Can science answer every question?

Which kind of question are you asking?

Compare how a rainbow forms with whether a painting of one is beautiful. Science can investigate light and visual perception. Beauty also involves a judgment. Similarly, evidence about the consequences of a decision can inform ethics without selecting everyone's moral priorities. UC Berkeley's Understanding Science explains these limits. They do not make questions about meaning unimportant; they make the kind of answer we seek important. Science and value judgments.

Here is an editorial example. Suppose a community must divide a budget between two worthwhile projects. Research might estimate their costs and consequences. But choosing how much to prioritize fairness, immediate benefit, or future benefit requires reasons about values. Gathering another measurement can improve the estimates while leaving that choice unresolved. The practical task is to state both the evidence and the priorities openly.

Why do scientists keep saying we do not know yet?

Because a question can have a supported part and an unanswered part. NASA describes dark matter through gravitational effects and observations such as gravitational lensing, while noting that proposed constituents have not been confirmed. Evidence for a phenomenon is not automatically an explanation of everything about it. Naming an unknown can help researchers design tests instead of filling the gap with a confident story. NASA: Dark Matter.

This distinction also tells you what to ask next. Does the unknown concern whether something occurs, how it occurs, or what it is made of? Those are separate claims. An answer to one does not necessarily answer the others. Conversely, an unanswered detail does not erase every observation already established. Ask the speaker to identify the exact unresolved point.

Are there limits to what humans can ever know?

There are limits to particular methods and measurements. NIST's measurement guidance explains that a measured value is generally an estimate and that reporting uncertainty is part of a complete result. Knowing a range can be useful knowledge, even when an exact value is unavailable. More precise instruments may improve an estimate; they do not justify pretending the current uncertainty is zero. NIST: measurement uncertainty.

For a claim that permits no observational distinction between competing explanations, science has no test to choose between them. That is a boundary of the method, not a certificate that the claim is true. Nor can we give a complete list of everything humans will never discover. Confidence is high about the need for testable evidence; predictions about all future knowledge deserve caution. A new test or observation could turn an open question into an answer. Continue with scientific proof and questions about meaning.