IEEE Technology and Society Magazine - March 2018 - 41

scientific discovery in existence, and it is too early to
generalize and propose new theories. However, being
"in the midst of ... a revolution" it is important for the
research community to re-examine views pertinent to
scientific discovery. This paper highlights how experience with robot scientists could inform discussions
in other disciplines, from philosophy of science to computer creativity research.

Scientific Discovery and Robot Scientists
The branch of Artificial Intelligence (AI) devoted to developing algorithms for acquiring scientific knowledge is
known as "scientific discovery." The pioneering work in
scientific discovery was the development of learning
algorithms for analysis of mass-spectrometric data [2]. In
the subsequent 50 years, much has been achieved and
there are now convincing examples in which computer
programs have made explicit contributions to scientific
knowledge (e.g., [3]-[6]). However, the general impact of
such programs on science has been limited. This is slowly changing as the expansion of automation in science is
making it increasingly possible to couple scientific discovery software to laboratory instrumentation [6]-[9].
Science is an excellent testbed for the development
of AI discovery systems:
■■ Scientific problems are abstract, but also involve the
real-world level knowledge.
■■ Scientific problems are restricted in scope - no
need to know about "Cabbages and Kings" - and
are also are extensible.
■■ Science assumes that Nature is not trying to deceive
us, so there is no need to consider malicious agents.
■■ Scientific knowledge is a public good when it is
openly available.
■■ Science is a worthy object of our study.
A robot scientist is an example of such an AI discovery system. The robot scientist is a physically implemented laboratory automation system that exploits AI
techniques to execute cycles of scientific experimentation [8], [9], [11]. A robot scientist automatically originates
hypotheses to explain observations, devises experiments
to test these hypotheses, physically runs the experiments
by using laboratory robotics, interprets the results, and
then repeats the cycle. The advent of robot scientists is of
significant philosophical and social interest. They are
also of practical interest as they have the potential to
increase the productivity of science: they can work cheaper, faster, more accurately, and longer than humans, and
can also be more easily multiplied.
The robot scientist "Adam" was the first machine to
autonomously discover scientific knowledge, i.e., to
autonomously both form and experimentally confirm
novel hypotheses [8]. Adam worked in the domain of
yeast functional genomics, and autonomously both
MARCH 2018

∕

generated functional genomics hypotheses about the
yeast S. cerevisiae, and experimentally tested these
hypotheses by using laboratory automation. Adam's
conclusions have been manually confirmed using gold
standard experiments.
The robot scientist "Eve" (Figure 1) was designed to
make drug discovery more economical, specifically for
neglected tropical diseases [9]. Eve integrates and automates library-screening, hit-confirmation, and lead generation through cycles of quantitative structure activity
relationship learning and testing. Using econometric
modeling, Eve was shown to economically outperform
standard drug screening. Eve has repositioned several
drugs against specific targets in parasites that cause
tropical diseases. One validated discovery is that the
anti-cancer compound TNP-470 is a potent inhibitor of
dihydrofolate reductase from the malaria-causing parasite P. vivax.

The Metaphysics of Robot Scientists
A major motivation for the automation of science is philosophical: if a mechanism can be built that is judged to
have discovered some novel scientific knowledge, then
this will shed light on the nature of science. To quote
Richard Feynman "What I cannot create, I do not understand" (on his blackboard at the time of his death). The
advantage of this approach to the philosophy of science, compared to traditional ones, is that it is constructive and objective. In building robot scientists one is
confronted with the need to make concrete engineering
decisions that relate to a number of important problems
in the philosophy of science: the relation between
abstract and physical objects, the nature of truth, the
relation between observed and theoretical entities, the
origin of hypothesis, the problem of induction, etc. This
approach to science is analogous to the AI approach to
understanding the human mind through the creation of
artifacts that can be empirically shown to have some of
the attributes of human minds [12].
We argue that the software/hardware isomorphism is
the key to bridging the physical/abstract dichotomy in the
metaphysics of science (Figure 2). The key to the power of a
computer is that computers implement abstract programs in physical devices. This is the insight that distinguished Turing from the other great logicians of his time.
Although the idea that a physical object can be isomorphic with an abstract system is at least as old as the abacus, a Universal Turing Machine is a uniquely powerful
physical/abstract device.
In a scientific investigation, to relate corresponding
abstract and physical entities requires the concept of
"truth." Within philosophy there are a number of competing theories of truth, including: correspondence, pragmatism, verification, and coherence. These theories are

IEEE Technology and Society Magazine

Table of Contents for the Digital Edition of IEEE Technology and Society Magazine - March 2018