From a materialist perspective, evolution is an expression of the laws of thermodynamics, and is thus both conservative (the first law) and non-conservative (the second law). It plays out developmentally, i.e. via feedback-driven symmetry-breaking processes, occurring within ‘bounded’ systems of interactions, that actualize (by progressively increasing the probabilities of) some (singular) possibilities at the expense of (all) others. Whatever is predictable about nature (and hence evolution) is so owing to systemic development.
Insofar as evolution involves chance it is unpredictable. Development nevertheless progressively leads to somewhat predictable ends. I say “somewhat” because the predictability of development increases with development. Immature systems are highly unpredictable, whereas mature systems behave predictably, and their continued development predictably ends with their extinction (or death). Fortunately for life, evolution is creative, continuously (and unpredictably, by chance) spawning new incipient systems with varying amounts of developmental potential. Some of these systems emerge developmentally via metamorphosis of old (‘senescent’) systems.
Science then, to the extent that it is a predictive (as opposed to explanatory) enterprise, can only deal with (relatively late) development, not evolution in general, or the particulars to which it by chance gives rise. Mechanisms come to be via development, and act in turn to constrain further development in more or less predictable ways. So, science lacks the capacity to foresee origins, and therefore cannot possibly direct the psycho-social-cultural-economic metamorphosis that is needed for the survival of civilized humanity. The best it can do is alert us that metamorphosis is needed.