DBMS don't do this by default because indexes aren't free. Each additional index means the database has to write to an additional location for (potentially) eve

DBMS don't do this by default because indexes aren't free. Each additional index means the database has to write to an additional location for (potentially) every insert/update. The feature would need to also keep track of which indexes are unused and remove them over time to avoid the massive write amplification that would come from building a new index for every query that crossed the threshold over time.