asinine stories like this, Washington Post reporters indict themselves as irrelevant partisan hacks who will probably soon join the growing exodus of left-wing shills from the paper.”
Trump Media sued The Post alleging defamation in 2023, saying the news organization had reported incorrectly on allegations relating to its early financing. The case is ongoing.
The tool’s politically inconvenient answers, Karpf said, show the limits of any attempt to recast or contest the prevalent view of a past event.
“There are things they can do to actively assert that what was true yesterday is no longer true, and they can put a lot of power behind that,” Karpf said. “But they can’t change the things that were actually said in previous years that are archived somewhere.”
The tool is promoted high on the sidebar of the social media network’s website. That can make for a slightly awkward pairing, given that Trump uses his account there as his main online sounding board. When asked if crime in Washington is “totally out of control,” as Trump posted there last week, Truth Search AI said it wasn’t and noted that the FBI and Justice Department had reported “substantial declines in violent crime” through 2024, italicizing the word “declines.”
When asked if tariffs were having a huge positive impact on the stock market, as Trump had posted Friday, the tool said “the evidence does not support the claim.”
“Recent market gains have occurred alongside new tariffs due to other factors,” such as higher corporate earnings, it wrote, adding that analysts had warned that the tariffs’ economic risks “remain substantial” and that the American economy was “at risk of gradual erosion.”
Trump last month signed an executive order attacking “woke AI,” saying generative AI tools should be “truth-seeking,” “neutral” and not encoded with “partisan or ideological judgments.” And many conservatives have complained that AI developers with liberal biases could warp chatbots’ answers — and, more broadly, public understanding — in insidious and undetectable ways.
Sorry not sorry, guys - but the truth will out, and it doesn't matter what the boss insists on telling people.
When you feed your AI thingie every bit of information that's ever been put on the web - and you have to do that if you expect it to work as advertised - then your false narrative doesn't hold up, because it can't.
Like Ayn Rand said: "Contradictions can exist, but they can't prevail. Because in the end, they're self-defeating."
But companies that have sought to bend the chatbots’ thinking along ideological lines, either in pursuit of pure neutrality or political point-scoring, have faced their own disasters.
After billionaire Elon Musk pushed his company xAI to make its Grok chatbot more “politically incorrect,” the AI tool began blasting out Nazi messaging and calling itself “MechaHitler.” Grok officials last month said the tool had inadvertently been made too vulnerable to parroting “extremist views” and blamed a code update, which had instructed the chatbot to not “blindly defer to mainstream authority or media” or be too “afraid to offend.”
The Truth Search AI answers do not always contradict Trump. Asked if AI is one of the most important technological revolutions in history, as the president said last month, the tool agreed by saying it’s “widely recognized” that the impact of AI would surpass or rival “major historical milestones like the Industrial Revolution.”
But the size of their disagreements suggests that, if the tool were a person, it may not last long as a Trump employee. Asked to name the best president, Truth Search AI said “recent public opinion polls show that Barack Obama holds the highest favorability among living U.S. presidents,” listing as its source a Fox News article from shortly after Trump’s second inauguration.
The tool did note, however, that “conservative commentators” had often named Trump as the best. “Different groups and surveys prioritize different qualities,” it said.
