We now understand a LOT of physics and chemistry. That's because it's observational, and because it's observational you can make predictions, perform experiments and get results. Other people can do the same as well, thus enabling not only peer review but actual verifiability.
This is why we understand physics and chemistry extremely well, biology OK but rapidly improving, and economics and public policy much less well. Because we can reproduce results in physics everyone either agrees on the truth or is generally regarded as a crackpot. Biologists can't produce the kinds of extremely clean data physicists can but it's still meaningful. In macroeconomics there's very little ability to perform experiments in a repeatable fashion, which results in everyone observing the world and then arguing about why things happened and nobody can prove that they're right or that others are wrong.
When you're dealing with math or computer science induction is a great tool that allows you to do pretty incredible things. But it does so because the nature of math or algorithms is stationary; the rules are 100% fixed from the beginning of time 'til the end (or so you can assume if you're building the system) and thus any assumptions are by definition correct. It's much harder to do this when you're dealing with things that can change over time (and thus potentially invalidate your assumptions), which is basically everything else. That's not to say that induction is worthless outside of math and computer science but it's use comes with more caveats.
I totally get why serious scientists don't want to even acknowledge that a creationist might raise a good point but once it's made the damage is done. If you don't respond at all they pile on saying "look he doesn't have an answer for X!" and if you do respond with a reasonable statement about "this is the best we know thus far and of course nothing can be known until we observe it directly" then they'll pile on with "see it's not guaranteed!" Both of those outcomes are extremely frustrating when you're dealing with someone who can't be convinced with any amount of evidence.
But failing to acknowledge that there is a difference between direct observation and historical/induction and that they create two classes of certainty which are not identical does the very tedious and difficult work of science a disservice in my opinion.
EDIT: To clarify I think what I'm mostly talking about is certainty. If you can't directly (or indirectly) observe something happening with some kind of experiment you can have a personal opinion about certainty and someone else can have another but neither one can be provably wrong. Once that happens there's real certainty and very little room for opinion anymore. Until then things are a bit more fluid. A well educated person's estimate of certainty might be much more accurate than a plebe's but it's not guaranteed to be right.