Artificial intelligence can make organizations even more efficient, technologically advanced and faster. When AI takes over routine tasks or even delivers new insights, it becomes a superpower. Five areas are important for the successful implementation of AI projects. With the right steps, introducing AI is not difficult.
Example: Sabrina works in the pharmaceutical industry. Each week, she receives an e-mail list of several hundred clinical trials. To make an initial assessment, she needs to open the link to each and every trial and review the summary and other parameters on their respective website.
At first, she really enjoyed doing it because it provided her with an overview of current research. But in the meantime, she is not so sure anymore that she is really able to identify potentially relevant trials. In addition, the pressure to provide a faster assessment of the publications is increasing and there is hardly any time left to read the studies in more detail.
Luckily, she now has an AI solution to assist her. It automatically analyzes all incoming information based on author, ingredients, medical conditions or therapeutic success. Keywords defined by experts serve as filters for excluding irrelevant contributions. This pre-screening gives Sabrina more time to thoroughly review the substance of the studies. It enables her to make better suggestions and the positive feedback from the specialist teams makes her work more enjoyable.
This real-world use case from a pharmaceutical company shows that AI can provide meaningful support and benefit for all stakeholders. Both the workforce and the company profit from it. However, before this can happen, a number of challenges need to be overcome. These concern five areas in particular: Data, technology, organizational processes, professionals, and legal & ethics.
Organizations can only realize added value from AI-based use cases if the right data is available in the right format. This is relatively easy for a specific use case, where AI can be implemented quickly and efficiently. However, in order to cover multiple use cases and transform into a data-driven organization as a whole, a comprehensive data strategy is required. It also needs to be taken into consideration that the search for the right data is not a one-time event but a continuous process. An adequate data management and governance framework, defining roles, processes, guidelines, etc. for the organization as a whole, is needed in this respect. A data strategy will provide a sustainable basis, not just for deriving use cases but also for deducting business and digital strategies—and hence the business model as well as in-house processes—in a data-driven approach.
Functional artificial intelligence technologies have been available for many years and are constantly evolving. Pre-trained AI services, for example, offer ready-made solutions for focused improvements in specific areas of application such image, text, and speech recognition, sentiment analysis, anomaly detection, and personalization. If these services are not sufficient, machine learning and data science platforms can be used to create custom models and data pipelines. You don't always have to start with building an entire machine learning platform with plenty of new hardware in your own data center to take the first steps. Organizations should look for ready-made AI applications that are available in the market and meet their needs in order to make their business processes smarter by using the right tool stack.
In any case, AI solutions need to be integrated into the organizational processes as they only generate a benefit if the results of an AI process are actually processed. In this context, it needs to be clarified who should process the results and in which applications they will be used. It is therefore important for companies to define who will be responsible for the AI-based process, who will develop it and where it will be operated. The management level needs to actively support the necessary activities by introducing agile processes and a change management process.
Different roles and specialist departments all play a part in the development and use of artificial intelligence. These include executives, data scientists, software and data engineers, as well as domain experts from the business functions. Data scientists and software engineers in particular are currently in high demand and it is essential for an enterprise to ensure that the different roles perform optimally by, for example, making sure that data scientists are not used for data preparation or project management. If these job profiles are not available in an organization, in-house AI skills need to be developed or leveraged by working with partners.
Compliance and data protection in particular constitute another challenge. Organizations need to verify which data can be legally processed, which actions need to be taken for being permitted to use the data, and how sensitive data must be anonymized. Specific rules for regulated industries and the transfer of data to the cloud are another aspect to be considered. In addition, the works council should be involved, especially if people fear for their jobs. Ethical issues concern aspects such as incorrect recommendations or possible discrimination as a result of implicit prejudice caused by data bias or insufficiently verified data entries.
All five areas deserve consideration if AI applications are to be successfully implemented and tangible AI use cases need to be identified across the different functions. This will ensure that the automation of day-to-day everyday tasks will provide clear benefits for the workforce. But before resolving all these questions, organizations can already start by drawing up a data strategy as well as choosing and implementing adequate technologies. It is advisable to set up an AI center of excellence or a business app center of excellence for the central management of the project.
AI can significantly impact and drive the digital transformation process. The major challenge is the generation of relevant data points, the management of the amount of data and excellent data quality. If you want to use AI tomorrow, you should start today by developing an adequate data strategy and choosing the right technologies.
For more information on the five most important areas and the actions to take for introducing AI, see the recently published Artificial Intelligence Experience Report by Campana & Schott and the Technical University of Darmstadt. The report describes concrete use cases in marketing, sales and product management and outlines approaches to solutions. AI will help you understand your customer, generate sales, and enable you to offer the right product at the right time.
Additional aspects and solutions concerning data and technology will be covered in additional best practice articles to be published in the coming weeks.
