Deduction, like induction, is a method of reasoning. Induction makes an assumption that a set of theoretical constructs that have been experimentally verified in a given set of condition can be utilized further to apply to problems that have not yet been subject to experimental verification. This is generally a very reasonable assumption under many conditions.
Historically however, applying inductive argumentation to problems has on occasion lead to incorrect conclusions. For example, it was believed that transmutation of elements, similar to alchemy was impossible from the middle ages until the early twentieth century. But this is wrong, transmutation is possible. The error stems from inductive application that chemical processes do not result in transmutation, to all processes. However, nuclear processes are capable of transmutation. Another example is that it was believed that the Earth was flat. Certainty, if you stand and look around, it does appear that it is flat. However, this is because the radius of the Earth is so large, that one cannot see the curvature with one’s eye. So if one believes one should be able to see something to believe it and one cannot see the curvature, one can inductively conclude that the Earth is flat. But this again is wrong. Other examples: it was commonly believed that two objects with different masses would fall at different rates in a gravitational field. This is wrong, also hypothesized by Galileo and experimentally verified (to some extent) when Galileo dropped objects with different mass from the tower of Pisa. And there are a lot more historical examples for which inductive reasoning has gone awry.
So it is possible for inductive reasoning to go awry, particularly when currently known methods can not adequately answer the problem posed or adequately describe an underlying phenomenon. In this case, one truly needs to think “outside the box” and for such problems deductive reasoning is called for.
Deductive reasoning does not start with the assumption that the problem posed can be solved by simply applying known theory. Rather, one needs to determine the solution by advancing new premises or hypotheses that are consistent with the solution of the problem. The logical consequences of these hypothesis are then further worked out to determine if they contradict experimental results or if they predict new outcomes that solve the problem at hand and also make new predictions beyond the inductive framework. N. Bohr, A. Einstein, Galileo, Newton, Maxwell, and many others utilized deductive reasoning to make new advances and develop entirely new theories.