It is well known that the best chess engines are now far stronger players than the strongest human players. But is it known what the smallest handicap is that makes them roughly equal?
No scientific studies on this that I know of, but chess.com has hosted some online matches between super GMs (like Nakamura) and top engines in the world (esp. Komodo if I recall correctly).
In these matches, the human GM is given one of the following odds from the starting position:
An extra exchange (so rook for knight).
The engine, playing Black, is missing its f7-pawn.
White gets two extra moves on top of his additional right to go first (I think).
Each of these are approximately +2 for White. So it seems that at the top level for humans and engines, a +2 advantage is needed to ensure rough equality.
Pretty humbling if you think about it. Wait until they perfect machine learning with engines...