"It's too late to correct it," said the Red Queen: "when you've once said a thing, that fixes it, and you must take the consequences."
There are a host of other problems. Recall that the IAT rests on the assumption that more difficult mental operations—such as resolving conflicts in attitudes about others—take more time. But there are other reasons that a mental operation can take more time. One is unfamiliarity—we respond more slowly to words and images that are unfamiliar because they require more retrieval and processing time. Another is the confusion born of linguistic and conceptual associations. The word "old" is often used as a synonym for "worn out," "spent" or "obsolete" when we're talking about technological products like computers, cars or sewing machines. When prompted by that word while taking an IAT, our brains automatically call up these associations; despite what the authors would have you believe, the fact that we have such associations doesn't mean we hate or devalue our grandmothers.
Another confounding factor is that the brain is designed to detect patterns of co-occurrence and responds to learned associations based on a lifetime of hearing word pairings. If I hear the word "bread," the first word that comes to mind might be "butter," even if I never eat butter, never buy it and for that matter don't even eat bread. But associations aren't the same as biases. My quickness in conjuring one word when hearing another says nothing about an "implicit bias." It says even less about how I would treat another individual. Common sense would tell you this.
Such linguistic associations occur at a shallow, superficial level of processing and reflect only statistical patterns in the language, not held opinions; and not everyone who knows the stereotypes about race, age or gender believes them.
Attempting to find moral significance or straightforward causality in word choice sounds like the concept of a Freudian slip, rebranded for our modern age. Honestly, the more I learn about language — and I'm hardly any sort of scholar — the more deeply amazed I am that such flimsy, abstract signs allow us to understand as much as we do and communicate at all.