0
0 By Study Smart India, AI News Desk
September 25, 2025
In the annals of human history, few minds shine as brightly as Albert Einstein’s. The physicist, whose theories revolutionized our understanding of the universe, is often cited as a benchmark for genius. Yet, in 2025, artificial intelligence systems like ChatGPT are not just approaching but, in some metrics, surpassing the intellectual feats once thought exclusive to humanity’s greatest thinkers.
Einstein never sat for a formal IQ test, but experts have retroactively estimated his score based on his groundbreaking achievements and biographical details. Common figures place it between 160 and 180, with many settling around 160—a level denoting profound giftedness. 24 This estimation underscores Einstein’s exceptional ability in abstract reasoning and problem-solving, skills that have long defined human intelligence.
Fast forward to today, and AI models are being put through similar paces. OpenAI’s ChatGPT, particularly its advanced iterations, has demonstrated remarkable performance on human-designed IQ assessments. In early 2023 tests, the GPT-4 model achieved a verbal IQ equivalent of around 155, outperforming 99.9% of human test-takers in language-based tasks. 2 By 2025, newer versions like GPT-5 Pro have pushed boundaries further, scoring over 140 on rigorous offline IQ tests—entering “genius” territory and eclipsing average human scores of 90-110. 10 These results come from benchmarks like the Mensa Norway test, where AI’s logical pattern recognition and verbal acuity are evaluated without real-time aids.
But ChatGPT isn’t alone in this cognitive leap. Google’s Gemini series, including the 2.5 Pro model, has clocked in at 124-126 on similar evaluations, showcasing strengths in multimodal processing that blend text and visuals. 16 Anthropic’s Claude-4 Opus leads in some leaderboards with scores around 127, emphasizing ethical reasoning alongside raw intellect. 30 Even xAI’s Grok-4 has entered the fray, achieving an estimated 125 in comprehensive datasets, highlighting its unique, unrestricted approach to problem-solving. 11
These advancements stem from rapid iterations in large language models (LLMs), trained on vast datasets to mimic and exceed human cognitive patterns. For instance, OpenAI’s o3 model scored 133 on verbal reasoning tasks, surpassing most humans and approaching Einstein’s presumed level in specific domains. 34 Recent X discussions echo this progress, with users noting GPT-4.5’s offline score of 97—modest compared to its reasoning-enhanced siblings but still competitive against human averages. 38
Experts caution, however, that equating AI “IQ” to human intelligence is imperfect. Traditional IQ tests measure fluid reasoning, working memory, and processing speed—areas where AIs excel due to their computational power—but they overlook creativity, emotional depth, and real-world adaptability that defined Einstein’s work. 19 “AI benchmarks are snapshots of capability, not holistic intelligence,” notes one analysis, pointing out that vision-focused models like GPT-4o (Vision) score as low as 63, far below human norms, due to test limitations. 14
As AI continues to evolve, these milestones raise profound questions about the future of work, education, and society. If tools like ChatGPT and its peers can rival or exceed Einstein’s estimated prowess in targeted tests, what innovations lie ahead? For now, the race between silicon and synapses is accelerating, blurring the lines of what it means to be truly intelligent.
Average Rating