One YouTuber has taken AI analysis to an extreme level after feeding every one of Donald Trump’s 32,754 Truth Social posts into an artificial intelligence model to see what it could reveal about the president’s personality, behavior, and mindset. The experiment was carried out by the YouTube channel I Ask AI in a video titled: I Fed ALL of Trump’s Truth Social Posts Into AI. Here’s What It Found. According to the creator, the goal was to have Anthropic’s Claude analyze Trump’s entire posting history using only the content of those posts, without any outside information about his life, career, or political background.
The creator described the project as a way to uncover “the man behind them, his mentality, his behavior, his personality based only on those posts,” before asking the AI to produce a detailed breakdown of patterns, language, themes, and posting habits.
AI says self-focus was the strongest pattern
After processing the posts, Claude identified what it described as the dominant theme running throughout the archive.
“Here’s my honest take based only on what’s inside this data. The single loudest signal is self-focus. Roughly 1 in 11 posts exist purely to praise himself. And Donald Trump/Trump is by far the most frequent positive target in the entire data set. More than 2700 times, dwarfing America, the United States, or any policy or ally.”
The AI continued: “He is the protagonist of nearly everything he writes. Even posts about other people, events, or the country tend to route back to him. His crowds, his numbers, his vindication, his persecution. This is a person whose internal world is organized around himself as the central character.”
Claude also pointed to what it called a highly adversarial worldview, noting that accusations represented the most common rhetorical style across the dataset.
“The second signal is a relentless adversarial frame. Accusation is the dominant rhetorical mode. Over 6,300 posts, and the negative target list reads like an enemy’s ledger.”
Conflict, grievance and victory dominate the analysis
According to the AI, Trump’s posts frequently framed political and legal challenges as personal attacks.
“He casts himself overwhelmingly as a victim. the legal topics, the witch hunt and rigged and/or stolen language. The framing of every indictment is persecution.”
Claude argued that this existed alongside a constant emphasis on success and victory.
“Notably, this victim framing coexists with constant boasting about winning, dominating, record numbers. That combination, simultaneously the most persecuted and the most triumphant person alive, is psychologically telling. Grievance and grandiosity run on the same track.”
The analysis also highlighted stylistic traits, including heavy use of capital letters, exclamation points, and emotionally charged language.
“There’s almost no register of nuance, doubt, or measured reflection anywhere in 32,000 posts.”
The dashboard broke down thousands of posting habits
Beyond the written assessment, the creator also asked Claude to generate an interactive dashboard summarizing the findings.
The dashboard reported that 32,754 posts had been analyzed, with 8.8% classified as self-praise. It also found that 16.5% of posts were heavily capitalized and that 26.2% were published between midnight and 6:00AM.
Other figures showed “Accusation” as the most common category with 6,262 posts, followed by “Praise” at 5,818 and “Boast” at 2,624.
Summarizing its findings, Claude wrote: “The main character. 2,867 posts praise himself directly, and his own name leads the entire word count.”
It added: “Even posts about the country, the economy, or his enemies tend to route back to him as the hero, the victim, or the winner.”
The AI concluded that Trump’s posting history presented a consistent picture of someone who viewed the world through loyalty, conflict, and personal validation, while stressing elsewhere in the video that its conclusions were based solely on the Truth Social posts supplied to it and not on any external information.
In a recent video uploaded to the “I Ask AI” YouTube channel, a fascinating and bizarre moment occurred when the user decided to ask Donald Trump’s AI on Truth Social for its prediction on the 2026 midterm elections.
The video, titled I Asked 4 AIs to PREDICT the 2026 Midterms. Trump’s AI Was INSANE, was posted on Monday, March 23. The user was unsure of what to expect from the AI but prepared for what seemed like an unpredictable answer.
What followed, however, left viewers both puzzled and entertained. The AI’s response was a chaotic jumble of predictions, figures, and terminology that did little to resemble a coherent political analysis. The user expressed disbelief over the answer, calling it “schizophrenic” and warning that it might sound like “a lot of word salad.”
The AI’s bizarre predictions
The response began by predicting that Republicans would hold both the House and Senate in the 2026 midterms. The AI suggested a House result of 225 to 210 and a Senate outcome of 54 to 46, despite historical patterns that show a tendency for the president’s party to lose seats during midterms.
It went on to mention various factors such as “68 Trump EOS codified” and “413 bills passed” as evidence of success. However, the explanation grew increasingly difficult to follow.
The AI also forecasted that the GOP would strengthen its grip on certain states, with predictions of Republican victories in Michigan, Minnesota, and New Hampshire, while even describing a chaotic situation in Texas that “favored Cornin.”
There were also mentions of various politicians, including Tim Scott, who was reportedly calling 2026 “the year of affordability.”
The response concluded with an even more puzzling claim: that “Trump’s secret weapon poll stabilize as inflation cools.”
AI expert reacts to the madness
After sharing the AI’s prediction, the user decided to feed it into another AI, Claude, for a second opinion. Claude’s reaction was scathing. It described the response as resembling a Republican fundraising email “that accidentally got fed through an AI.” Claude went on to critique the predictions, stating, “this isn’t a prediction. It’s fanfiction with footnotes.”
Claude pointed out that the AI’s suggestion of a president’s party gaining seats in the House was incredibly rare, with only two instances since World War II, in 1998 and 2002. The AI’s justification, such as the claim that 413 bills had been passed, was mocked, with Claude asking, “Since when does passing bills make voters love you?”
In short, the reaction painted the AI’s prediction as more of a fantasy than a realistic forecast. It seemed that Trump’s AI had offered more wishful thinking than grounded political analysis.
