Skip to main content
In the News   
Ex: Europe, Middle East, Education

Environment: What Big Data and Machine Learning Have to Offer

ARTICLES - 26 October 2020
Key Points
1

Big data and the different branches of machine learning are first-rate tools in the fight against global warming. They make it possible to grasp the complexity of reality and reveal factors that were previously invisible.

2

Agriculture is an example where digital tools make it possible to search for causal relationships between multiple factors. The same is true in the mobility sector or the supply chain management sector.

3

To make progress, data sharing policies must be put in place with a simultaneous implementation of mechanisms, to encourage players to measure and control their environmental impact, such as the carbon dividend.

The launch of 5G saw the emergence of a debate between the pro and anti 5G community, which in many ways was indicative of a poor understanding of what this technology is and what its externalities are. It has also revealed how we are still poorly equipped to enter a world where the environment is becoming an over-determining factor in public policy and in our daily actions. 

Managing the complex and the invisible

The management of global warming must deal with complex and invisible elements: 

Complex because the environment, like 5G, is massively based on multiple factors. It does not make much sense to isolate a variable from its environment. Modelling therefore requires models of a new paradigm.

Invisible because, like 5G's electromagnetic radiation, carbon dioxide is invisible — which can lead figures like Trump to deny the existence of global warming.

To meet these challenges, digital technology — in particular big data and the different branches of machine learning — show great potential. This is because these fields have two characteristics that seem important to keep in mind: they know how to manage complexity, and they reveal the invisible. 

The real breakthrough will be to create models that interact with massively multifactorial environments. 

To meet these challenges, digital technology — in particular big data and the different branches of machine learning — show great potential. This is because these fields have two characteristics that seem important to keep in mind: they know how to manage complexity, and they reveal the invisible. 

A striking example: agriculture

Currently, there is widespread global deliberation on what a virtuous agricultural model could look like. 

Is it organic farming without inputs? Is it intensive agriculture? Is it so-called preservation agriculture? How can we reduce the environmental footprint of agriculture — which accounts for about 20% of greenhouse gas (GHG) emissions — while producing in a qualitative manner? How can we limit water capture? How can we develop a better understanding of erosion and a more effective way of dealing with it? What are the emissions of various pollutants?

Faced with this complexity, traditional methods come up against our cognitive limits. Many agronomists across the globe believe that robotization, and more broadly precision agriculture, will enable the productive, quantitative and qualitative leap that must be made in order to feed the predicted 2050 global population, without further poisoning our ecosystem. 

Going further, the real breakthrough will be to create models that interact with massively multifactorial environments, in which hundreds of parameters of a very heterogeneous nature have to be taken into account. To achieve this, our specialized approaches are no longer suitable. This is true in many other domains (smart cities, supply chains, etc.). This is a major paradigm shift, which should lead us to rethink our educational and academic systems, our production methods, our administrative organization, our way of building political consensus, etc. 

The complexity of urban environments and supply-chains 

Our urban environments also face challenges of complexity. Cities directly or indirectly emit 60% of the world's GHGs. However, schools are occupied only 20% of the time, office buildings 30% of the time, and public transportation is 130% full for less than 10% of the total time. 

The relationship between flows, stocks and infrastructure is largely deficient and, in reality, poorly understood. Much more could be done with a better understanding of these many interrelated factors. For example, with the help of artificial intelligence, we could rethink the challenge of energy efficiency improvements.

With the help of AI, we could rethink the challenge of energy efficiency improvements.

Expert systems would be able to optimally indicate the most suitable techniques, building by building, and according to the specific characteristics of the structure, for the best possible result at the lowest possible cost. Today, what largely undermines the effectiveness of these energy efficiency improvement policies is specificity, insofar as the trades, buildings, renovations and administrative processes each have very distinctive and specific criteria. Beyond that, there is a lack of access to the data needed to implement these new approaches, and of course the appropriate skills to build these systems. 

In supply chains, which, let's not forget, are responsible for almost 70% of the footprint of production sectors, we could do the same. According to the European Commission, between 2000 and 2017, the filling rate of trucks in the European Union grew by 14%. Has it ever occurred to anyone that this increase can only be due to information systems, in other words the Internet? The next challenge is to greatly increase their resilience, to reduce their environmental footprint by having a more traceable supply.

The question of data 

Obviously, the first thing we need, if we want to be able to use the potential of big data and machine learning, is data, a lot of it. Yet it is astounding to observe how different the situation is from one field to another. In some fields, such as air transport, it’s been more than twenty years since data has completely shifted to platform-based systems. All the data — the identifiers of an aircraft, its flight parameters, our reservations, etc. — has been standardized worldwide and is shared on this scale. However, in agriculture, the opposite is true: the data is opaque, especially when it serves the interests of intermediaries. Consumer associations and farmers' unions frequently denounce these market characteristics in which intermediaries succeed in creating rent-seeking situations due to the intentional opacity that reigns in the intermediation of this type of product. 

The most effective device seems to me to be the carbon dividend [...], which consists in introducing a carbon tax starting at €50 per ton on all economic activity in the European Union.

In urban environments, such as in the field of mobility for example, it is very largely the American platforms that hold the data. However, it is difficult to deny that this data is potentially of general interest. We need to make it so that we can access this data. Private players should not be the only ones accessing the data in order to develop innovative services from it. In supply-chains, just as in industrial sectors, there is still an important issue of harmonization. We need to encourage players to work towards this harmonization, guaranteeing sovereignty, resilience and economic performance.

Encouraging the consideration of negative externalities

The challenge of regulation does not end here: it must also ensure that externalities appear clearly and are associated with a cost that makes it possible to limit them, while at the same time continuing to stimulate innovation. The most effective device seems to me to be the carbon dividend defended by the Institut Montaigne, which consists in introducing a carbon tax starting at €50 per ton on all economic activity in the European Union. This tax would also apply to imported products and its proceeds would be paid in full to every European household. Thus, for the French, who emit about 7 tons of carbon dioxide per year, it would bring in about €350 per person to the State, to be shared among the population. The big carbon consumers would lose, the most frugal would gain. 

Moreover, this tax would have an important competitive benefit for the EU. It is already the most advanced of the major economies in terms of emissions per unit of GDP, with 0.18 kg of carbon dioxide generated per dollar of wealth produced — where the United States stands at 0.32 kg/$ and China at 0.59 kg/$. 

Can the EU afford to go at it alone under these conditions? It not only can, it must, because doing so would inevitably influence all those planning to export to Europe. The European market, which is comparable in size to the American market and much larger than the Chinese market, cannot be bypassed.

The European market [...] cannot be bypassed.

Positioning Europe

It is in the context of this competitive recovery that the European Union could maintain its environmental advantage. To do this however, it is necessary that, strengthened by this new capacity to understand its environment through data, regulations be created to level the playing field and limit anti-competitive situations, as well as review industrial policies to favor those sectors that integrate externalities.

The issue of skills cannot be avoided either. This includes more coders, of course, but also more transdisciplinarity to deal with a consubstantial environment of complexity which is also massively multifactorial. This factor may seem secondary, yet it is essential. What is slow to emerge are university curricula that take this reality into account. We need schools for engineers, technicians, climate scientists with training in code and data use, but also data scientists with political training, and research chairs linked to multivariate models. It is at this expense that we can hope to understand and act upon the coming world in which, once again, the environment will be a central factor. 

 

Copyright : ALEX WONG / GETTY IMAGES NORTH AMERICA / Getty Images via AFP

 

See also
  • Commentaires

    Add new comment

    About text formats

    Commentaire

    • Allowed HTML tags: <a href hreflang> <em> <strong> <cite> <blockquote cite> <code> <ul type> <ol start type='1 A I'> <li> <dl> <dt> <dd> <h2 id='jump-*'> <h3 id> <h4 id> <h5 id> <h6 id>
    • Lines and paragraphs break automatically.
    • Web page addresses and email addresses turn into links automatically.
    • Only images hosted on this site may be used in <img> tags.

...

Envoyer cette page par email

L'adresse email du destinataire n'est pas valide
Institut Montaigne
59, rue la Boétie 75008 Paris

© Institut Montaigne 2017