The “AGI debate” misses the point: AI will be better than the human alternative long before it's good or free from errors and imperfections 💡
tl;dr;
OpenAI just released #GDPval, an evaluation measuring AI performance on real economic tasks across 44 occupations. The results? AI models are approaching human expert quality while being 100x faster and cheaper. 😮
https://lnkd.in/e4sw5Eu7
GDPval spans 44 occupations selected from the top 9 industries contributing to U.S. GDP. The GDPval full set includes 1,320 specialized tasks , each meticulously crafted and vetted by experienced professionals with over 14 years of experience on average from these fields. Every task is based on real work products, such as a legal brief, an engineering blueprint, a customer support conversation, or a nursing care plan.
Key highlights:
Quality: Leading AI models (GPT-4o, o4-mini, OpenAI o3, GPT-5, #Claude Opus 4.1, #Gemini 2.5 Pro, and #Grok 4) are now approaching the quality of work produced by industry professionals. In blind evaluations across 220 tasks, Claude Opus 4.1 performed best overall, particularly excelling in aesthetics like document formatting and slide layout, while GPT-5 stood out for accuracy in finding domain-specific knowledge.
Progress: There's been significant improvement over time - performance more than doubled from GPT-4o (spring 2024) to GPT-5 (summer 2025), following a linear upward trend.
Speed and Cost: AI models can complete these tasks roughly 100 times faster and 100 times cheaper than human experts, though this doesn't account for the human oversight, iteration, and integration work needed in real workplace scenarios.
Practical Implication: The researchers suggest that for tasks where models perform particularly well, trying AI first before human experts could save both time and money. The research indicates AI is rapidly closing the gap with human expert performance while offering substantial efficiency advantages.
#ai #transformation #disruption #change #agi