Eliminating the Disappointment of Artificial Intelligence

By Steffen Kuhn, Global Practice Lead Innovation & Special Assets /Head of Digital Engi-neering Center/Managing Partner and Manuela Mackert, Chief Compliance Officers, Deutsche Telekom [ETR: DTE]

Steffen Kuhn, Global Practice Lead Innovation & Special Assets /Head of Digital Engi-neering Center/Managing Partner and Manuela Mackert, Chief Compliance Officers, Deutsche Telekom [ETR: DTE]

Artificial intelligence, a buzzword that is currently of great concern to the economy in the course of digitization, is on the advance and various technological hurdles seem to have gradually disappeared. This is one reason why companies are increasingly asking themselves: “How can I use this technology profitably for my business?” In order to prevent disappointment and frustration during the implementation phase, it is imperative that the business perspective, customer perspective and ethical issues are considered in the overall view, in addition to the technological possibilities.

The use case must be clearly outlined, and an economic assessment must be done. An over-arching strategy helps to classify the technology and can lead to further synergy effects. We now know far over 50 percent of data, AI and IoT projects fail because at least one dimension has not been considered deeply in advance.

However, the central question that companies should always ask themselves before investing money in expensive data specialists and infrastructure is still too rarely asked at the be-ginning: “What should the collected data be used for and how should it be prepared for this purpose?” Answers to these central questions are the way to put quality before quantity. A comprehensive consideration of the individual relationships with each other, from end to end, is essential for the following steps.

In an economy that is growing increasingly data-driven, it is important to face the transformation to a data-driven enterprise and clearly evaluate the benefits and added value of da-ta. It is important to analyze and evaluate the data processes within the enterprise at an early stage by defining which data is needed, which data is available internally, or where it is necessary to access external data.

With Data Thinking, Detecon has created an overarching framework that covers all relevant aspects. From observation of the market environment, analysis within the company in order to develop new, data-driven applications to prototype development, test phase, and introduction. The needs of customers and users are identified and creative solutions for data-driven challenges are modulated.

The early integration of data and AI experts and the correspondingly developed data-based methods Canvas, and checklist-templates ensure that there are no unnecessary breaks and that nothing is forgotten. The continuous data reference ensured by the technical expertise enables relevant trends to be identified at an early stage in the conception of solutions and thus the latest technological approaches and standards to be incorporated. The participants benefit from new perspectives, which create out-of-the-box ideas that give the company innovative access to data-oriented solutions.

The technological feasibility is validated by proof-of-concept, which show the business impact of the developed solutions and thus prevent misguided investments at an early stage.

Prototypes are the focus here, as they illustrate the usability of the solution approaches to complex problems. With the help of the data-thinking methodology, this process is no longer a black box! By integrating different departments, continuous proof-of-concept, and an exact reflection of data relevance, a very accurate assessment of the ultimate success of a data-oriented digitization strategy can be made.

In addition to the technological and economic aspects, it is also important to take ethical aspects into account. They usually address the unspoken core of customer requirements. Digital ethics is not a mainstream or marketing aspect. Only if customers have confidence in the products and services, they will buy them and recommend them to others.

It is about clarity, transparency, security, responsibility and trust. Deutsche Telekom develops and uses AI to exploit the benefits of this technology (process optimization, abuse detection, improved customer satisfaction, etc.). However, this also holds the challenge of using the AI according to the rules and avoiding, for example, an independent existence of one´s so-called “black boxes” of the AI.

Against this backdrop, AI guidelines for Deutsche Telekom's AI processes and products were developed more than two years ago - incidentally as one of the first companies in Germany and the world to do so. They are based on Deutsche Telekom's business model and on in-depth discussions with internal and external CI experts, our employees, customers and various representatives of our civil society. We represent the contents of our AI guidelines as a company so that our customers can continue to place their trust in us.

It is therefore of great importance to us that customers and users are enabled to handle new technologies on their own responsibility. Programmers and technicians who deliver and improve these technologies need to do so responsibly and know what to look for

In addition, self-learning systems need defined and initially maintained limits within which they may act. For example, the AI guidelines define who is responsible for which AI system and function.

With the AI guidelines, a valid foundation stone was laid for dealing with AI at DT. But as with a Business Code of Conduct in the analogue world, further steps, regulations, and processes are needed to further formulate the topic and implement it sustainably. For quality assurance the internal processes are used by means of a test matrix with points for the consideration of new AI products. Furthermore, new AI products and services are tested for compliance with the guidelines in our Privacy and Security Assessment (PSA). AI projects are advised directly. An internal seal of approval is awarded for the successful assessment. A first business impact of the initiative can already be seen: For example, cooperation’s with health insurance companies and pharmaceutical companies that pay attention to secure and confidential infrastructure and that have approached us via the topic of “Digital Ethics.” In addition, the sustainable implantation of a “Corporate Digital Responsibility” makes a positive contribution to increasing sustainability ratings.

One project worth highlighting here is the creation of a modular end-to-end product and solution offering (Conversational AI Suite) based on “best-of-breed” partner products in the field of AI. In pilot projects with our business customers, different partners and solutions are implemented and tested. AI also helps our digital assistants to improve recognizing user questions. In return, it learns from past conversations how users formulate certain intentions and can therefore better respond to additional information. The digital assistance complies with the self-binding guidelines of Deutsche Telekom for AI as well. The core here is that customers always know where they stand and whether they are chatting with a machine or a person.

In conclusion, it can be said that only a structured overall view of technology, business environment and ethical aspects, from the early project phase to the rapid testing of the solution, will help many companies to find the right AI solution for their company and to secure customer loyalty.