The organisation of tomorrow will be built around data using emerging technologies. Big data analytics empowers consumers and employees. This will result in real-time decision making and a better understanding of the changing environment. blockchain enables peer-to-peer collaboration and trustless interactions governed by cryptography and smart contracts. Meanwhile, artificial intelligence allows for new and different levels of intensity and involvement among human and artificial actors.
When big data analytics, blockchainand AI are combined, it will change collaboration among individuals, organisations and things. When implemented correctly, these technologies can significantly improve consumer engagement, increase transparency, reduce costs and improve production efficiency or service delivery. Organisations can now leverage data and embed learning in every process. As such, data-driven organisations will experience great opportunities to deliver personalised products and services and remain competitive.
However, with great opportunities come great responsibilities. The organisation of tomorrow should be aware that there should be a balance among the technology used, the data analysed, the products created, the impact on the environment and the profit pursued. Without such balance, relentlessly pursuing shareholder can quickly damage your customers’ privacy and rights as we have so clearly seen with Facebook.
A data-driven approach is extremely powerful, especially with the right tools to use that data. However, it should not be at the expense of your customers, your employees or the environment. Here are three reasons why the organisation of tomorrow comes with great responsibility and how to ensure that all stakeholders benefit from your data-driven approach:
Algorithms are black boxes. The more advanced they become, the more difficult to understand the reasoning behind decisions. This would not be a problem if algorithms would not make any mistakes. Unfortunately, algorithms are created by biased humans, and most algorithms are trained with biased data. AI preserves the biases inherent in the dataset and its underlying code, resulting in biased outputs that could inflict significant damage.
AI is not immune to the ‘garbage in, garbage out’ imperative and organisations should prevent the use of biased data when dealing with AI. Biased training data can negatively affect business outcomes, which is why data governance is important to ensure high-quality, non-biased, data.
When AI can collect unbiased data, for example, through unbiased sensors, unbiased automated decision-making becomes possible. However, unbiased data alone is not enough, as algorithms are also affected by the (cognitive) Bias of human designers. Developers who create algorithms are, by definition biased. Whether it is the company culture, their upbringing, their education, or where they live. All these variables influence a person’s character and with that their work. In addition, people tend to suffer from cognitive biases — that is, they look for what they know (focusing on data that reaffirms beliefs), see patterns in data in which none exist (due to the illusion of understanding), ask the wrong questions (and ignore evidence) and overestimate their knowledge (resulting in tunnel vision). Even more, Bias can also appear in the selection and tweaking of an algorithm by the data scientist, even when the machine learning algorithm did not require any data.
Consequently, achieving unbiased algorithms is highly challenging, especially when the rationale of the algorithm is not clear. However, having a multi-eyes principle when writing the algorithm as well as being aware of the possibility of bias can help a long way.
