- TalentGenius
- Posts
- 136 and climbing: when AI took the IQ test
136 and climbing: when AI took the IQ test
AI has brought a version of Moore’s law to every company and every individual.
Malcolm Frank, CEO
May 18, 2026
136 and climbing: when AI took the IQ test
Somewhere in a quiet corner of the internet last week, a website went live that gave OpenAI's GPT-5.5 an IQ of 136. Anthropic's Opus 4.7 came in at 132. Google's Gemini 3.1 Pro, 131. The site, aiiq.org, plotted more than fifty frontier models on a human bell curve and, in doing so, handed every CIO, board chair, and curious uncle the one thing they had been quietly craving: a single number.
Is it a tidy number? Yes. But also a misleading one? Maybe. Researchers at Anthropic and Microsoft have shown that the error bars on these leaderboards are wide enough to reorder the rankings. (Cynics posit “of course they did” as their scores were slightly lower.) But these arguments mask the larger point, that AI’s “IQ” is steadily increasing. For example, GPT-4-turbo registered an estimated 75 in late ‘23; thirty months later, the leaders are flirting with genius level. The trajectory is real.
This progression opens a much bigger question: Where are we on the path to superintelligence? An IQ of 136 puts today's best models in the neighborhood of a sharp MIT undergrad (which feels about right when you chat with Claude or ChatGPT).
Now extrapolate. What happens at 200? At 500? Tim Urban, in his seminal Wait But Why essay, framed it this way: trying to grasp the output of ASI will be like a chimpanzee trying to grasp how humans built a skyscraper. The chimp sees the building, but can't begin to contemplate the engineering, the financing, the steel. He can't formulate the questions. In fact, he probably doesn't even register it as something that was built.
So, are we racing to superintelligence in the next two years? Maybe. But the more useful question is “What to do now?” Moore's Law taught us one thing above all: the vendors who stepped off the curve became irrelevant. AI has brought a version of that law to every company and every individual. You have to get on the curve, regardless of where it's taking us.
Don’t miss out! [WEBINAR] Govern AI or Be Governed By It: What Every Board Director Needs to Know Now
A board-level briefing with the Corporate Board Member network on managing AI risk, oversight, and competitive impact with confidence.
When: Tuesday, 19 May, 1pm - 2pm EDT
What You’ll Learn
How to govern AI in the enterprise without falling into risk, misuse, or legal blind spots
What proactive, effective AI oversight looks like and why fear-based approaches fail
How to track AI-driven market shifts, competitor moves, and valuation impact at the board level
Reply