Part 5/9:
While some researchers dismiss language models as mere prediction machines, recent evidence challenges this view. For example, a astrophysicist shared that GPT-5 Pro independently rediscussed scientific results in under 30 minutes, arriving at solutions comparable to those of human researchers. This suggests that GPT-5, and likely GPT-6, are starting to generate genuine new knowledge—an advancement from "AI summarizing science" to "AI doing science." Such capabilities could accelerate innovation dramatically.
AI's Expanding Functionality: Beyond Text
New Features on the Horizon
Trademark filings and rumors reveal several anticipated features for GPT-6: