Online examination is widely used in distance education. Automated test paper is the key technology of online examination. The problem of generating test paper is the solution of multi-objective expected value, and it often has multiple solutions. For solving multi-objective function, the advantage of artificial intelligence algorithm is more and more obvious. Among them, the multi-objective optimization of genetic algorithm and ant colony optimization is more efficient, and can be more competent for the automatic test paper generation of the database technology curriculum. The application of artificial intelligence algorithm in test paper generation is discussed. The index system of test paper generation is constructed, and a mathematical model of multi-objective constraint is established, and the multi-objective expectation is optimized. The experiments results demonstrate that artificial intelligence algorithm has the highest success rate, with an average of more than 98% (including 100% of ant colony optimization, 96% of genetic algorithm), while those other than the artificial intelligence algorithm have low success rate, with the random variables 62%, backtracking method 84%. The application of artificial intelligence method, especially genetic algorithm and ant colony optimization, improves the efficiency of automated test paper generation. It meets the needs of various actual test paper generation, and makes online examination very well applied.