"I'll play diplomat and say I think what Elon meant to say is that Mark's view of this is optimistic. I do tend to take the more pessimistic view that it will be hard to keep general AI in a box," said Costolo on CNBC's "Squawk Alley."
"If you hypothesize that we could create intelligence greater than ours, it's pretty easy to jump from there to think that intelligence would be able to figure its way out of any sort of logical box you try to put it in," he added. "Mark tends to be much more of an optimist on the future of technology."
During a Facebook Live broadcast on Sunday, Zuckerberg said that doomsday scenarios about AI are "pretty irresponsible."
A Twitter user posted an article about the Facebook chief's comments. Musk, responding to the tweet, said that he has already spoken to Zuckerberg about this and that the Facebook CEO's "understanding of the subject is limited."
The debate came amid talks about how long it will be before computers become smarter than humans — if that ever happens.
Costolo also acknowledged that any impactful technology in the past has had some sort of downside.