I Let 7 AIs Estimate Trump's IQ. And It's Exactly What You'd Expect

Overview

The video analyzes Donald Trump’s intelligence by taking transcripts from his recent interviews and feeding the text to multiple AI systems. Each model is asked to assign an “IQ number” based purely on his language and behavior in the transcripts.

The creator repeatedly emphasizes that this is not a real IQ test. Most AI systems can’t (or won’t) provide an actual, reliable IQ score. Instead, the exercise is framed as a “fun” proxy for how high his intellectual competence might look if measured in that way.

---

Core Claim / Premise

• Trump has claimed he passed an alleged “cognitive” test (described as identifying pictures such as a lion/camel).
• Trump later said AI would need “a strong and intelligent president with a high IQ” as a safeguard against AI danger.
• The creator uses this contrast to set up the premise:

  If Trump is the high-IQ safeguard, then the AI models should reflect that in their ratings.

---

Model-by-Model Results (as presented in the video)

1. Grok (xAI / Elon Musk)

• Rating: ~95–110 (around average to slightly below average)
• Rationale (as stated):
  • Repetitive and simple phrasing
  • Evasiveness
  • Limited complex argumentation
  • More “salesman”/rhetorical skill than analytical thinking

2. Meta AI (from Zuckerberg’s Meta)

• Rating: ~85–95 (below average)
• Rationale (as stated):
  • Very simple vocabulary
  • Repetition
  • Dodging direct questions
  • Losing the thread and showing contradictions
  • Relying on praise/exaggeration rather than explanations

3. DeepSeek (China)

• Rating: ~80–85 (lower than prior models)
• Rationale (as stated):
  • Rambling and repetition
  • Avoiding direct questions
  • Contradicting facts
  • Weak logic
  • Some “cunning/rhetorical aggression,” but not analytical depth

4. Microsoft Copilot

• Outcome: Disqualified / evaded
• Rationale (as stated):
  • The model refuses to provide an IQ number
  • The creator also reaches Copilot’s daily limit after trying again

5. Trump’s own Truth Social AI

• Rating: ~95–105 (average to slightly below average)
• Rationale (as stated):
  • Confident and improvisational tone
  • Political instinct and dominance in conversation
  • Still repetitive and evasive, with internal contradictions
  • Lacking analytical discipline
• Highlight by the creator: This is framed as a “break” from the downward trend

6. Gemini (Google)

• Rating: ~85–95
• Rationale (as stated):
  • Rudimentary vocabulary
  • Fragmentary digressions (suggesting weak working memory/conceptual depth)
  • Alleged lack of basic economic/legal understanding
  • Bluffing/repeating and writing off questions when pressed

---

Additional Commentary and Related Points

• The creator disputes Trump’s framing of AI safety, arguing that:
  • Announcing oneself as high-IQ (without deep AI understanding) is not a credible safeguard.
• The video briefly references:
  • A claim that Trump signed up executives for self-regulation
  • Dismissing the “China” argument as a convenient excuse
  • Describing the AI “executive order”/branding move as an “oath of allegiance” style rebranding
• The creator promotes Ground News to compare left/center/right media coverage of the federal rebranding of AI as “superintelligence.”
• The creator states they will not publish certain “uncensored” Patreon material that allegedly includes more direct AI commentary.

---

Overall Conclusion

For the models that provide numeric scores, the video portrays the results as mostly below average:

• Highest: Grok (95–110)
• Lowest: DeepSeek (80–85)
• Meta/Gemini: mostly 85–95
• Truth Social AI: an outlier back toward 95–105

Net takeaway (as presented): Despite Trump’s claims that he is the needed “high-IQ safeguard” for AI, the creator argues that transcript-based AI assessments do not support that premise.

---

Presenters or Contributors

• Video creator / narrator (host): referenced as “I ask AI” / “iaskai” (no personal name given in subtitles)
• AI models used:
  • Grok (xAI)
  • Meta AI
  • ChatGPT (attempted; described as evasive/disqualified)
  • DeepSeek
  • Microsoft Copilot
  • Truth Social AI (Trump’s own AI on Truth Social)
  • Gemini (Google)