WTF roland, you are failing basic probability. If you eat 3 things with 4% probability, ofc the last one still has only 4% chance, but the total chance of beeing poisoned eating the 3 does increse.
Lets do it the easy way, using inverse probability which makes it easier when trying to calculate "at least 1" cases. The probability of beeing poisoned after 3 times with 4% each time is 1 - the probability of not beeing poisoned, so:
0,96*0,96*0,96=0,88 probability of not beeing poisoned.
1-0,88= 12% chance of beeing poisoned, close to what Vik said
EDIT: the result of 12% beeing the same as 4%+4%+4% is just a coincidence. You dont sum them up, but the fact several order of which one is the poisoned one are valid, plus the possibility of more than one, and so on adds to the probability. All that is simplilfied by using inverse probability like I did. Another way would be to calculate all possible outcomes (first one beeing poisoned +second one + first 2, and so on). And all that happens to be 12% same as 4*3, but its just a coincidence.
Still, its 12% like Vik said
And still more posts saying he is wrong xD