When will artificial intelligence surpass human intelligence?
Humanity is undoubtedly on its way there, but we are still far from the point of superintelligence.
So there is no date yet.
Even the experts are very divided about when and whether this will be achieved at all, because both the hardware and the software will have to develop significantly further for this to happen. In this respect, superintelligence currently only exists in Hollywood films and science fiction novels. Just as there is no so-called general AI that can adapt to a variety of different things. It is different in the case of weak AI, where AI is developed and trained for a very specific purpose, such as playing chess or Go or driving autonomously. Today’s technology can already do that. There will be huge advances here in terms of better products and productivity increases. However, Germany must be careful that it does not completely lose touch, as it did when the internet was invented.
You are at the forefront of this with IBM Watson. What can your AI do in contrast to other AIs from Google, Amazon, Apple or Meta?
IBM has always played a pioneering role. For example, in the last century, the Deep Blue chess computer was an epoch-making system that won against the then world champion Gari Kasparov. Then there was the IBM solution used in the American television show Jeopardy: here the AI was able to answer the very complicated and diverse questions from various fields of knowledge faster and better than the highly intelligent contestants. This was an AI application in the field of natural language processing. So today IBM has a wide range of AI tools and services with which they can implement a large variety of AI projects – in language processing, in machine learning, in deep learning.
And how is IBM different from the other big tech players mentioned?
Everyone has developed certain strengths in their field. IBM is strong in research, among other things: more than 3,000 researchers around the world work every day at IBM Research on technologies for the world of tomorrow. IBM Research offers a strong pipeline of innovative artificial intelligence and machine learning tools and applications that feed into the Watson portfolio. Take the Debater project: it is the first AI system that can debate complex topics with humans. The goal is to help people build persuasive arguments and make well-informed decisions. The AI is able to debate at a high level – that is, to develop and exchange arguments. Key capabilities of Project Debater have been integrated into the Watson portfolio.
What topics are you currently concerned with? And which industries do you see as the forerunners in terms of AI?
The topic of language processing is very much in vogue right now. Many companies are working on the automation of FAQs. The technology that is used here are so-called smart assistants. If they answer text questions or chats, they are chat bots, if they also have a voice, a text-to-speech service on top, they are voice bots and if you also connect document databases, they are information bots. Or a combination of these. All this belongs to the field of intelligent or smart assistants. Think of insurance companies, for example: Getting the most important questions answered or making changes to contracts without personal contact. These things can be automated.
From the media discourse we know AIs that scan CVs, machines that analyse voices and also bot application systems. What can an AI do to distil the most suitable candidates?
Of course, digital tools can make it easier to match job supply and demand. In the past, this was done with a simple keyword search. With increasing specialisation, this is changing. You might be looking for a software developer in a complex environment who knows exactly three special programming languages, two libraries and two foreign languages in a certain country – so you have a combination of X criteria. And now you want to find the one that best fits your requirements among thousands of CVs, possibly in different languages. Here, AI can help to find the right person much faster. In addition, there are digital diagnostic tools that are becoming increasingly automated and better, thus creating greater comparability on the candidate side.
What solutions are available in the field of HR analytics?
HR Analytics goes far beyond what I have just explained. For example, at IBM, employees receive suggestions on what training might be relevant based on their job profiles. Workforce diversity also plays an important role at IBM. We actively promote this because we believe that we are more successful with heterogeneous teams than with homogeneous groups. Of course, this requires a good analysis.
What about the ethical aspects of AI-based selection and assessment systems? With regard to alpha and beta errors in applicant selection, can AI be used reliably here? Or is an algorithm just a reflection of the distortions of reality?
Recruiting analytics must be designed as a perfectly coordinated interaction of a wide variety of parameters and key figures. One example is the AI-based application IBM Watson Recruitment. An algorithm calculates the “success score”, the probability of how successful the applicant will be in his new position and how well he fits into the company. Influencing factors are an applicant’s skills, education and previous roles, their duration and the companies they have worked for. But data from current employees who are successful in a similar position is also relevant. The algorithm has to be adapted specifically for each company and checked again and again for possible biases.
Ethics and AI – what role does it play?
IBM’s approach to trustworthy AI puts ethical principles and people at the center respectively. We see a fundamental responsibility as a company to promote trust in technology. It is about ensuring that users know what AI does, that it is transparent, explainable and robust against disruption, that it can withstand cyber-attacks and, in particular, that it is also fair, i.e. treats all people equally. To ensure this, there will be the so-called Artificial Intelligence Act from the EU. If we in Europe manage to become known for trustworthy AI, this will be a competitive advantage comparable to “Made in Germany” in mechanical engineering.
How pronounced is the acceptance of top management when it comes to AI use?
Of course, large companies have had their own departments and teams working on this for years. But there are now also many medium-sized companies that are considering how they can improve and supplement their products and processes with AI.
What skills do decision-makers need today in order to choose the right solutions, handle the tools correctly and then derive decisions?
I consider openness to modern technologies to be one of the essential competences of a modern leader. It is important to be able to roughly classify tools and techniques in terms of what they can and cannot do. At IBM in Munich, for example, there is an experience center where we constantly receive companies that can look at more than 30 AI applications there and then get a feeling of what is possible and what is not. Secondly, a manager must be able to create a culture of innovation, support new ideas, give space to projects. It’s also about investing in staff training. And it’s about collecting, providing and understanding data. Because without data, there is no AI. Strengthening all these skills and then thinking in terms of processes, techniques and applications, prioritising project ideas according to feasibility and investment volume, and then investing in implementation – that’s what matters.
Where does Germany stand in international comparison in the areas of AI research and AI application?
Germany is traditionally good at research and has many globally leading universities, be it the TUM in Munich, the KIT (Karlsruhe Institute of Technology) and the DKFI (German Research Centre for Artificial Intelligence) in Kaiserslautern. We are top in the number and quality of scientific publications. The problem is the transfer from science to practice, the translation of ideas into products and services. High potentials often go abroad, where they sometimes have more opportunities, are better paid and are better supported in start-up projects.
What can be done about it?
We are getting better. For example, venture capital AI funds have been set up here in Germany in order to be able to keep up with the international competition. With the support of the Quandt family, TUM Munich has created the Unternehmer-TUM, one of the largest AI hubs in Europe, from which numerous IPOs have already emerged. This is absolutely exemplary and must be further expanded.
How can AIs like Watson contribute to a better world and sustainable living? And how do you link the two central fields of action – AI and strategic sustainability goals?
Even when it comes to sustainability, there is no getting around data and data analysis. Without a robust data base, it is not possible to analyse where the organisation or suppliers stand, intervene in the right places or produce a verifiable sustainability report for investors. While it is necessary to set up a data- and science-based sustainability strategy, where we as IBM support with data-based decision models, most companies are still struggling with the burdensome reporting, new regulations and obligations for disclosure and regular reporting. This blocks the view of the opportunities that lie in the topic of sustainability.
Which utopia do you consider realistic with regard to AI? To be striking, based on two film AIs that would probably easily pass the Turing test: An “Ava” from Ex-Machina – manipulative, calculating and ingeniously emotional, surpassing humans? Or a “Sam” like in the film Her – just as affectionate, but never with the hybris of wanting to be superior?
These are typical examples where the possibilities of current technology are overestimated. AI is not yet that far advanced, and our brains function quite differently. In this respect, I believe that an Ava from Ex Machina, a humanoid woman who is much more intelligent than all humans, is a thing of the future. If at all, it will be a very long time before that is possible. But on the other hand, there are already many ways to use AI in the company today. These include, for example, applications in customer care, risk management, IT, planning or supply chain management. And the opportunities to do that today are massively underestimated. We talked about this. What AI will make possible will give companies massive competitive advantages. That means there is no way around dealing with it intensively.
Wolfgang Hildesheim holds a doctorate in high-energy physics. He started his career at CERN and DESY. After more than ten years in research and consulting, he took on a leading sales position in Big Data & Communication Intelligence. Hildesheim has been working for IBM since 2007. He first headed the Automotive, Aerospace and High Tech Practice, from 2009 he led the Big Data Industry Solution Business in Europe and since 2012 he has been responsible for the development and IBM’s Watson, Data Science and Artificial Intelligence Business in Europe with a focus on DACH, where he implements projects with clients such as BMW, Deutsche Bahn, Telekom, Fraunhofer and HUK. Hildesheim is also a member of the German government’s AI steering group, Bitkom representative for AI standardisation issues, author and key note speaker.