OpenAI’s latest benchmark, GDPval, shows that its ChatGPT models are now performing at levels comparable to human experts across 44 professions. The evaluation tested real-world tasks in nine major industries crucial to the U.S. economy, with the benchmark designed by professionals averaging 14 years of experience.
OpenAI’s newest model, GPT-5, recorded a 38.8% win and tie rate against human experts. In comparison, Anthropic’s Claude Opus 4.1 achieved a higher 47.6% rate, highlighting growing competition in AI for workplace applications. Experts note, however, that real-world jobs often involve more than the tasks assessed, and many AI pilot projects still struggle to produce measurable revenue gains due to low-quality output.
In a separate announcement, OpenAI launched Instant Checkout, allowing U.S. users to buy products from Etsy and Shopify stores directly within ChatGPT. Powered by the open-source Agentic Commerce Protocol developed with Stripe, the feature supports payments via credit cards and digital wallets. The launch boosted Etsy and Shopify shares, reflecting market confidence in AI-driven e-commerce.
Meanwhile, Microsoft and Anthropic are expanding their offerings with autonomous AI agents capable of handling more complex professional tasks. Analysts say these developments signal a shift toward AI becoming increasingly integrated into both work and commerce, though limitations remain.
Overall, the advancements demonstrate AI’s growing capabilities in professional and commercial settings, reshaping industries while still facing significant practical and economic challenges.
Follow on Google News
