02.03.2016 – Von Dirk Solte. In order to discuss the probable future economical and societal impact of Big Data and Analytics, the term Big Data & Analytics (BD&A) has to be defined precisely. That is anything but easy. Sometimes maybe a joke can convey the essentials: Trying to get a correct answer to a question via BD&A, figuratively speaking, is like trying to get inside knowledge about the behaviour of a pig by analysing all the dishes of a huge number of potluck dinners with mathematical methods.
That is a very brief description of what we could think about BD&A. On a serious note, there is a big difference between conventional ICT and data driven innovations that lie ahead with BD&A:
As a first aspect, BD&A solutions provide a kind of “manifest what” by extracting value from a flat and unstructured “datafied universe of information-shreds” with unknown veracity. This helps to answer questions on the basis of approximations and correlations.
The difference is that conventional data and analytics provide a kind of “know what” instead of “manifest what” by implementing “know why” as value. Due to this approach questions are answered on the basis of implemented causation. Thus, the difference is mainly correlation in the case of BD&A instead of causation in the traditional field.
Based on this very rough but hopefully precise enough definition, including the differences of classical and BD&A solutions, the results of a study, which was part of a large OECD-project “Data-Driven Innovation: Big Data for Growth and Well-Being”, presented at the “OECD Global Forum on the Knowledge Economy 2014” in Tokyo, are summarized.
Firstly, let us assume that ecological, societal and economical sustainability via a green and inclusive market economy is our common goal.
Secondly, let us assume green and inclusive growth of Gross Domestic Product (GDP) with a fair distribution pattern of entitlement and participation is a prerequisite for societal and ecological sustainability.
BD&A have a big potential in this direction:
We should keep in mind that the current level of employment, meaning the number of employed people in the OECD countries is just over 500 million. We should compare this number to a probable workforce in the next 30 years of about seven billion people. So on the one hand, BD&A could have a positive impact on the direction of green and inclusive growth and a green and inclusive market economy but on the other hand, it is important to mention that BD&A could also have negative impacts.
There is the problem of total transparency and as a second aspect, the possibility of technologically-induced unemployment. Nevertheless, we should also reflect on what has already been outlined concerning the impact of BD&A on employment: Correlation and statistical evidence lie at the heart of BD&A; the results and manifestations are qualitatively sanctioned by the law of large numbers.
That is why BD&A accelerates the societal move regarding decision-making “from causation to correlation” or “from ‘know why’ to ‘manifest what’”. In other words, machines will tell us what to do and when, via pattern recognition and quantitative reasoning.
There is a warning that has to be given here, which is perfectly described and illustrated by Nassim Taleb, who says “Be careful. Black swans do exist!” For example, the financial crisis was not predicted by econometrics, which uses the methods applied in BD&A. Furthermore, the big rating institutions classified a huge amount of securities “triple A” using quantitative risk management and these all became toxic papers a few minutes later. Thus, the ideal solution would be a combination of humans and machines.
The key political challenges result from the following facts:
Firstly, there are skills necessary far beyond the data-focused part of science, technology, engineering and mathematics (STEM). Only by teaching orientating knowledge far beyond STEM is it possible to enable people to detect those “black swans” in advance via qualitative reasoning instead of quantitative reasoning. This requires huge and multidisciplinary know-how and know-why.
Secondly, the regulation of responsibility and accountability is a necessity in order to enable people to “override” machine- and correlation-based decisions in situations where qualitative reasoning or even – what could be called – creativity or intuition contradicts statistical evidence.
The following question leads to such a situation where causation contradicts correlation: What is the predicted impact of BD&A on employment?
McKinsey predicts the potential of BD&A as five or up to ten percent on productivity growth over the next few years.
Since productivity growth is mainly the optimization of automation it will be medium-wage jobs of the last few decades that are at stake. This would cause the following impact on OECD countries. It has already been outlined that there are just over 500 million people presently employed in the OECD countries. Five to ten percent of productivity growth would primarily cause a job loss of 26.75 – 53.5 million jobs.
What’s left for humans to do? On the one hand, there are high-wage jobs that need a high level of creativity and the ability to handle situations of undecidability. On the other hand, the overhang of workforce supplies employees for low-wage jobs that do not require many skills but require a high level of what could be called “manuality”, which means the ability to handle non-routine manual tasks and requires a high level of “sociality”.
What is the good news about employment? Eventually there will be new opportunities for employment because of BD&A.
The study predicts 1.6 to 1.7 million job opportunities by 2018, which could turn into seven million potential new jobs in the OECD countries.
The big question is: How will we fill the job opportunity gap, which will be approximately 20 to 45 million jobs in the OECD and how will we fill the possible job gap worldwide, which will be much higher?
There is a correlation argument that answers this question in a statistical form and says: “Don’t worry! The principle of creative destruction will remain true in the future.” Nevertheless, a big “if” has to be added to this statement if we follow this correlation argument. The reason is the fact that we are faced with a big challenge. The successful attainment of a “sustainability-triple” – massive green growth of GDP, fair and balanced participation and at the same time solving the problem that our current ecological footprint is about 1.5 planets – requires a big transition and this challenge and the resulting needs are predicted by investigating the situation via correlation and causation.
The causation argument concerning the question of the predicted impact on employment is different. Causation tells us: “This time is different”. The argument is that we will probably have technological unemployment due to an accelerated innovation cycle and the highest innovation speed throughout history. One argument that makes this plausible is Moore’s law and another argument comes from Prof. Sergej Kapiza, a member of the Club of Rome who has figured out that throughout history the speed of innovation has been a quadratic function with respect to the number of people living on earth. Thus “technological obsolescence” of human capabilities within the living workforce population has to be considered even if we consider lifelong-learning.
The key political challenge resulting from this analysis would be to invent societal innovations and to implement corresponding regulations that are needed to ensure fair entitlement and participation: A fair share in GDP.
This huge challenge requires well-founded political efficacy through the teaching of broad skills far beyond a data-focused STEM education and lifelong-learning.
We could conclude this analysis with the question: Will the statistically evident paradigm of “creative destruction” to overcompensate the expected job losses in the medium-wage segment remain true in the near future due to the BD&A-induced productivity growth?
Alternatively, we could ask the question: Will we witness a high level of “technological unemployment” and a “winner takes it all” economy? When answering these questions, causation contradicts correlation!
Thus, the key recommendation is: Seriously consider the possibility that BD&A will have negative impacts on employment and equity in the years to come.
Figuratively speaking, the key challenge is to create a global and adequate musical arrangement to effectively “dance with the machines” in sustainable harmony!
Big Data and Analytics – What are the perspectives? Lassen Sie uns diskutieren, ich freue mich auf Ihre Meinungen und Kommentare an firstname.lastname@example.org
Die Kolumne „Zur Debatte“ erscheint regelmäßig im Magazin Versicherungswirtschaft. Mehr zu dieser Ausgabe