Why don’t we use AI to solve the right problems?

When it comes to artificial intelligence, consumer technologies like Siri, Alexa and targeted advertising capture the public imagination. What do these AI applications have in common? They analyze preferences in order to influence behavior, and their contribution to humanity’s progress is limited. In my keynote at the United Nations’ AI for Good Global Summit earlier this week, I argued that industrial AI applications offer much greater potential for shaping a better planet than their better-known consumer counterparts. Below, I’d like to share some of the key points from my talk. I’m looking forward to a great discussion!

If research alone were the measure of human progress, one could say that we live in the best of times. Worldwide, there has never been so much investment in research and development as today. Investments in AI, in particular, have skyrocketed. In 2016, investments in AI in North America, Asia and Europe were in the range of $26 billion to $39 billion. That’s between one and two times higher than NASA’s annual budget. While comprehensive data on AI investment in 2018 has not yet been announced, one can safely assume that such investments have only accelerated. One example: last year, Shanghai’s city government announced a $15 billion fund for the development of AI.

“The rise of powerful AI will be either the best or the worst thing ever to happen to humanity. We do not yet know which.” Scientist Stephen Hawking spoke these noteworthy words in 2016. So far, we haven’t made much progress in resolving the issue he raised. We still don’t know if AI will deliver the greatest or the most horrible outcomes for humanity.

An inflection point

Today, in 2019, we’re at an inflection point for artificial intelligence. We can compare this situation with 1989, when geopolitical and technological disruptions of tremendous scope coincided: The Wall fell. The East-West conflict that we had experienced for decades was resolved. That same year, 1989, also saw the invention of the Internet, a technology that has had huge impact on our existence. In the meantime, digital technologies have fundamentally changed how we lead our lives.

The most pivotal of these digital technologies is AI. What are we using this powerful technology for so far? Here are some examples:

• Speech recognition: our living rooms are now equipped with fancy voice-controlled virtual assistants. We can ask them to play music, put together shopping lists, or read the weather forecast for us.
• Targeted advertising: AI can help personalize the advertising that we see – so that it can be fully tailored to our needs and preferences.
• Spreading trends: AI has now already brought artificial people to life – personalities created to influence our purchasing behavior. We can hardly tell them apart from real people.
• Relevant news: We’re no longer the ones who determine what news sources keep us informed. Instead, artificial intelligence prepares a selection for us. AI defines our news feeds in social media, in part by analyzing what it assumes to be our preferences and then using that information as a basis for decisions.

Capabilities like these surprise us, impress us and even amaze us. In each of these examples, AI analyzes our likes and dislikes and subsequently takes on a role in shaping our behavior. It knows us better than we know ourselves, so-to-speak. We’re being influenced. Some would even say we’re being manipulated.

Not on track to reach key Sustainable Development Goals

There is also another inconvenient truth: Despite the tremendous budgets invested in AI, despite the remarkable examples of machine-mind capabilities – we’re not making sufficient progress when it comes to solving some of our planet’s most fundamental challenges.

We already know what the real problems are: They’re defined in the United Nations Sustainable Development Goals (SDGs). The SDGs are the roadmap for shaping a better future for the generations to come.

Let’s look at a few key SDGs:

• SDG #2, “Zero Hunger”: World hunger is on the rise again. 821 million human beings were undernourished in 2017, up from 777 million in 2015.
• SDG #3, “Good Health and Well-Being”: Every day, 10,000 people die because they lack access to affordable healthcare.
• SDG #7, “Affordable and Clean Energy”: One billion people live without electricity. And if current policies and population trends continue, nearly 700 million people will still be without electricity in 2030.
• SDG #10, “Reduced Inequalities”: In 2018, the combined wealth of the world’s 26 richest people equaled the wealth of the 3.8 billion people who make up the poorer half of humanity.
• SDG 13, “Climate Action”: 18 of the 19 warmest years on record have occurred since 2001. At the climate conferences in Paris and Katowice, the world made political progress in fighting climate change, but achieving a drastic turnaround of our emissions is still pending.

Industrial AI: great potential for good

What if we solved the real problems in this world – by using the power of industrial AI?

• Imagine affordable and clean energy for all. Today’s AI solutions are already capable of adjusting the powerful gas turbines in electricity plants to optimize operation. To do that, they must perfectly balance a great number of parameters – such as combustion dynamics, fuel efficiency and emissions. As they learn, machine minds can become better at this task than even the most experienced human technicians. Similar AI technologies could be applied to wind turbines. Data analysis and predictive maintenance are already helping to avoid downtimes and lower the costs of wind parks. In the future, AI solutions could configure entire wind farms for maximum yield by taking complicated aerodynamic effects into account. Learning algorithms have the potential to make renewable power even more efficient and cost competitive.
• Imagine vibrant, livable communities with zero carbon emissions. Small- and large-scale power plants that use renewable sources of energy, a range of storage solutions (from power-to-gas facilities to batteries in private homes), electric vehicles, smart buildings, and many similar approaches can help achieve this goal. Yet running, stabilizing, and optimizing these intricate systems is nearly impossible for the human mind. AI can jump in and help us find the right answers to these complicated questions – and run infrastructure more smoothly and efficiently than ever before.
• Imagine a world with affordable, sustainable goods for all. Today, we’re capable of designing, simulating and testing products and even entire production lines with specialized software. And AI solutions can already enhance specific production steps: They can ride “piggy-back” on existing industrial controllers and boost their capabilities. In the future, AI will help entire factories self-organize – for instance, to reconfigure manufacturing processes to shift from one product to another, find solutions to unforeseen events, resolve interruptions or reduce the consumption of energy and other resources. All this will greatly improve the efficiency and speed of our factories. Put differently: AI will make products more affordable and eco-friendlier.
• Imagine enough food for all, without waste. As of today, 500 million tons of food a year are lost due to planning and handling issues. Another 350 million tons spoil each year due to inadequate storage conditions and time lags in transportation. In total, that’s 850 million tons of food waste –a payload equivalent to a line of large trucks that extends ten times around the globe! A.P. M?ller – M?rsk (APMM), the world’s largest container shipping company, is harnessing digital technologies to eliminate this kind of food waste. At APMM, 100 percent of the refrigerated containers are connected. They send data on temperature and humidity in order to ensure that food is kept at the correct temperature during transportation – thus minimizing food waste. Based on data from the containers, AI could help identify typical problems and delays in the shipping chain and propose logistical improvements. Such optimizations would not only help eliminate food waste, but also make the global trade of other goods more cost and energy efficient. APMM transports 25 percent of all refrigerated containers worldwide.
• Imagine cures for humanity’s most dreaded diseases. Today, we’re capable of creating a virtual 3D model of an individual patient’s heart. This so-called “digital twin” integrates anatomic, electrophysiological, biomechanical, and blood-flow data. It can be used to find the ideal position and setup for a pacemaker device, for instance – before and even during an operation. The great frontier in healthcare is to expand this “digital twin” and create a highly sophisticated and accurate simulation of processes in a patient’s body. AI will be the technological backbone. Sophisticated approaches like these will make it easier to understand health conditions comprehensively, predict the viability of therapies and plan customized treatments.

Making industrial AI an engine for progress

These examples provide but a glimpse of the tremendous potential that industrial AI offers for helping meet the SDGs. To realize this potential, we’ll have to find a good balance: If we use AI too little, the results will be insignificant. If we overuse AI, we may suffer negative consequences. Some of the key questions are:

• How can we make productive use of big data without sacrificing privacy?
• How can we create and operate powerful platforms without allowing them to become monopolists that abuse their market dominance?
• How can we reap the advantages of autonomous systems without losing control?

Many of these tensions can be resolved if we take the right decisions today, at this inflection point.

Making industrial AI an engine for progress has some important prerequisites on the operational side:

• Focus: We should focus future AI investment on industrial applications. Public research funding as well as public-private R&D partnerships could play a key role. And attracting talent to industrial AI development is just as important: Companies and universities should reach out to the public, especially to talented students, to attract bright minds to this exciting and promising field.
• Infrastructure: One way to accelerate industrial AI is to build state-of-the art 5G networks. 5G will make it possible to connect nearly everything in the real world to industrial AI – from cars, fridges and buildings to trains, traffic lights, wind turbines and energy grids. Once this connectivity has been established, AI can enhance all these real-world applications.
• Data: All forms of AI need data to learn and improve. The greater the quantity and the higher the quality of this data, the better AI will become at serving its purpose. This is also true for industrial AI. Public and private institutions could cooperate for this in fields that are of special relevance for social progress. We can ensure the availability of data for industrial AI (and avoid data monopolies).

All these prerequisites are about accelerating technological advances in industrial AI. Making AI an engine for global progress also requires the right political and ethical principles, however. These should include:

• Trust: AI development should be intertwined with privacy, security, and explainability from the very beginning. “Privacy” means that industrial AI should be developed, trained and operated exclusively on the basis of data that doesn’t infringe on personality and ownership rights. “Security” means that powerful encryption and security mechanisms are not to be introduced as an afterthought but rather integrated into AI architecture from the outset. “Explainability” means that AI should never be a black box. Developers shouldn’t just extend the capabilities of AI – but also expand our understanding of how AI arrives at decisions and where ethical deficits may arise. What is more, we need new systems of public oversight over the mechanisms and frameworks that govern AI in critical areas.
• Accountability: Another crucial requirement is the need to assign clear responsibilities. We can’t allow ambiguity when it comes to the question of who’s responsible for an AI application’s functions and possible malfunctions. We must prevent the delegation of overall responsibility!
• Enhancement: Industrial AI should enhance the work of human beings. A machine can easily solve complex optimization tasks, for instance. However, one needs a human mind – and public oversight – to set the framework for critical AI tasks: What would success look like? Which outcomes have priority over others? The more cooperative the interaction between man and machine, the more acceptable AI will be for our societies – and the more AI will contribute to social progress.

Positive impact on society will be a key differentiator

The OECD’s principles on artificial intelligence, adopted by 42 nations in late May, cover some of these prerequisites and principles. So do the Ethics Guidelines for Trustworthy AI, which were presented in April 2019 by a high-level group of AI experts set up by the European Commission. The impact of non-binding principles like these will depend on their implementation, of course. Nevertheless, they send a clear message: Shaping AI in ethical ways is not just a nice-to-have.

I am convinced: Going forward, positive impact on society will be the key differentiator for AI. And this is where I see a promising opening for Europe. So far, Europe has been a distant third in the global AI race behind the U.S. and China, both in terms of investment and in terms of technological breakthroughs. If we put our energies behind it, responsible, industrial AI could become Europe’s USP.

Why? Because Europe has exactly what it takes: a strong industrial base plus strong social responsibility:

• Europe has deep know-how in industrial processes, reflected for example in its world-leading output of science and engineering publications. This expertise is of great value for shaping industrial AI.
• Europe has a proven track record of balancing technological progress and social benefit. The General Data Protection Regulation (GDPR), for instance, has been a great step ahead for data privacy worldwide – with a positive impact far beyond the EU’s borders. Moreover, the EU has repeatedly shown its determination to curb the power of tech giants and investigate monopolistic practices.

All in all, people around the globe have an interest in Europe being a strong entry in the global AI race. After all, what’s happening now really counts: Industrial AI will reshape the critical infrastructure that enables modern life – from healthcare to energy, from transportation to manufacturing.

On the surface, the AI debate we’re leading these days is about technology. If we dig deeper, it’s about something much more important: the future of our societies and our planet.