If you are serevd a random pancake in a random orientation, then each of the burnt sides occurs with equal probability as top side. But in two out of these three cases, the other side is burnt as well, so the answer is $\frac 23$.
To argue with Bayes, note that $P(\text{burnt-top}|A)=0$, $P(\text{burnt-top}|B)=\frac12$, $P(\text{burnt-top}|C)=1$ where $A,B,C$ are the three pancakes. If $P(A)=P(B)=P(C)=\frac13$, you obtain $P(B|\text{burnt-top})=\frac13 $ and $P(C|\text{burnt-top})=\frac23 $ from Bayes: $$ P(B|\text{burnt-top})=\frac{P(\text{burnt-top}|B)P(B)}{P(\text{burnt-top})}=\frac{P(\text{burnt-top}|B)P(B)}{(P(\text{burnt-top}|A)P(A)+P(\text{burnt-top}|B)P(B)+P(\text{burnt-top}|C)P(C))}=\frac{\frac12\cdot\frac13}{0+\frac16+\frac13}.$$
Since (given a burnt top) the distinction burnt vs. non-burnt bottom is precisely that between $C$ and $B$, the answer to the original question is - again - $\frac23$.