AITask choice and the divide over AI's capabilities
One post argues that AI has made only modest progress on “wordcel-type” tasks but gotten much better at “shape-rotator” tasks—a partial explanation for differing views of its capabilities.
TLDR
A user suggests that the tasks people use AI for partly explain their reactions to it. In their view, AI remains “very mid” at “wordcel-type” tasks while having improved substantially at “shape-rotator” tasks. They argue that using AI only for the former makes it easier to ask, “what's the hype about?”
Combined views
142.7K
2 Sources, first seen 22h ago
Task choice and the divide over AI's capabilities
One post argues that AI has made only modest progress on “wordcel-type” tasks but gotten much better at “shape-rotator” tasks—a partial explanation for differing views of its capabilities.
TLDR
A user suggests that the tasks people use AI for partly explain their reactions to it. In their view, AI remains “very mid” at “wordcel-type” tasks while having improved substantially at “shape-rotator” tasks. They argue that using AI only for the former makes it easier to ask, “what's the hype about?”