With this article, I am sharing my personal learnings from creating an Artificial Intelligence-enabled UI. […]
Introducing Artificial Intelligence in your organization?
Eleven tips from Deloitte's Artificial Intelligence experts
Everyone has an opinion about it, but not many organisations are actually doing anything. Therefore, Artificial Intelligence raises a lot of questions. How do ensure that your organisation takes advantage of the opportunities? How will Artificial Intelligence be embraced by your employees and truly embedded in your organisation? And where do you start?
Last year Deloitte Netherlands created the Artificial Intelligence Center of Expertise (AICE) to accommodate its Artificial Intelligence organisation. We are pleased to share with you the 11 most important lessons that we learned doing so.
1. Ensure that you have the support of the leadership and embedded in the strategy
The establishment of an Artificial Intelligence organisation can be seen as an organisational change. And as with more or less all changes, it is important that the change is supported by the leadership and that a clear vision is formulated.
The leadership must back the establishment of an Artificial Intelligence organisation and be prepared to fund it and give it sufficient time. For instance, to manage the Artificial Intelligence community and to experiment with this new technology. It is important that the leadership understands the usefulness and necessity of this investment. Does the leadership realise how Artificial Intelligence will influence the organisation?
“At both the global and North West Europe level, the focus on new technologies such as Artificial Intelligence and Blockchain is an aspect of Deloitte’s innovation strategy. Within Deloitte North West Europe, we have appointed a responsible partner for Transformational Solutions, who currently focuses on Artificial Intelligence and Blockchain. This means that we have also decided to focus on Artificial Intelligence in our organisation in the Netherlands” – Richard Roovers, Partner Deloitte Netherlands, Innovation Lead Transformational Solutions NWE
2. Create general awareness of Artificial Intelligence among the employees
Even for technology-driven organisational changes, ‘it is not (at) all about technology’. Getting your employees to accept and participate is just as important. To do this, it is important that people are aware of Artificial Intelligence and also understand how it will affect their organisation.
With Artificial Intelligence, this is difficult, because it is intangible. Therefore, within Deloitte Netherlands, we developed an awareness campaign to help our employees becoming aware of Artificial Intelligence. We did this with our AI ‘AIME’ co-worker. AIME gently introduced our employees to the idea. This aroused their curiosity. Our employees could then voluntarily sign up for our ‘Artificial Intelligence for dummies sessions’ to learn more about the subject.
“There was great interest in our Artificial Intelligence for dummies sessions. Our employees are eager to know more about Artificial Intelligence. The sessions make it easier for them to discuss the Artificial Intelligence applications with customers and the people around them. The sessions give our colleagues a better feeling of what is and what is not possible. Even our CEO participated! — Arjen Lettinga, organiser of the ‘Artificial Intelligence for dummies sessions’, Deloitte Innovation.
3. Get the Artificial Intelligence experts on board
Before AICE, numerous people within Deloitte Netherlands were already working with Artificial Intelligence. Our organisation is organised in a number of business units, this meant that the various Artificial Intelligence initiatives were found in different parts of the organisation. The people in one business unit were sometimes unaware of the activities that were taking place in another business unit. Therefore, the first step was to identify what Artificial Intelligence developments had already taken place in the organisation and to bring together all of the people who had a technical Artificial Intelligence focus.
Starting by involving people with a technical background has various advantages. They are often ‘early adopters’ and therefore know more about the subject. As a result, they already know what is (and in particular what is not) possible and they are often more driven by content and less by commerce. This means that they can quickly find common ground with respect to content, meaning that department boundaries vanish. This is important when building an organisation-wide community.
We therefore appointed a technical Artificial Intelligence leader within every business unit. When selecting the correct people, intrinsic motivation is more important than job level. However, we asked people in the more senior positions to think about who would be the right person in their unit (refer to Item 1 for leadership commitment). It is important, in addition to intrinsic motivation, for the chosen people to have an understanding, and that, for instance, they have included it in their annual plan so that they want to and are able to devote time to the subject.
“Involve people in the program who have technical knowledge, a hands-on mentality, sufficient time and are not afraid of changing the status quo. Intrinsic motivation and knowledge are more important than job level. Giving senior leaders a role in the program might be the obvious thing to do, but can be counterproductive because it can impact existing relationships” – Ewout Bolhuis, Director, Deloitte Innovation, the Netherlands
4. Make choices and set clear objectives
Together with the chosen Artificial Intelligence Tech leaders, in a number of sessions, we have drawn up our joint annual plan. It is important to bring all of the different ambitions into line and to make choices (see the example 100 euro exercise below). When making choices it is good to take (technical) feasibility into account.
When drawing up the annual plan, it is wise to include the ‘Low-hanging fruit’ at the start of the process. For instance, by communicating what is already happening in the area of Artificial Intelligence or by achieving quick successes. This increases confidence in the program, also among the leadership team.
“In one of the first sessions, all of the Artificial Intelligence Tech leaders were asked what activities they wanted to include in AICE. This resulted in a long list, that would not be possible to realise in just one year . To address this, we used the ‘100 euro exercise’. In this exercise, the leaders had to divide a notional 100 euro among all of the activities. This gave us insight into the differences between the departments and which activities should be prioritised. It helped us to make choices and generate support for the plan” – Marjolein Vlaming, Program Manager AICE
5. Build the Artificial Intelligence community but be selective
Soon after gaining a picture of the technical community, we received requests from people with more of a ‘business focus’ to become part of the Artificial Intelligence community. In every business unit, we appointed an Artificial Intelligence business leader. Involving people with a business focus, in addition to – and after appointing – the technical people, has various advantages.
Technical people often focus on their technical solution. This is a positive approach, but if it goes too far there is a risk that something is developed that is not based on an actual business problem (‘develop for the sake of developing’). People with a business focus are in close contact with the customer and therefore they can identify the problems actually faced by the organisations. That said, technical people better understand what is and is not currently possible. They can prevent customers being promised something that cannot be achieved. By involving both groups, you create a situation that allows the optimum solution to be offered to the customer.
However, it is important to involve both groups differently in the Artificial Intelligence organisation. What interests the Tech community is often uninteresting for the business community and vice versa. We therefore made the conscious decision to cooperate on some activities (for instance some community events) and to keep other activities separate (for instance training sessions).
“Artificial Intelligence is a popular theme. Even so, when building the community it is important for people to have the same ambitions. The interests of technicians are different to those of employees with a business focus. Technicians want to get a greater understanding of, for instance, the various technologies. It is important to serve both groups, but each in their own way.” – Wouter Pepping, Artificial Intelligence Lead Risk Advisory, Deloitte Netherlands
For people with a Business focus, we made it a condition that they want to focus on Artificial Intelligence in their daily work. We do not make people who have only an interest in the theme part of the community. For this group of employees, we organised the ‘Artificial Intelligence for dummies sessions’ that gives them a basic knowledge.
We also included our IT Department and Risk & Reputation Office in the organisation. This lets us optimally facilitate the community (for instance when buying software) and also identify and rapidly address data and privacy issues.
Our method of classification resulted in the following groups:
“There is a threat hiding behind every opportunity. Therefore, it is good that I have been involved in the AI program from the very start to be able to give timely advice or identify risks” – Alfons Koenders, Deputy Confidentiality Officer/National Security Officer, Risk & Reputation office, Deloitte Netherlands.
6. Outline clear and realistic expectations: Artificial Intelligence is not a panacea
In particular in the first stage of your Artificial Intelligence program, but also later, it is crucially important to manage the expectations concerning. In general, within organisations there are two (mis)conceptions:
- That Artificial Intelligence is the solution to all problems
- That the organisation is already able to do everything with Artificial Intelligence
It is important that both the leadership and the sales people understand the current (im)possibilities of Artificial Intelligence. This prevents awkward situations arising, for instance that things are promised that cannot yet be realised.
To be able to finally succeed as an organisation, in this initial phase the emphasis must be on building a community, gaining experience and experimenting. Therefore, it is possible that in this first phase sales are limited or absent, that less is possible than envisaged and that mistakes will be made when experimenting. The leadership must have the courage to invest, in spite of possible setbacks, and to maintain a long-term vision.
“Artificial Intelligence is not a panacea. It is very possible that for a certain problem Artificial Intelligence is not the solution, but that, for instance, a simple regression analysis or RPA will suffice. It is important to continually consider whether Artificial Intelligence is the best way to solve the customer’s problem. People who talk with customers about Artificial Intelligence must be aware of this” — Willem van Willigen, Artificial Intelligence Lead Financial Advisory Services, Deloitte Netherlands
7. Ensure there is technological support
It is important that the time and money that is invested in the program is also used to facilitate the community in the area of ICT. Consider, for instance, establishing a data platform, a code repository (GIT) for sharing code or configuring collaboration tools.
“A code repository (GIT) for sharing code can facilitate knowledge sharing across department boundaries” – Roderick Lucas, Artificial Intelligence Lead within Tax & Legal, Deloitte Netherlands
“In general, this group has an above-average interest in new technology. This offers us the opportunity to experiment on a small scale within our organisation with new software, for instance collaboration tools or Artificial Intelligence cloud solutions. This allows us to kill two birds with one stone” — Jan-Pieter Kuyt, Artificial Intelligence Leader within IT&WS, Deloitte Netherlands
8. Appoint a program manager from a ‘neutral’ department
As indicated above, Deloitte Netherlands is made up of various business units. In spite of the fact that all of the units belong to the same organisation, business units can have (un)conscious preferences that hamper knowledge sharing. Because the Innovation department serves a general, company-wide interest within Deloitte Netherlands, it is easier for us to take on a connecting role. Time and money has been assigned to the program and a program manager has been appointed.
“It is very helpful that the Artificial Intelligence program is coordinated from a central department and that we can exchange knowledge with other units in the organisation. Managing the program requires wide-ranging expertise, which is not always on hand within your own unit. The clear meeting structure and the regular contact we have ensure that the program becomes part of our job” – Twan van Gool, Artificial Intelligence Lead Audit, Deloitte Netherlands
9. List the Artificial Intelligence competencies
When you have identified the community, it is important to investigate which competencies you have available and at which level. This gives you a clear picture of which business units are progressing well or lagging behind, which Artificial Intelligence skills are missing and where the experts can be found. The result of this analysis forms the basis for, among other things, the Artificial Intelligence training plan (what training, to whom, when), and the classification of community members in workflows (business, technology) and level (potentials, experts).
“Together with a few colleagues who have PhDs in Artificial Intelligence, we developed a framework to map out our Artificial Intelligence competencies. The framework addresses hard skills (for instance whether one has knowledge of certain technologies) but also knowledge of other relevant aspects including ethics” — Naser Bakhshi, Artificial Intelligence Lead Consulting, Deloitte Netherlands
10. Offer a safe learning environment
Many people are unfamiliar with Artificial Intelligence. It is also a very wide domain. It is very possible that people have, for instance, experience with classification problems, but have never done anything with Natural Language Processing. As a result, there is a high chance that community members do not yet know or understand some aspects. It is essential to ask questions. Record what has been learned and share it, so that others do not make the same mistakes.
It is realistic to assume that mistakes will be made in the initial phase. This makes it important to think about risk avoidance, but it is also important to think about how you can identify any compliance or security breaches as fast and effectively as possible and correct them. Involving the Risk & Reputation office (see Tip 5) can help here.
11. Communicate about Artificial Intelligence
Communicate a lot and frequently. Make clear to the organisation what has been done in the area of Artificial Intelligence within the program and repeat the message. Do not assume that the knowledge will be known immediately. Specify short-term successes and long-term progress.
Communicate at both the organisational and personal level. Enter into a dialogue with colleagues. It concerns a complex subject. Time must be allocated for explanations and, where required, clarification. Discuss things not only with the ‘enthusiasts’ but also with the sceptics. This will help you to learn about and respond to what is going on.
Artificial Intelligence is currently a hot theme in many organisations, in both the board room and among the employees. This is the moment to start to introduce it into your own organisation!