As humans, we're naturally inclined to believe that nothing, not even our clever creations, can surpass our intelligence, at least in critical areas. After all, we've already ceded superiority in raw speed, calculation, and visual acuity to machines. Admitting inferiority in intelligence will force sweeping changes in every realm of life for which we are least ready. So, while that day of reckoning seems inevitable, we keep hoping it's not imminent with the desire to disregard any contrary evidence.
This denial drives many to nitpick AI's flaws, like a broken record. Surely, deficiencies exist in generative AI's processes and outputs. However, those fixated on imperfections or brandishing AI as a statistical parrot overlook its eerie ability to conjure entirely novel things.
As we have written before (https://bit.ly/43zITJV), GenAI's prowess at the theory of mind tests offered an early clue that our decades-long search for the path to creating intelligence might have been over. A recent study (https://bit.ly/3QhY9FS) highlights AI's budding talent for systematic compositionality, another marker of advanced intelligence.
Without getting twisted by these studies, one can only look at the machines' new-fangled abilities to generate entirely new things out of thin air, from images to synthetic proteins to scientific hypotheses. The jury is out on the quality of synthetic creations, even if they will only get exponentially better.
To illustrate AI's mimicking of human imagination, I asked ChatGPT to ponder the language underlying the universe's governing laws, assuming it's not man-made math. I requested something no human has conceived before. Its response below may be technobabble, but it's impressively abstruse.
One needs to be cautious on many fronts, including in optimism over the abilities. Equally important, however, is for people in all walks of life to begin thinking about what all they might want to do if they have access to machines that can think at a level of complexity higher than what best humans have.
