GPT-5 launched a few weeks ago amid MUCH hype from Sam Altman and OpenAI. The release comes as Altman has been suggesting since December that AGI — a form of AI that can handle most knowledge worker tasks — was right around the corner.
"We are now confident we know how to build AGI as we have traditionally understood it." —Sam Altman, January 5th, 2025.
So is this the case? Has OpenAI really created an LLM that can augment significant portions of your work and mine?
No. Not yet.
It's notably faster for most queries and more accurate on most benchmarks. It hallucinates less often. These are all critical improvements for using AI in business, but we're not remotely close to taking humans out of the loop.
The one area where I've seen notable improvement is in the GPT-5 Pro model, which effectively replaces the o3 Pro model—the previous top tier. I don't have benchmarks, but GPT-5 Pro seems slightly slower. However, for every topic I've had it research or discuss, it's delivered very high-quality answers.
OpenAI claims it has PhD-level intelligence. I'm not sure how to evaluate that claim, and I'm skeptical. But it will undoubtedly unlock research capabilities for many professionals that previously would have required anywhere from a couple of hours to several weeks of human work, depending on the topic and scope.
Which brings us full circle. While it definitely wasn't worth the hype, it's a very capable release. Had OpenAI simply not set the bar so high, this would have been seen as a successful launch.
If you aren't figuring out how to use AI to add quality, services, or efficiency, you're falling behind. Not in one giant leap, but one incremental release at a time.
