That's not entirely accurate.
In 1980 there definitely were some polls that showed Carter up late, like the infamous Gallup poll that had him up 8 with 2 weeks left. But by and large the polls showed Reagan with the momentum and almost all the polls at the end showed Reagan leading, including the final Gallup one.
Gallup Presidential Election Trial-Heat Trends, 1936-2004
In 2000, while I can't find a great list of public polls, what I can find shows that altogether they did very well, (with this source finding an error of barely a percentage point overall and that 7 of the polls included overstated Bush while just 2 overstated Gore).
Presidential Poll Performance 2000 | NCPP - National Council on Public Polls
In 2004, polls by and large showed Bush leading, with no polls outside the error range you'd expect from a bunch of polls done well.
RealClearPolitics - Election 2004 - General Election: Bush vs. Kerry
They also weren't really that off in 2016, but that's been pointed out to death in this thread already.
None of which says of course that polls in May are particularly predictive for what happens in November.