Scientific method can be understood as the following steps: formulating a hypothesis, designing an experiment, carrying out experiments, and drawing conclusions. Conclusions can feed into hypothesis formulation again, in order for a different (related or unrelated) hypothesis to be tested, and we have a cycle. This feedback can also take place via a general theory that conclusions contribute to and hypotheses draw from. The theory gets to represent everything we have learned about the domain so far. Some of the steps may be expanded into sub-steps, but in principle this cycle is how we generally think of science.
This looks quite simple, but is it really? Let’s think about hypothesis formulation and drawing conclusions. In both of these steps, the results are bounded by our imagination and intuition. Thus, something that doesn’t ever enter anybody’s imagination will not be established as scientific fact. In view of this, we should hope that scientists do have vivid imaginations. It is easy to imagine that there might be very powerful findings out there, on the other side of our current scientific horizon, that nobody has yet been creative enough to speculate about. It is not at all obvious that we can see the low hanging fruit or even survey this mountainous landscape well – particularly in an age of hyper-specialisation.
But scientists’ imaginations are probably quite vivid in many cases – thankfully. Ideas come to scientists from somewhere, and some ideas persist more strongly than others. Some ideas seduce scientists to years of hard labour, even when the results are meagre at first. Clearly this intuition and sense that something is worth investigating is absolutely crucial to high quality results.
A hypothesis might be: there is a force that make bodies with mass attract to each other, in a way that is inversely proportional to the squared distance between them. To formulate this hypothesis we need concepts such as force, bodies, mass, distance, attraction. Even though the hypothesis might be formulated in mere words, these words all depend on experience and practices – and thus equipment (even if the equipment used in some cases is simply our own bodies). If this hypothesis is successfully proven, then a new concept becomes available: the law of gravity. This concept in turn may be incorporated into new hypotheses and experiments, paving the way for ever higher and more complex levels of science and scientific phenomena.
Our ability to form hypotheses, to construct equipment and to draw conclusions, seem to be human capacities that are not easy to automate.
Entities such as matter, energy, atoms and electrons become accessible – I submit – primarily through the concepts and equipment that give access to them. In a world with an alternate history different from ours, it is conceivable that entirely different concepts and ideas would explain the same phenomena that are explained by our physics. For science to advance, new equipment and new concepts need to be constructed continually. This process is almost itself an organic growth.
Can we have automated science? Do we no longer need scientific theory? (!?) Can computers one day carry out our science for us? Only if either: a) science is not an essentially human activity, or b) computers become able to take on this human essence, including the responsibility for growing the conceptual-equipmental boundary. Data mining in the age of “big data” is not enough, since this (as far as I know) operates with a fixed equipmental boundary. As such, it would only be a scientific aid and not a substitute for the whole process. Can findings that do not result in concepts and theories ever be called scientific?
If computer systems ever start designing and building new I/O-devices for themselves, maybe something in the way of “artificial science” could be achieved. But it is not clear that the intuition guiding such a system could be equivalent to the human intuition that guides science. It might proceed on a different path altogether.