Will Artificial Intelligence save the day?
23 JUL 2022
Over the last couple of years, the research field of Artificial Intelligence (AI) has made impressive progress. Automated translation services, voice-controlled digital assistants, self-driving cars, world-class algorithms and other novelties have long made it to headlines – and our homes.
Driven by curiosity, scientists from a wide range of disciplines continue to push artificial intelligence ever further. Could future developments in this field be the solution to some of our most difficult challenges?
Looking back, before looking ahead:
The idea of machines being able of thinking and learning has fascinated scientists since the mid-20th century, when Alan Turing raised the question of whether a machine might someday achieve the same level of intelligence as a human being in 1950. With the first introduction of robots in 1969 and a computer program defeating a world chess champion in 1997, Alan Turing’s question suddenly seemed less far-fetched. The basic concept behind these advances has since been applied to a wide variety of other problems and has been successful in applications such as voice recognition for digital assistants, translation software, and self-driving cars.
Only ten years ago, the use of deep neural networks marked the beginning of the recent wave of interest in machine learning and artificial intelligence. Computers modeled on the human brain – so-called neuromorphic computers – have the potential to act as a catalyst for future AI and machine learning applications. While artificial and natural intelligence will gradually converge in future, the limitations of this technology have also come to light: AI today still needs enormous amounts of training data and energy – and it can be easily fooled. At the end of the day, human characteristics, such as their inquisitiveness, their courage to innovate and their curiosity, are difficult to replace.
The bottom line is: AI and intelligent machines are supposed to serve the needs of humans – and not the other way around.
AI will become higher-performing – and more affordable
Image classification is a key concept of AI, and it’s used in a variety of practical applications. In the smart factory, image classification helps to spot production defects, self-driving cars use it to identify objects in their surroundings and doctors can detect tumors more reliably. The Artificial Intelligence Index Report 2022 found that since 2018, training times for an image classification system have improved by 94.4%, while the cost has decreased by 63.6%. This tremendous development can also be seen in other task categories such as object detection and language processing, enabling a more widespread adoption of AI technologies in a commercial frame.
One important aspect to keep in mind when talking about AI concepts like this is a phenomenon called algorithmic bias. One great example of this surfaced in 2019, when researchers found an algorithm used on over 200 million people in the United States. When predicting which patients were likely to need additional medical care, white patients were heavily favored over black patients. By now, luckily, the bias has been reduced by around 80%, but it underlines the importance of minimizing biased data and researchers. If we want to use AI to support humankind in their day-to-day lives, it is key that we feed the algorithms with diverse data sets to prevent biases that stem from programming and data sources.
We know that digital business models must be ethical. Consistent with our efforts of developing and adhering to rigorous bioethical standards, we have developed our Code of Digital Ethics to serve as the fundamental guideline for digital ethical questions and created our Digital Ethics Advisory Panel. We know that as early as in the developing stage, a number of different perspectives is necessary to minimize biases and to really create algorithms that can be applied in our very diverse reality.
Tackling today’s challenges with AI
At EMD Electronics, we see the opportunity to leverage AI to tackle supply chain challenges - of today and tomorrow - on a global scale, as these are some of the main difficulties for the semicondutor industry. With Athinia, the collaborative data analytics platform introduced together with Palantir Technologies Inc. in 2021, we allow device manufacturers and material suppliers to come together and exchange data safely. This collaborative approach can unlock new insights, for example regarding quality or processes, providing a new level of supply chain transparency and potentially sustainability. In this context, Athinia helps companies to drive data-based decision-making by leveraging AI and machine learning.
The benefits of this platform become clearer when looking at a concrete example: The chip shortage has effects on a number of industries, while the automotive industry is arguably most affected. Traditionally, each company along the supply chain focuses on optimizing their own processes, keeping a lot of the knowledge and expertise internal. However, due to the growing complexity of the industry, it is more necessary today than ever to look at the interactions of entities along the supply chain rather than single organizations. Bringing datasets from manufacturers and suppliers together, parameters that could impact the performance can be revealed, shortening time to market significantly. By ensuring no disruptions along the supply chain as well as providing the necessary insights into material optimization, Athinia can support manufacturers that currently are simply out of capacity.
Collaboration is key
Another example for a collaboration to make use of AI is the "AI Quality & Testing Hub", a project between the state of Hesse, Germany, the VDE and TÜV. This Hub, which is supposed to be established later this year, aims to bring together research, development, standardization, testing methods and infrastructures, and experimental spaces under one roof, while also offering a space for dialog and discourse.
I am convinced that collaboration between politics, science, and academia as well as corporations is key to truly leverage AI to its full potential.
Therefore, I highly appreciate that Hesse is pushing the idea of an Artificial Intelligence Agenda – Hesse is the home of our Merck KGaA, Darmstadt, Germany headquarters. The goal is to combine innovative strength with the state’s core values to develop AI responsibly and in compliance with data protection. To pursue this, the state has published an AI Agenda and identified five fields of action:
- Promoting AI innovation and AI applications
- Strengthening AI research and teaching
- Generating interest in AI and strengthen AI competencies
- Developing AI in smart government
- Sustainably developing computing infrastructure for AI
With the TU Darmstadt being one of Europe's leading universities in the field of AI, the Financial Center Frankfurt and strong pharmaceutical, mobility and logistics sector, Hesse is definitely an interesting and highly dynamic AI ecosystem. It is pleasing to see different initiatives to further support this key field of research. As an example, the state of Hesse has recently announced an investment of ten million euros in the establishment of an AI innovation lab at the Hessian Center for Artificial Intelligence hessian.AI. With our many years of experience in AI research as well as application in various fields across Healthcare, Life Science and Electronics, we are ready to contribute knowledge and expertise in future collaborations. I believe that AI has enormous potential, and while we should not rely on it to save the day, I am excited for what this technology will unlock in the future – for businesses as well as for society as a whole.