When techs attack


When techs attack

-article also published on Linkedin on 7 June 2017 –

The world is all excited about the emergence and development of new technologies in such a wide variety of disciplines as the natural sciences, life sciences, knowledge of our universe and engineering sciences.


To put this craze into perspective, the world also thought that it was at a major turning point in its evolution a hundred years ago.

Albert Einstein had just published his theory of general relativity, the vaccine against tuberculosis was developed, the poles had been visited, Ford’s Model T began production, and the first airlines were opening, just thirteen years after the Wright Brothers’ first flight.

It is always hard to judge one’s own times and it won’t be until later that we can look back and designate which recent discoveries were important. And yet, with improvements in our means of communication and in the accessibility of information, the cycles of adopting new technologies are getting shorter.


In the engineering sciences, computer hardware is still undergoing major improvements, whether in terms of computing, memory or communication capacities. Most smartphones exceed the entire Apollo space programme computing power. It is possible to buy today a CPU one million times more powerful than in 1987, thirty years ago, for a similar price, a few thousands euros.

The price of sensors to capture information is going down, so more and more data are being produced, stored and shared.


With data and the ability to process it, people are processing data, finding patterns, analogies, correlations, predictions. The old fashioned statisticians are being replaced by data scientists enhancing theories formulated decades ago.

Copying data is as easy as copying algorithms. Creating an algorithm at the same time similar and different from a patented one is usually relatively easy when creating fake data is useless. Algorithms are not the scarce resource, but data used to train computers and data to sell better and faster is the new oil.


Artificial Intelligence has different aspects. A quick and simplistic way is to look at it as tool box of logic, optimization, sophisticated statistics and probabilities for pattern recognition. Very catchy results have been widely publicised with Watson, AlphaGo and various self driving cars. These systems can exceed the best human experts, becoming a form of superhuman intelligence.

When Man was hunting elephant or lion, he was surpassed in strength and expected catch up with intelligence. Man has then built machines to exceed his own force. And now for the first time, Man has built an ‘intelligent’ machine that seems to surpass his intelligence, at least a kind of intelligence.

This is triggering all types of hopes and fears.

At this point, demons raise their ugly heads, such as HAL, the murderous computer in 2001, A Space Odyssey, or the world dominated by machines in Matrix. And all the doomsday prophets are relaying similar ideas. A parallel can be drawn with danger related to genetics, all the fear regarding eugenics, culling, genetic enhancement or discrimination.

At the other end of the spectrum, naïve are expecting a world of arts and sciences where Man is served by Machines, a new idyllic paradise.


A technological innovation can bring a significant advantage to those who own it or to those who have access to it. There are many examples of this in the military field (gunpowder, aircraft, nuclear weapons, etc.)

Nobody knows yet if they will be in a position to take advantage of artificial intelligence as an owner of algorithms or data, renting them to third parties, or as a user, paying to benefit from them, or if they will be incapable of accessing them – an outcast. This anxiety brought about by the prospects created by artificial intelligence in the workplace is producing a reaction in the form of conservative attitudes that are amplified by the prospects of having to manage more outcasts.


Artificial intelligence is not a new subject. It was conceptualised several decades ago by such famous contributors as Kurt Gödel and Alan Turing. The change is the recent examples that are leading us to believe in its generalised adoption, possibly as quickly as the adoption of the smartphone. If this is the case, we will have to undertake discussions on how we organise society in terms of the division of labour and the distribution of wealth, or on how this intelligence is controlled.

It should be pointed out, however, that so far the artificial intelligences deployed are expert systems (weak artificial intelligence). Many of these systems, like neural networks, operate using a black box method. They give solutions without being able to explain why a solution is chosen, which is unsuitable for human logic and laws.

General intelligence systems (strong artificial intelligence) that can select useful data for solving a problem, notably poorly formalised implicit data, or that have a capacity for innovation, improvisation or creativity and can explain their reasoning, have yet to be invented.


AI-based expert systems have two possible operating principles.

They either operate on a sophisticated statistical or probabilistic process, deriving their ‘knowledge » from training that finds similarities or patterns in data with or without the support of a user. This is used for voice or text processing, or sentiment analysis.

Alternatively, they operate using a rule engine, fed by a subject matter expert; however, the difficulty here lies in adjusting the precedence between the various rules. Such systems are used today for landing planes by autopilot, for example.

Typically a stat-based system operates at 80% accuracy, while a rule-based system only works in 80% of cases, as the remaining 20% cases are still to be planned for. Unless 100% is achieved, such systems would remain assistant rather than autonomous or independent AI.


Fintechs provide services for core activities of Investment Management.

Client management is improving with the use of KYC facilities, client profiling, text and voice interaction management and community management; including meetic-like services between investors and asset managers.

Support functions, such as legal and compliance are benefiting from new lobbyists, regulatory or compliance watch services. Support function fintech can appear less exciting than front office fintech. However, RegTechs are upgrading their digital services to achieve the same standards.

There are now alternative sources of information on companies and their activities, and services like social media sentiment analysis. For risk management, the well-known Robo-Advisors should be assessed, just like API-based pre- or post- trade services, or even liquidity risk assessment services.

When techs attack, everyone should adapt