From Nick Kalliagkopoulos – Prime Ventures (@kalliagk)

There is a lot of hype on Artificial Intelligence (AI). Gartner placed Machine Learning on top of its Hype Cycle for Emerging Technologies in 2016, while at the same time forecasting only 2-5 years to mainstream adoption. As a field it is not that new. It was firstly described in a paper of Alan Turing, published in 1950 and titled Computing Machinery and Intelligence. Since then many years have passed, with multiple developments on the field. Still, the  AI field has not fulfilled the huge expectations.



I am personally highly interested on Artificial Intelligence and its transformation potential. I want to share 5 points that I consider important on this technology.

But first lets give 3 very basic definitions.

  • Artificial Intelligence: Intelligence exhibited by machines. AI is a technology that tries emulate human performance and reach its own conclusions. The basic idea behind the AI field is that human intelligence can be described so precisely, that a program can be developed to simulate it.
  • Machine Learning: Machine Learning is an approach to achieve AI. While in traditional computing, computers are fed with specific instructions, in AI approached, computers are capable of learning from data and coming to their own conclusions.
  • Deep Learning: Is a technique to implement machine learning

1)  The time of AI has come
Almost seventy years after Alan Turing published his original paper, I believe the time for AI is here. There are three main reasons for that.a) Computing power is becoming faster and storage is becoming cheaper.
b) The above, in combination with numerous cheap sensors are driving an exponential growth of data generation around us. Humanity is creating data sets that never existed before. Technologies in the broad field of AI have the potential of enabling businesses to make better sense of these data.
c) The quality and speed of the AI algorithms has increased. AI driven products are already out there, improving the performance of search engines, financial trading, fraud detection, loan origination and more.2) AI will create new ‘Winner takes all markets’ due to the Data Network Effects that it inherently creates
In a famous article published in Harvard Business Review in 1996 (entitled Increasing Returns and the New World of Business), Brian Arthur articulated the point that although the basic economic principles were based on the assumption of diminishing returns, the new economy that is fuelled by technology has shifted to one of increasing returns. The assumption of diminishing returns means that products or companies that get ahead in a market eventually run into limitations, so that a predictable equilibrium of prices and market shares is reached. In contrast, the reality of increasing returns stems from the fact that there are mechanisms of positive feedback that operate within markets and industries and result in those who are ahead to get further ahead and those who lose advantage to lose further advantage. This is creating ‘Winner takes all markets’.I believe that AI has the potential to magnify ‘winner takes all’ markets even more, and create new winners. This mainly derives from Data Network Effects. Data Network Effects originate from the fundamental principles of Machine Learning. More users in a product will lead to the generation of more data. More data will then be used to train the algorithms and create a better product. Therefore strong data network effects reduce the threat of new entrants and increase the importance of data ownership (due to the monetisation possibilities and increase of switching costs).
​​ Picture

3) Good luck disrupting Google, Apple and their likes
Startups in the 90s or 00s were trying to disrupt companies like IBM. Startups today are trying to disrupt companies like Google, apple, twitter and Microsoft. Of course these companies are completely different type of incumbents, who understand the potential impact of new technology. Most of them are already highly active in the space and are capable of attracting high profile data scientists and engineers. Larry Page (Google CEO) mentioned in October 2000 that ‘AI would be the ultimate version of Google. We are nowhere near doing that now, however we can get incrementally closer to that and that is what we work on’. Fast forward 16 years to today, and Larry Page mentions that ‘We have been building the best AI team and toold for years and recent breakthroughs will allow us to do even more. We will move from mobile first to an AI first world’.4) There is substantial momentum at the moment
According to GP Bullhound $4.2 billion were invested in 591 deals in 2016 in the AI space. At the same time 40+ acquisitions took place. CB Insights published a very nice graph (below), visualising the acquisitions of AI companies over the past years. Google, Apple, IBM, Intel and Microsoft are continuously acquiring tech companies.


5)  White collar jobs will be replaced. 
Technology has always been replacing jobs. The industrial revolution made many jobs obsolete. Most of the jobs that were affected though were blue collar jobs. AI has the potential to replace typical white collar jobs, like lawyers, accountants and even doctors. I believe there will be a fundamental shift in professional service organisations. Machine Learning and AI are moving from hype to adoption, and this can result in significant disruption in multiple enterprises. Very interesting areas with huge potential of AI and Machine Learning include healthcare (genomics, drug discovery), Autonomous Vehicles, Agriculture (optimizing seed planting, sorting of vegetables, identification of sick plants), Finance (credit scoring, compliance), Personal Assistants and conversational bots.My thesis
AI is overhyped and is increasingly used as a buzz word. Therefore, investors need to be cautious and look for real technology. I personally believe the next big startup (and hence VC) successes will be in vertically focused companies (broad horizontal platforms will attract large tech giants that will be a tough competition), with an enterprise focus (consumer focus takes longer, is more competitive and thus hard to monetise), that are exploiting data network effects (reducing the treat of new entrants and creating a viral loop) and own the resulting data (offering further monetisation possibilities, better algorithms and higher switching costs).  ​​