Artificial Intelligence and Machine Learning: Vision and Reality
Business Topics

Artificial Intelligence and Machine Learning: Vision and Reality

Today, artificial intelligence (AI) and machine learning (ML) are not just topics for discussion in professional IT circles. Sociologists too are busily examining the social implications of these new technologies; personnel experts are discussing whether the new technologies are good or bad for tomorrow’s working world, and jurists are asking themselves what legal consequences will result from the use of these new technologies. But what about the companies still waiting to implement these technologies?

Their view remains divided, a fact made clear by the results of the current Trend Study Customer Data Management. Eighty-three percent of respondents said that they don’t use any AI/ML methods at all, although 72 percent said that they view the future role of AI/ML as important for marketing, sales and service. By contrast, 23 percent did not (yet) dare to give an assessment.

What stands out in discussions about KI / ML is that – especially in Germany – only the FUTURE role of these technologies is mostly debated. In contrast, if one looks over the pond to the USA, or to the Far East, it’s clear that the use of KI / ML has long been viewed as "The Future is Now!" As a result, it is repeatedly reported that German companies risk losing their position in the field.

Operational Scenarios of AI/ML in Marketing, Sales and Service

In the USA, large technology companies in particular have been dealing with the new technologies for a long time. The search machine giant Google has been carrying out pioneering work for several years. As a rule, the wide mass of the public tends to follow the widely publicised projects, such as Google’s self-driving car, and overlook (or simply don’t know) that AI technologies were first integrated into Google Search with the Hummingbird update back in 2015 as part of the RankBrain project.

Other established operational scenarios are voice recognition systems such as Siri, Cortana, and Alexa, or the Supercomputer IBM Watson.

Nevertheless, some German companies are already using KI / ML methods. For example, Telekom uses specialist software for processing altered contracts and return bookings – thus replacing work formerly done manually. And Lufthansa provides a chatbot to help customers find the lowest air fare.

In addition to these application examples (where it’s mainly a question of automated processes replacing manual work), AI /ML methods are being used increasingly for creating prognoses of future behaviour patterns, based on experiences gained until now. In professional circles this is referred to as predictive analytics.

Based on historic data, algorithms create prognoses for the future development of business processes and customer relationships. In the field of marketing and sales today, predictive analytics already provide support for planning marketing and sales campaigns. 

Comparable processes are deployed in the financial sector for claims management and identifying fraudulent activities (e.g. money washing). Production and manufacturing companies already use the technology for (predictive) maintenance, whereby algorithms calculate the place and time of the next servicing or maintenance work, thus proactively warning of an impending breakdown of a particular machine or plant.

Obstacles on the Path to using AI and ML

And what hinders companies from implementing AI/ML methods? Here it’s worth looking back at the Trend Study Customer Data Management 2018 mentioned at the beginning. It shows the main reason for the delay being “missing knowhow”. Apparently – in Germany at least – companies lack the “skills” necessary for successfully implementing these new technologies. Generally speaking, and despite all previously mentioned operational scenarios, it seems that real understanding of the future technology is still missing. The study reveals the second most common argument for not using AI /ML technologies being that “there is no apparent need for/advantage to be gained from the use of AI /machine learning methods at this time”. 

These two deficits must be rectified as soon as possible, otherwise German companies will lose ground in international competition.

And one last important aspect can also prove to be a "brake" on the use of KI / ML technologies: The quality of a company’s customer data. Here we must return to the predictive analytics procedure mentioned above; the algorithm “learns” to create example prognoses for future sales and marketing campaigns - using the company’s own customer data – but irrespective of data quality. Therefore, poor quality (obsolete, incomplete, imprecise, etc.) data will cause the algorithm to “learn” from inaccurate material – and thus create an equally inaccurate prognosis. If the person training the algorithm is unaware of the poor quality of the data, they will trust the forecasts - even though they are wrong.

It’s here that the results of the Trend Study Customer Data Management 2018 underline the deep gulf and uncertainty currently prevailing in German companies faced with this topic: Forty-two percent of survey respondents saw a close relationship between the use of AI / ML technologies and the quality of customer data, with 40% failing to assess it at all.

Conclusion: The use of AI /ML methods for marketing, sales and service is not a long-term vision – it is reality in most application scenarios today. Businesses waiting to address this issue should do so as soon as possible and build the skills and qualifications they need.

The relationship between correct forecasting and the quality of customer data plays a central role, particularly in the area of predictive analytics, and this presents an additional challenge. However, the advantages for one's own competitive situation of knowing the future buying behaviour of customers based upon correct prognoses, make tackling this challenge more than worthwhile.