A useful digital experiment controls variables. In y=mx+b, hold b fixed while changing m to isolate the effect of gradient; hold m fixed while changing b to isolate the intercept effect.
The conjecture must be stated before or during testing, then challenged with positive, zero and negative cases where relevant. One attractive graph is not proof of a general rule.
Digital tools accelerate variation and comparison, but reasoning remains mathematical: identify what changed, what stayed invariant and whether the evidence supports the claimed generalisation.