The Highlights of our Thought Leader Interview with Dr. Andreas Liebl
on Leadership and the Rise of Intelligent Machines
What is the state of corporate AI adoption today? How might we build a competitive edge by testing and scaling innovative AI solutions? And what are opportunities and threats posed by the ever-increasing pace of development?
The following interview was conducted Christian Spier and Dr. Laura Bechthold as part of the Future of Leadership Initative’s (FLI)’s project titled “Yes, Mr Robot! Leadership and the Rise of Intelligent Machine”.
- The speed of AI development poses a crucial challenge for our society. If the consensus building system of our democracy is slower than the speed of development, we cannot keep up with regulation.
- We need to establish a culture of experimentation in which humans and intelligent machines can learn together and make mistakes with limited effects on the economy and society.
- For efficient AI innovation, we need start-ups to experiment with new technology use cases in an agile way and corporates to adopt and scale solutions.
- In the long run, AI value creation will reach a significant proportion and hit currently higher-paying jobs, whereas the human value creation will more lie in fields such as education, child- and elderly care.
- AI is good at solving problems that are too complex for humans to understand. We need to be able to trust real-time “black box” models but explain their actions retrospectively.
ABOUT DR. ANDREAS LIEBL
Before joining UnternehmerTUM, he worked at McKinsey for five years and received his doctorate from the Chair of Entrepreneurship at the TU Munich. In his doctoral thesis he accompanied 120 start-ups for one year and examined their corporate culture and identity.
WHAT ARE THE MOST DISRUPTIVE DEVELOPMENTS WHEN IT COMES TO THE RISE OF MACHINE INTELLIGENCE?
On the one hand, you have developments on the algorithmic side. There, we have seen many developments in recent years in the architecture and concept of neural networks such as LSTMs*, CNNs*, RNNs*, as well as reinforcement learning and GANs*. One of the key questions there remains how to be more efficient in applying those concepts to real-world tasks.
The second big area of advancement has been made with computational power enabled by customized hardware. With GPUs, TPUs**, and IPUs**, we have different types of processing units specifically designed for neural networks, especially on the model training task but also on the inferencing side.
The third one is data availability. It is definitely a large effort to for instance coordinate Industry 4.0 components with a multitude of sensors ready to provide data which we did not have a couple of years ago.
Beyond these three developments, we think that an overarching component of AI adoption is augmented reality because it is the interface between the digital and physical world. And if we have automated agents, there needs to be a way of communicating and interacting with these types of systems.
[Author’s note: *the above-mentioned technical terms refer to different types of neural network architectures: LSTM = “long short-term memory network”; CNN = “Convolutional Neural Network”; RNN = “Recurrent Neural Network”; GAN = “Generative Adversarial Network” **the above-mentioned technical terms refer to different types of processors, some of which are tailored to more efficiently perform specific machine learning tasks such as matrix multiplications. They are sometimes summarized as “Neural Processing Units”: TPU = “Tensor Processing Unit”; IPU = “Intelligence Processing Unit”]
LOOKING AT THE CURRENT PHASE OF FAST DISSEMINATION OF AI TECHNOLOGIES, WHAT ARE THE BIGGEST CHALLENGES FACING BUSINESSES AND SOCIETY TODAY?
If you look at the status quo of companies today, in most cases you have POCs (“proof of concept”). There are only few cases where you’ll find implementations of AI. So, the step from POC to an implemented solution is quite a hard one for them.
A second challenge which represents one of the foundational challenges of AI is the speed of development since the technology changes so fast that you need to ask yourself what your core USP really is. What you have worked on in the past three or four years could be obsolete next year already. So, you need to constantly stay on top of development which most companies cannot afford.
This topic of development speed not only poses an implementation challenge but also one for our society because the consensus system of democracy is simply slower than the speed of development and we cannot keep up with regulation. For the topic of “deep fakes”, for example, the ethical discussions take longer than the technology development which makes it extraordinarily hard for our society to tackle this challenge.
DO YOU FACE A THREAT OF INERTIA EMERGING FROM THE SPEED OF TECHNOLOGY DEVELOPMENT?
Yes, but I see this more in Western democratic societies than in societies like the Chinese one because they are more autocratic systems. They can really push these technologies and if we want to keep up with their speed of development, we need to find a way to more intelligently apply these technologies than we do now.
ABOUT THE PROJECT
SO WOULD YOU SAY THAT WE CAN LEARN TO A CERTAIN EXTENT FROM THE AUTOCRATIC SYSTEM IN CHINA WHEN IT COMES TO TECHNOLOGY ADOPTION?
Only to a very limited extent since there is no way for us to mimic this type of system. We need to find our own way of smart technology adoption because what works there not necessarily works well here. It behaves similarly to startups and established companies. Both have unique strengths and established companies should never try to mimic startups because they simply lack the speed and have different requirements from the beginning and we can transfer that to nations or economies. We need to identify the strengths and unique assets here in Europe to focus on them and integrate them into our innovation process to successfully adopt AI.
WHICH SOLUTION ARE YOU ENVISIONING FOR THIS CHALLENGE?
I think that you need to have a space in which you can experiment fast and can see what works and what doesn’t. Also, you need to have system in place to be able to scale what is working well.
Looking at the interplay between established companies and startups in our economy, you need the startups to experiment on a small scale which solutions are working well. But then you need to create an environment to really push the startups to become significantly large if they are doing well and the same holds true for our society. Ideally, we’d have the freedom to test assumptions before defining laws and regulations and I think that it needs to be a very agile process which you monitor very closely.
IMAGINE 10 YEARS FROM NOW: WHAT IMAGES AND MOST SIGNIFICANT USE CASES COME TO YOUR MIND WHERE OUR SOCIETY COULD BENEFIT FROM ARTIFICIAL INTELLIGENCE? WHAT COULD BE BETTER THAN TODAY?
I like to think about AI application in an analogy for how we work. What we did in the last technological revolution, is to adapt our work to best support IT systems. We broke down processes and defined them in such way that we can easily digitize and work with them to manage supply chains, whole processes and so on, by that creating work environments that are good for computers but not for humans. With the help of AI, we can turn this around to create workspaces that are designed for humans and tasks in which humans are good at.
On top of that, what we predominantly see in the Scandinavian countries is that they tie in AI innovation to the Sustainable Development Goals and thereby helping to solve fundamental problems such as fighting hunger or climate change.
With individual assistance for health, AI can help to provide access to medical support for underserved regions that have a low density of medical professionals. Individual assistance for agriculture can help you farm in the best way possible with given soil and weather conditions. So, AI can really act as a decision support system for people to perform best in their respective environments.
On the other hand, AI is going to take over a significant amount of the value creation in our economy. If we are smart about designing the system the AI operates in, we can increase economic equality and make sure that AI is benefiting us all.
HOW ARE YOU AND YOUR ORGANIZATION CONTRIBUTING TO SHAPE THIS POSITIVE FUTURE?
We try to have a very neutral perspective on AI. For us, AI is a toolkit and it is our obligation to use it responsibly. There are three pillars that we at appliedAI are pursuing to accelerate the adoption of AI in Europe:
Strategy – One core assumption is that every company has the same questions about AI and most of these are non-technical ones, e.g. about culture, processes, and methodologies. Since we are not targeting specific technologies with this pillar, the knowledge is in most cases not proprietary and we can therefore distribute it more efficiently to companies. We structure this process as an “AI strategy house” where we publish content pieces and also advise companies on their transformation processes.
Academy – Second, once you have an understanding of the core concepts, we help companies bring this knowledge into their organizations and help educate thousands of people within their organizations to get started. We do not focus too much on tech education since there is a lot of great content out there but rather on AI project management, strategic implications, and we also train domain experts on how they can leverage AI for their fields of expertise.
Engineering – The third part which we call “learning by doing” is about prototyping and engineering. In most cases it is not relevant for companies to create knowledge on their own because AI develops so fast that you cannot keep up if you don’t have experts in your team. A good solution for that is to collaborate with partners, in many cases AI startups, which is why we created Startup Landscapes and help the companies find the right partners for their projects. If companies want to build up on core competencies, we support them with joint development projects.
LET US MOVE TO MORE CRITICAL ASPECTS OF MACHINE INTELLIGENCE. LET US THINK AHEAD AGAIN AND ENVISION A WORLD IN WHICH AI HAS UNLEASHED ITS FULL NEGATIVE POTENTIAL. WHAT WOULD BE WORSE THAN TODAY? IN WHICH AREAS OF SOCIETY COULD MACHINE INTELLIGENCE HAVE A NEGATIVE IMPACT?
Again, I think that the main challenge is that the pace of technological change is faster than we humans are capable of changing ourselves. The dangerous scenario here is that this development could lead to an “elite” which is capable of working with AI and the rest of the population being left behind. If the value creation in the economy is increasingly driven by AI, the wealth inequality between these two groups can become enormous. Now, I do not fear that machines could become intelligent enough to rule the world as much as humans feeling left behind and civil unrest emerging between these two groups due to inequality rising to an extreme extent.
IN YOUR OPINION, WHAT ARE THE “WHY” QUESTIONS LEADERS HAVE TO ASK THEMSELVES TO PREPARE FOR THE AI REVOLUTION AND MOVE TOWARDS THE POSITIVE FUTURE SCENARIO?
Given that AI is going to transform the way we work and creates a lot of uncertainty about the future, it is a vital leadership task to provide visions on what we want to do with this technology and to provide room for testing out things and communicate what is working well to give guidance to people that are affected by this transformation. What I see at the moment in our society is a set of tasks that we set to achieve but without any shared goal in mind on what we want to do with this technology in the future.
IN WHICH ENTITY OF SOCIETY DO YOU SEE THE RESPONSIBILITY TO SET THE PACE AND MAKE SURE THAT NO ONE IS LEFT BEHIND?
In my personal view, I think in Europe the organization best suited for that is the European Commission and I think they are doing a good job with the high level expert group now working on the ethics guidelines on trusted AIs. They now start to form a communication agenda around this whole technology which is human-centered trustworthy AI and from that derive measures for companies to adopt which I think that it is a good start.
WHAT ARE THE MAJOR SKILLS, COMPETENCIES, AND ATTITUDES THAT YOU NEED IN THE FUTURE OF HUMAN-AI COLLABORATION?
We really need to establish a culture of experimentation in which humans and intelligent machines can learn and improve together. That implies that, at the beginning, mistakes will be made and we need to create environments where we limit their effects on economy and society. But only through that we can test and scale the application of intelligent machines and enable humans to perform the tasks in which they are sustainably better than machines, for example empathy, creativity, and interpretation of research. Generally humans are better than machines in those kinds of tasks where you draw implications of a full situation and assess it under the given circumstances. Interestingly, the jobs in which this skillset is relevant are in education, elderly- and childcare – jobs that are not as well-paid as those that require lots of domain expertise and are currently getting automated such as a doctor’s examinations for cancer or complex contracts and purchase agreements of lawyers.
WHAT IS YOUR RECOMMENDATION FOR YOUNG TALENTS HOW TO PREPARE THEMSELVES FOR A FUTURE WITH AI?
A basic understanding about these technologies is really helpful. I would also say that it is worth investing in learning the mathematical and statistical foundations, especially for those whose actual field of expertise is unrelated to math/statistics/computer science. For instance, we have psychology students using voice recognition AI to identify depression and I think that the next wave of innovation will come from domain experts transfering methodological AI knowledge to their own fields.