Team works on a project - 3 reasons why AI projects fail

3 reasons why AI projects fail

Companies expect a lot from artificial intelligence: from higher productivity to increased automation and shorter time-to-market. There is no doubt that AI projects have the potential to fulfil these expectations. However, as with all newer technologies, there is currently no broad pool of experience that teams can draw on.

The risk of failed projects is correspondingly high in the field of artificial intelligence. This is confirmed by a study conducted by the International Data Corporation (IDC) in May 2019. For the IDC study, 2,473 companies worldwide were surveyed. Failures in AI projects were mentioned by most respondents. A quarter of the companies stated an error rate of 50 percent.

But why do so many AI projects fail?

This is a particularly important question for companies that are planning an AI project or are already in the implementation phase. The following three reasons often lead to AI projects not developing as successfully as expected or even failing.

1) Not enough investment in employees

Gartner, the world’s leading research and consulting company, sees the greatest advantage of artificial intelligence in the fact that employees will be able to dedicate themselves mainly to more demanding and essential activities in the future.

On the one hand, highly qualified employees such as data scientists are therefore required. On the other hand, all those employees who are involved in AI projects and have no previous experience in this field should also have “data literacy” and be further qualified. “Data Literacy” refers to the ability to read data, work with it, analyze it and use it as a basis for decision-making.

The different backgrounds and experience of employees should also be taken into account. So-called business and IT alignment is all about bringing together different experts and following a common path that fits the corporate strategy.

2) Strategy is not sophisticated

According to the IDC study, only 25 percent of the companies surveyed have developed an enterprise-wide AI strategy – even though two-thirds of them stated an “AI First” culture. This trend is confirmed by a study conducted this year by the market research and consulting firm Lünendonk & Hossenfelder. In this study, 33 CIOs, CDOs and AI managers from large companies and corporations were surveyed. According to the study, only one in four of the large companies analysed has a dedicated AI strategy or even a definition for AI. Not even every second of the few AI projects is in productive operation in the surveyed companies.

Without a clear goal and defined processes, it is difficult to meet the high expectations on AI projects. The reasons for a missing strategy can be complex – from a lack of skilled employees to silo thinking and reservations within internal structures.

3) Lack of data quality

One of the biggest challenges for many companies is data quality. It must be excellent, otherwise machine learning algorithms cannot be trained correctly. Moreover, predictions or recommendations for action would be correspondingly inaccurate or wrong if the quality is poor.

Outdated data, duplicates, incorrect or even missing information inevitably lead to the failure of AI projects. The first step in any AI project is therefore to clean up the data and bring it from differently structured or unstructured sources into a clean system.

Do you need support in organizing your data? Please contact us!