Spiros found this remarkable 1-hour talk by Paul Dirac in New Zealand in 1975:
Recall that Dirac was born in 1902 and died 1984; so he was 73 in the video above. He is a serious man but there's a lot of room for his memories as well as some jokes.
There are actually four talks. The first one above is on quantum mechanics; the second one is on quantum electrodynamics; the third is on the magnetic monopoles; and the fourth one is on his largely incorrect large number hypothesis (variation of universal constants).
Even if you focus on the first lecture on quantum mechanics, there's a lot of interesting stuff to listen to. He was impressed by the successes of the old Bohr model of the atom. He identified the key property of Heisenberg's quantum mechanics to be the non-commutation of the observables. He thinks that Heisenberg himself was afraid of this feature – much like almost every originator of a great idea is afraid of the key revolutionary relationship because it could also very well spell the doom for his pet idea. But Dirac himself wasn't afraid. He took it seriously.
At the end, he also mentions that he doesn't believe renormalization unless the quantities are renormalized by small amounts. He explains the success of the renormalized quantum field theories by comparing them with the successful old Bohr model of the atom which was fundamentally wrong, too. Well, Dirac was both right and wrong in his counter-revolutionary attitude. He was wrong because the success of renormalization is no accident, of course. He was right because it's possible to reorganize the field theory in such a way – a way using counterterms – that the renormalization is small, indeed. However, you must still accept that there are counterterms that cancel the loop divergences. I am not sure whether Dirac understood the Renormalization Group and similar insights of the 1970. He may have already been too old for that.