Artificial Inteligence (AI) is a very broad topic with a multitude of different perspectives, living in the
intersection of mathematics (esp.
statistics and logic), computer science, engineering (esp. robotics), economics, psychology (esp. cognitive
science), linguistics, biology (esp. neuroscience) and philosophy. Depending how you define AI, its history goes
back either to very recent research, starting around 1956 with a “Summer Research Project on Artificial
Intelligence”, or to the great ancient Greek philosopher Aristotle himself. Thus, an extensive
view on AI is naturally impossible within a few pages. Here, AI is introduced on a high level with emphasis on two
point of views: technological and innovation-theoretical.
In its very core, AI “attempts not just to understand but also to build intelligent entities” (Russell et al., 2010).
To be able to do so, Minsky (1961) stated five different subdomains of problem solving that science needs to
solve. For a given problem instance, a machine or entity needs to use (1) Search techniques to traverse the
solution space, (2) Pattern-Recognition for improving time and space efficiency, (3) Learning for
generalizing from past experience, (4) Planning for allocating scarce resources like computational time and
finally, (5) Induction for gaining knowledge out of all findings. But there are still plenty of dimensions in
which one can distinguish different types of AI or respective fields of AI research. These include:
- Should this entity be rational, or human-like (so per se irrational to some degree)?
- Is it enough to be able to simulate thinking for decision-making or should it also act
autonomously in a certain environment?
- Is it able to fulfill only one task very well (“applied” or “narrow” AI), or any given task
(“full” AI or “Artificial General Intelligence”)?
- And finally, does the AI truly possess a conscious mind or just pretend to have one (“strong” vs.
For most of the business applications, the type of AI is weak, applied, mostly rational and thinking-orientated.
Particularly, one sub-area of AI, namely Machine Learning (ML), is the main driver of the core problems of
Pattern-Recognition and Learning.
As an innovation, AI and ML is still far from being widely adopted in business. On the one hand, one can argue with
the general diffusion pattern of new technology, which often follows an S-curve. This reflects the fact that
diffusion rates of any innovation usually first rise and then fall over time. The popular “Epidemic Model” explains
this behavior by arguing that the limiting factor of technology diffusion is the “lack of information available
about the new technology, how to use it and what it does”. This might be true to a certain
degree for AI, since even today society and businesses are struggling to understand AI. On the other hand, AI
history is specifically characterized by a series of hypes, quickly followed by phases of disenchantment between
1970s until the mid-2000s, so-called “AI winters” (Bengio, 2016). Bengio argues that missing computational
power, a lack of data and insufficiently sophisticated algorithms yielded poor practical results, which is the
reason for the decline in AI research.
Nevertheless, since 2005, AI got an accelerating wave of interest. This hype is mainly driven by one special ML
technique, called Deep Learning (DL). DL is based on a relatively old technique, known as Artificial Neural Network
(ANN), which is in turn inspired by the architecture of the human brain. The hype around DL rests upon multiple
success stories, where the algorithms archived human or even super-human level performances, especially in the field
of computer vision (e.g. for autonomous driving, face recognition) natural language processing (e.g. for chatbots
and virtual assistants), or board games (e.g. Go, chess). Cockburn et al. (2017) state that DL will have a large
impact on the whole economy, because it represents an Invention of a Method of Inventing (IMI) as well as a General
Purpose Technology (GPT) at the same time. IMIs “enable a new approach to innovation itself, by altering the
‘playbook’ for innovation in the domains where the new tool is applied”. They argue that DL is indeed an IMI since
it may be able to substantially "automate discovery" and "expand the playbook" in the sense of opening up the set of
problems that can be feasibly addressed. Furthermore, DL is also a GPT. This means that DL can be characterized by
the potential for pervasive use in a wide range of sectors and the role of an "enabling technology", opening up new
opportunities rather than offering complete, final solutions. Ultimately, since AI is a GPT and a IMI at the same
time, it is arguably one of the most important innovation enabler of our time.
Bengio, Y. (2016): Machines Who Learn. Scientific American, 314(6), 44–51.
Bresnahan, T. F., & Trajtenberg, M. (1995): General purpose technologies ‘Engines of growth’?
Journal of Econometrics, 65(1), 83–108.
Cockburn, I. M., Henderson, R., & Stern, S. (2017): The Impact of Artificial Intelligence on
Innovation: An Exploratory Analysis. In NBER (Ed.), NBER Conference on Research Issues in
Artificial Intelligence. September 2017, Toronto, Canada.
Geroski, P. A. (2000): Models of technology diffusion. Research Policy, 29(4-5), 603–625.
Gubrud, M. A. (1997): Nanotechnology and International Security. Fifth Foresight Conference on
Molecular Nanotechnology. Retrieved from Foresight Institute website:
foresight.org/Conferences/MNT05/Papers/Gubrud/ on 16/08/2018.
Minsky, M. (1961): Steps Toward Artificial Intelligence. Proceedings of the IRE, 49(1), 8–30.
Nilsson, N. J. (2010): The Quest for Artificial Intelligence: A History of Ideas and
Achievements. Cambridge, New York: Cambridge University Press.
Russell, S. J., Norvig, P., & Davis, E. (2010): Artificial Intelligence: A Modern Approach. 3rd
ed. Upper Saddle River, NJ: Prentice Hall.
Searle, J. R. (1980): Minds, brains, and programs. Behavioral and Brain Sciences, 3(3), 417–424.
About the author