Others have pointed out Hume's "you can't get an ought from an is." In other words, descriptive facts do not provide prescriptive rules.
However, Kant did set out a response to this with his categorical imperative, and Einsteins' answer, while correct in one sense, may contain seeds of its refutation. The scientific urge to seek causes in nature and the "scientific method" do model a kind of morality.
First, human beings are implicitly unified in contrast to "nature." To cooperatively seek causes in "nature" is to not seek them in nefarious human or divine acts. Science actively directs us away from human guilt, blame, sins, and rivalry. So there is a kind of historical negation of negation.
Second, science is, above all, a method of conditional agreement, so that, as Popper said, "our hypotheses can die for us." Or, he might have added, instead of others.
Finally, in the Kantian CI sense we might follow science to rationally derive the universal rules necessary for deriving universal rules. Habermas has, I believe, attempted to translate this into a concept of language and communication. If we are to communicate as social beings at all, certain rules are necessary.
So, while science cannot "discover" morals out there in nature, it may itself be abstracted into a model of what is required of rational, moral, communicative beings.
While this veers away from Einstein, utilitarians might also argue that sociobiology, cognitive science, game theory, and other areas may offer moral insights once the most basic optimal outcomes are postulated, not that I buy it.