Artificial intelligence allows humans to focus on the tasks that really matter. Rather than replacing us, AI can take care of steps in the chain of inference, predictions, decisions and subsequent actions that are too time-consuming, tedious or difficult to do at scale for the human brain. To make AI work effectively and ethically in any setting, it must be paired with organizational, cultural and human capabilities. In this article we will introduce three elements from the DAIN Data & AI Maturity Model (DAMM); what we call the “human” enablers.
This is Part 3 of our four-part article series explaining how we approach and advise our clients on their data & AI transformation journeys, introducing the DAIN Data & AI Maturity Model.
- Part 1 introduces the DAIN Data & AI Maturity Model (DAMM).
- Part 2 dives deeper into the strategic impact drivers.
- Part 3 (this article) details the organisational and human enablers.
- Part 4 gets to the core technologies of data & AI transformations.
Building the Organization and Culture
Data transformations do not happen overnight; they require changes to an organization, its culture, and its ways of working. An important aspect is that data analysis is not only “data science” reserved for the data team but a change in how decisions are made by thoroughly analyzing data. To underline that, some organizations have even opted to use the term “decision science” in place of data and analytics.
To be successful with decision science, a close interaction between business and technology stakeholders is a must. The data team members need to understand the business they are dealing with. Similarly, the business stakeholders should have a reasonably good understanding of the concepts of data science, especially in how to apply data use cases for their business area. Having this understanding on both sides helps with the stakeholder interaction considerably.
Since businesses, corporate culture, and skills differ from company to company, there is no one size fits all solution to organizing your data teams. While a center of excellence may be the right structure in certain settings, its role and influence must change as the organization is becoming more “data mature”. At the ideal stage, it is important to design the organization with clear purpose and structure, to have the right competencies available in the right teams dispersed throughout the entire organization.
Apart from improving the understanding between stakeholders, formal governance processes need to be implemented to ensure decisions, responsibilities, and escalation paths are clear for everyone. This model is called an operating model for Data & AI development, and its purpose is to ensure all enablers and impact drivers described in this model are working seamlessly in practice. This is a broad topic and we will cover it more thoroughly in a later chapter of this article series.
The purpose of a good data culture is to ensure that the investments put into improving the overall data maturity are utilised by the organization to the full extent. Coming back to the term decision science, a good data culture means that data is sought to back decisions, whenever relevant data and analysis can be made available. The organization should transform from gut-driven to data-driven.
Data and AI means new skills
Working with data and advanced analytics are new disciplines for organizations. With new disciplines arrive also new demands for human skills. While organizations typically already have internal or external skills for managing data, data warehouses and report development, managing greater volumes of data and advanced analytics require new skills. Additionally, as described under Organization & Culture, understanding and applying data to decision making are important to ensure the efforts will have a sustainable impact on the business. Hiring or contracting skills to an organization is difficult in today’s market where data talent is in high demand. Therefore, a clear Human skills plan is needed to ensure the relevant capabilities are available when and where needed.
Starting with a data transformation in an organization typically requires the use of external resources to help with putting a plan together and potentially help with the implementation of the first data use cases. Using externals is logical in the first stages when your own data team is non-existent or small. Externals that have worked and learned from similar projects in the past are invaluable in helping to give the right direction and making the first success stories. However, using external resources extensively will increase the budget and decrease the development of in-house skills. The goal is to find the right balance between externals and internals. At optimum, the external resources ensure that the boat is moving to the right wind angle, downwind, and upwind, while internal resources are continuously learning new manoeuvres and the leadership is telling where the whole team is headed.
Similarly to the balance between externals and internals, the data team should have the right balance between seniors and juniors matching the overall data maturity. As demand for skills often varies between projects, a variety of seniority helps assign tasks and keep the team motivated. Moreover, a well-designed career path and harmonized titles in the data team help with attracting talent, juniors and seniors alike. Leading organizations can attract the best-in-class talent, which further boosts their maturity.
Data use cases will have limited impact if the organization is not ready to exploit the results for improving the business and operations. Data literacy is the ability to read, understand, create, and communicate data as information. Throughout the whole organization, data literacy skills are needed to ensure the insights from the analytics team are correctly understood and appropriate actions are taken.
As a whole, there are many skills to be fulfilled in the transformation. Some can be hired, some outsourced, but more importantly, the whole organization’s knowledge of data and AI must be lifted. This is necessary to meet general data literacy requirements, establish a culture of innovation with data, and allow smoother communication between business and data teams. In addition to general training, special trainings should be available for the data team members to upskill their expert knowledge.
Keeping in mind Privacy and Ethics
Privacy and ethics of using data and AI are important and hot topics. While these don’t necessarily speed up the data transformation, they ensure that using data conforms to our society’s legal and ethical standards. Failing to comply with either standard may result in severe consequences, which is why we have dedicated an entire section to the topic.
Data privacy, sometimes also referred to as information privacy, is an area of data protection that concerns the proper handling of sensitive data including, notably, personal data but also other confidential data, such as certain financial data and intellectual property data. Personal data means any information that can be used to distinguish or trace the identity of an individual, or when combined with other personal or identifying information that is linked or linkable to a specific individual.
In the European Union, UK, California, and many other regions, data privacy legislation defines a frame for how privacy data is to be handled. At the minimum, data privacy should comply with these legal and possible contractual requirements. However, this is often not enough. The basic level means that the mandatory requirements are fulfilled, but not that the privacy-related responsibilities are clear and processes efficient. Further to avoid the unnecessary leak of privacy data the general awareness of privacy should be raised in the organization.
There are no clearly defined rules for evaluating ethics, as there are for data privacy. The European Commission has issued a proposal of an AI law that may become the first benchmark for regulating Data and AI, but that will be in effect earliest in 2022. Meanwhile, organizations should define their own guidelines for ethical use of data, and establish governance processes to enforce them. Guidelines should be defined from a starting point of what is the generally acceptable and fair use of data both within the industry and the broader societal context. What sets leaders apart in this area is the ability to translate, communicate and embed the general guidelines into the day-to-day operations of every employee working with data. Clear guidelines illustrate to the organization what data privacy and ethics mean in practice.
Transformation starts with people
Transformations are done by humans. Having the right human skills, a culture that values data and a supporting organization for data transformation make the journey smoother. To make sure you are making progress, it is recommended to do the data maturity assessment repeatedly, continuously understand where you are on the path, and ensure no enablers are falling behind.
To learn more about the journey it is recommended also to read a guide written by two of our founders: How to define and execute your Data and AI strategy.