Why do AI projects fail? For example, because there is no real AI strategy in the company, because of poor data quality, or simply because expectations are too high. However, companies often disregard a key factor in their change process: change management. The AI experts Dr.-Ing. Akhauri Prakash Kumar and Ralf Hülsmann explain how the consistent involvement of employees contributes significantly to the success of AI projects.
Dr.-Ing. Akhauri Prakash Kumar: I would not go so far as to say that interest in artificial intelligence is already declining. On the contrary: we are only at the beginning of the AI career. The technologies now are more mature and the costs are manageable thanks to low-cost computing power from the cloud. But what we are currently experiencing is a certain frustration that is becoming widespread in practice. However, this has less to do with the benefits of AI or the lack of solutions, but rather with the way companies introduce AI. Artificial intelligence which bypasses people, i.e. which doesn’t involve the employees in the specialist departments, does not work.
Ralf Hülsmann: Artificial intelligence is nothing more than new software or a new machine that a company installs. These kinds of projects have also failed in the past if they were implemented from the top down, by a manager. The employees directly affected by the introduction of new technologies were not informed or were informed too late in the process, and they then switched to blockade. There are a number of projects that have clearly failed because people refused to cooperate. Companies should learn from this.
Dr.-Ing. Akhauri Prakash Kumar: It's in most people's nature. Change is often perceived not as an opportunity but as a danger. They are confronted with something new that they aren't familiar with, something that takes them out of their daily routines. And maybe it overwhelms them or, worse, costs them their jobs. And the notion of artificial intelligence still triggers a reaction among employees: there's a computer or a machine that will make me obsolete. It is completely human that this is met with little approval.
Ralf Hülsmann: This alone will provide only limited help. It also depends on who explains it and on what occasion. When a manager first explains to his team which processes can still be improved, which key figures do not quite meet expectations, and then suggests that an AI solution would solve these problems, it is already too late. To the employees this sounds like: you can't do better than that and therefore we are looking for a better solution. Then it doesn't matter how well you explain artificial intelligence, it won't change anything in the face of the existing skepticism.
Dr.-Ing. Akhauri Prakash Kumar: Among other things, the rationalization that this could entail. AI is taking over my job. They won't need me much longer. It is therefore quite specifically a matter of job losses. The aim is not to cut jobs, but to improve processes, service, and products and thus, ideally, to secure the company’s competitive edge. To do this, companies need the know-how of experienced specialists, without which artificial intelligence remains just a useless piece of computer code. AI projects therefore only lead to success when people are involved.
Ralf Hülsmann: Artificial intelligence needs experienced personnel to feed it. Experts have learned over the years to recognize certain patterns in their job. Artificial intelligence provides the opportunity to recognize patterns that humans cannot recognize despite their experience. Together they take the leap and try to find solutions that humans have not thought of or could not see before. But making the most of it requires the experience of experts.
Dr.-Ing. Akhauri Prakash Kumar: This is comparable to a game of chess, where a player is supported by artificial intelligence. The algorithm quickly derives a response to the opponent's last move. And not only is it faster than an experienced chess player. The chess computer also finds solutions that the player wouldn’t come up with. However, the solution proposed by the computer provides the chess player with time to contribute his experience. He knows the tactics of his opponent, his strengths and weaknesses. Maybe that's why he decides to make a different move. But what is certain is that they are stronger together. The purpose of artificial intelligence is not exchanging a human being for a machine, but expanding his skills and knowledge.
Dr.-Ing. Akhauri Prakash Kumar: Actually, it's quite simple. One factor for important success is being transparent from the outset and forming a team that reflects all roles, hierarchies, and tasks of a department, process, product, or service. Only this team will be able to determine which adjustments could possibly be made with an AI solution. Together, they develop starting points and define common goals. Experts in artificial intelligence and data analytics should then be available to answer questions around which AI solution and which data will be used, i.e. the technical side of the project.
Ralf Hülsmann: It is important to understand that AI applications – no matter how specific – can affect all areas of a company. This is why the involvement of employees from several areas is so crucial. AI touches on corporate strategy, corporate culture, technology, leadership, organization, and communication.
Ralf Hülsmann: Successful change management can only be implemented if there is transparent, open, and appreciative cooperation in the company. If a company still has an opaque prescriptive culture that is imposed by the management, i.e. from the top down, change management and AI projects cannot be successful. In this respect, the corporate culture is decisive for the success of AI projects. This means that it may be necessary to first change the culture and then think about artificial intelligence. This has been clearly shown in practice.
Ralf Hülsmann: Working in task-oriented teams, with agile methods, and with a collaborative approach are extremely good catalysts for artificial intelligence. People today are more involved in decision-making processes and are closer to the action. The error culture has also changed. If something doesn't work as expected, you don't look for the culprits, but learn from it and continue. This encourages entrepreneurial thinking and facilitates access to new ideas. When a team is given responsibility for an AI project, when they experience artificial intelligence themselves and see the benefits for themselves, people feel taken seriously and are more motivated.
Dr.-Ing. Akhauri Prakash Kumar: Top-down is the worst approach available. Especially in times when companies are changing from a matrix organization to a group organization. The important thing is that everyone can contribute their skills. I like to compare this to a symphony orchestra. Everyone plays their own instrument and contributes their individual skills. But only together can the musicians put on a great concert.
Dr.-Ing. Akhauri Prakash Kumar: In order to coordinate projects and bring experiences from one project to the next, this certainly isn't a bad idea. But it is very important that the person in charge of AI brings in more than just a financial point of view, otherwise AI projects will again be evaluated only from the top down, from the perspective of turnover and profit. The person responsible for AI in particular must ensure that change management is taken seriously and plays an essential role.
Ralf Hülsmann: Absolutely. Artificial intelligence needs clear governance, which must also be controlled. Otherwise, there is a risk that AI projects could get out of hand from a data protection perspective. After all, it is almost exclusively about data. At Telekom, we established guidelines for artificial intelligence at a very early stage. The keyword here is digital responsibility. The guidelines define how we at Deutsche Telekom want to deal with AI and how we will develop our products and services based on artificial intelligence in the future. We have also committed ourselves to transparency and disclose, for example, how we use customer data.
Ralf Hülsmann is a proven IT and AI expert. In the Strategy & CTO Office in the area of Public Cloud Managed Services at T-Systems, his current thematic focus is on the use of artificial intelligence in the cloud. Since 2010 he has been working as an Enterprise Architect at T-Systems in various roles on the development of the cloud portfolio, partnering and strategy.
Dr.-Ing. Akhauri Prakash Kumar is Head of Artificial Intelligence and Senior Sales Evangelist at T-Systems. With more than 35 years of experience in the semiconductor industry, Dr. Kumar also holds a bachelor's degree in mechanical engineering, a diploma in electrical engineering, as well as holding doctorates in artificial intelligence, electrical engineering, and mechanical engineering.
