OpenAI just marked 10 years with big news: The release of GPT-5.2, a model designed to master knowledge work, and a bold prediction from CEO Sam Altman that superintelligence is now practically inevitable in the next decade.
The new model, which follows an internal “Code Red” directive to accelerate development, introduces significant improvements to make it better at executing complex, real-world tasks from start to finish than previous models.
This release signals OpenAi is moving from abstract IQ scores to metrics that track how well AI performs real-world jobs.
To discuss GPT-5.2’s potential impact and OpenAI’s first influential decade in business, I talked it through with SmarterX and Marketing AI Institute founder and CEO Paul Roetzer on Episode 186 of The Artificial Intelligence Show.
Intelligence Beyond IQ
Early on, the AI industry measured progress using benchmarks that looked a lot like standardized tests, essentially IQ tests for machines. But according to Roetzer, we have hit a ceiling with those metrics.
“The tests of IQ are basically saturated,” says Roetzer. “When you’re trying to evaluate these models against standardized tests that a human might take, the AI is already there.
It’s at basically the top human level, if not beyond top human level at a lot of these tasks. So it’s really hard for all of us to feel the difference when we’re talking about increases in IQ points.”
Here Comes GPT-5.2, Adding Value to GDP
With the release of GPT-5.2, OpenAI is leaning heavily into a new benchmark that evaluates the model’s ability to perform 1,300 specific tasks from 44 occupations that contribute to Gross Domestic Product. The result is real-world deliverables, such as creating legal briefs, engineering blueprints, and nursing plans.
The results are eye-opening: OpenAI reports that GPT-5.2 Thinking achieved a win rate of roughly 71% against human experts in direct comparisons.
“We need to start measuring against real work because that’s what we’re going to know when economic disruption is around the corner,” says Roetzer. “And I would say we’re there.”
Faster, Cheaper, “Better” Than Humans?
The implications of this GDPval, as the benchmark is called, go beyond quality. OpenAI’s data suggests that frontier models can complete these professional tasks roughly 100 times faster and 100 times cheaper than human experts in these fields.
In their announcement, OpenAI described this as a tool to “unlock even more economic value for people,” emphasizing how the model can help with spreadsheets, presentations, and coding.
“Notice they don’t say we want to replace more jobs,” says Roetzer.
The subtext regarding human job displacement, however, is hard to ignore.
A Changed Mission
OpenAI released GPT-5.2 during its 10th anniversary, inviting a look back at the company’s origins.
Founded in December 2015 as a non-profit, its original mission was to advance digital intelligence to benefit humanity, without a need to profit. The original announcement vowed to distribute power evenly and share patents with the world.
Compare that vision to now, and the contrast is stark.
“Nothing in that paragraph is true anymore,” says Roetzer, referring to the founding statement.
Today, OpenAI is a dominant commercial force, racing against global competitors, including Google, to achieve Artificial General Intelligence (AGI). In a commemorative post, Sam Altman reflected on the journey, saying the company is now “almost certain to build superintelligence in the next 10 years.”
OpenAI’s evolution from an idealistic non-profit research lab to a powerhouse driving the global AI arms race is one of the most significant business stories of not only the past decade, but of all time.
