In the three years since the public launch of ChatGPT, we have witnessed a cognitive leap that would normally take a biological species millions of years. We have moved from AI that could “mimic” a graduate to AI that systematically outperforms PhD-level experts in complex reasoning.
Tracking the IQ Jump (2022–2025)
The progression from GPT-3.5 to the “Reasoning Models” of late 2025 (such as the o1 and o3 series) reveals a startling acceleration. Using benchmarks like the Wechsler Adult Intelligence Scale (WAIS), we can track this ascent:
| Model Generation | Approximate Timeline | Estimated Verbal IQ | Human Equivalence |
| GPT-3.5 | Late 2022 | ~115–125 | Above Average / University Student |
| GPT-4 | Early 2023 | ~155 | Genius Level (99.9th percentile) |
| Frontier (o1/o3 Class) | Late 2025 | 180–195+ | Beyond Human Benchmarks |
By December 2025, frontier models have moved beyond simple pattern matching. With “Chain-of-Thought” processing, these systems can now spot their own logical errors and adjust strategies mid-task. In recent GPQA (Graduate-level Physics/Science) benchmarks, these models scored 78%, significantly higher than the 65% average achieved by human PhDs in the same subjects.
The ‘Cognitive Debt’ Warning
While the AI’s IQ is soaring, human cognitive engagement is showing signs of “atrophy.” A landmark June 2025 study from the MIT Media Lab titled Your Brain on ChatGPT used EEG monitoring to track brain activity during essay writing.
- Connectivity Collapse: Participants using AI showed the “weakest overall coupling” in brain regions associated with memory and semantic processing.
- The ‘LLM-to-Brain’ Hangover: Perhaps most worrying, users who relied on AI for their first three tasks showed significantly lower neural connectivity even when they were asked to work without AI in the fourth session.
- Memory Ownership: Users struggled to recall or quote their “own” work when it was generated by an LLM. This suggests that outsourcing thought bypasses the neural “encoding” required for true learning.
The 2025 UK Job Market: A Junior Role Crisis
The IQ explosion has hit the UK white-collar sector with the force of an “economic earthquake.” Because AI can now perform the work of an associate or junior researcher with superhuman precision, the “bottom rung” of the career ladder is disappearing.
Graduate Contraction: According to Indeed UK, 2025 has been the toughest market for graduates since 2018. Big Four accountancy firms have reportedly cut graduate recruitment by up to 29% as AI takes over routine data work.
The Junior Gap: A study by King’s College London (October 2025) found that while total employment in AI-exposed firms fell by 4.5%, the losses were concentrated almost entirely in junior positions (down 5.8%).
Senior Wage Premium: Conversely, senior roles have seen a pay increase of over £1,300 as firms pivot toward “Senior-Heavy” teams where one human “Director” manages a fleet of “AI Agents.”
The New Professional Value
As we head into 2026, the definition of “Intelligence” is changing. Raw analytical output (the ability to process data, write code, or draft reports) is now a cheap, infinite commodity.
Your value as a professional has shifted from Production to Sanction. The world no longer needs you to generate the 180-IQ answer; it needs you to have the Judgment and Moral Accountability to decide if that answer is the right one for the client.
