Remember May 2018? That was the month the expansion of the EU General Data Protection Regulation (GDPR) was enacted. Months in advance people were up in arms about what this new regulation would mean for software vendors and email marketing (newsletters). Well, and then May came and went and, surprisingly, life continued in a normal and unexcited manner. Wouldn’t it be great if we could maintain that level of normalcy and calm, while engaging in an even more extensive discussion about data that has the potential to radically change the way we work and live? Let’s dive right in.
In order to have a calm discussion about data and artificial intelligence, it helps to approach the topic with openness and curiosity. Let’s move away from extremes! Neither should we go crazy experimenting with data to see what might come out in the end, nor is it helpful to promote an image of robots that are more intelligent than humans and will, sooner or later, take over the world. Instead, let’s take a closer look at all the positive developments in technology that already exist today and that are based on humane and ethical principles and see how, together, we can develop them further. When we use data sensibly, we can achieve great things. If we collect it based on ethical principles, analyse it responsibly and embed it in “open organizations” that encourage participation, individuality and diversity, then data-based technology can facilitate a new level of freedom for everyone. Employees could finally fully unleash their talents and strengths, allowing HR managers to orchestrate all this new potential. As a result, organizations would become more efficient and productive.
So how do you collect and use data ethically correct?
Good data serves people, and not just a few but, in the best case, everyone. When you break that down to a company, not only the upper management levels benefit, but also the people in lower levels, regardless of department or job title. In order to assess whether algorithms and AI are being processed in an ethical way, we need criteria that can be used as guides. The “Ethikbeirat HR Tech”, an ethics committee for HR tech, has suggested a set of guidelines to evaluate data quality. They are organized in three categories:
- Data collection: Why is data being collected? How is it being collected? Who is collecting it?
- Information extraction: What is the data being analyzed for? How is it analyzed? How is it converted to useful information?
- Decision: Which conclusions are being drawn from the data analysis? How are these conclusions being drawn? How are decisions and recommendations derived?
Within these three evaluation categories there are specific questions that organizations can use as a checklist for the handling of data. And since we love checklists just as much as flawless data, we will show you how it can work by giving you a glimpse into our Tandemploy Data Check. As would be expected, our matching software is also based on data and smart algorithms that use that data to generate recommendations. We are developing the tool according to the following highest ethical standards (imitation desired!):
1. Everyone understands the goal and benefit.
Employees of companies that use our matching software, have a specific need and a specific goal in mind. For instance, they might need special know-how from a colleague in another department, are are looking for a mentor to prepare for a new task or want to get involved in a project. Overall there are 18 application areas for collaboration and exchange across departments and positions. Companies choose the applications that fit their needs and make them available for all employees to use. The employees then decide for themselves which networking approach they want to take.
2. Nothing happens by accident.
Our SaaS has a strong foundation that our fundamental attitude, which puts employees and the employee experience at the center of all thought processes, is virtually built into. Based on that, we worked with experts to develop our algorithms in-house. We also received support and consultation services from scientific institutions. In addition, we ourselves continuously research and use the experiences our customers relay to us. Our software not only inspires knowledge exchange and continuous learning in organizations, it itself is also constantly learning and adapting.
3. The algorithm thinks – but it’s the people who guide.
„People matter“ – that is the Tandemploy philosophy that permeates through everything we do. Also in our matching tool, the user has the last word. Nothing happens unless the user actively sets it in motion. Even the algorithms don’t act uncontrolled. The ontology on which our software is based is the result of six years of manual maintenance – and thus, first and foremost human intelligence, which is gradually being supported by technological intelligence.
4. Everyone knows what they’re doing.
The logic behind our software is easily comprehensible by everyone. The matching of colleagues follows a clear pattern that results from the interests that the users indicate: What topics are you interested in? What would you like to learn about? What knowledge can you share? Also fundamental factors like location and language may be relevant.
For example: With topics like “job sharing” and “projects” the sought after skills play a key roll. For the Job Sharing module, people with similar skills are matched as they are, after all, going to be sharing a position. With other modules, such as the Onboarding Buddy, general interest and a certain amount of time worked with the company are enough to be matched. Skills don’t play a roll here. Each of our fields of application has its own logic that is understandable for everyone and ensures that everyone only enters the data that serves their personal goal.
5. We encourage others to tolerate and support change.
Companies that use our software are responsible for the consequences. Therefore, we constantly emphasize that technology and culture must fit together. If companies promote the use of a networking tool amongst their employees, they should be prepared to not only tolerate the changes in work flows and structures that result from it, but also to encourage and maintain these changes – even if that means redefining their own management role.
6. We collect as little data as possible.
Our aim is to promote big changes without organizations having to lay themselves completely bare or bend over backwards. Change begins with the individual, so it was clear to us that our software had to be operated by the employees themselves. They have complete data sovereignty and only share data about themselves that is relevant for their development and networking goals. Specifically, that means that the data collected will vary depending on what kind of match is being made. A personalized evaluation of an individual is not possible. In other words, if required, HR management can receive an aggregated and anonymized skill map, which they can use to identify skills and learning needs in the organization, but they can not obtain insight into the skills and learning interests of individuals if they are not publicly visible. In addition, users can create a pseudonym in order to stay anonymous.
7. We create awareness that and how the machine decides.
Whoever utilizes our tool to network with colleagues or to put their hat in the game for a job, a project or an event, knows that the information will be processed automatically. The user fully understands that “the machine” (and not a human) decides whether he or she qualifies as a project partner or to whom he or she is suggested as a matching partner. The logic behind the way the data is being processed is comprehensible for everyone involved (see point 4).
8. It’s people who handle the data – not data that handles the people.
When entering your data into the Tandemploy Matching Tool, you can edit, add or delete it at any time. Nothing happens without the employees explicit action. This bottom-up approach values every individual in an organization and acknowledges their unique ability to organize their own professional development, allowing them to be the owners of their career path.
9. Our code is based on equal opportunities.
One of the most important requirements for AI and data-based technology is that it makes fair decisions and doesn’t discriminate against anyone based on gender, skin color, origin, religion or sexual orientation. We prevent programmed discrimination in the development of our SaaS.
Two examples: Senior employees will never be considered a “better match” than other employees. Job sharing users can indicate that they are interested in taking on a management position in the future and therefore ONLY want to network with colleagues who already work in management.
We generally work with virtual waiting lists as demonstrated by the example of onboarding buddies: The onboarding buddies will be suggested in order of registration, without any preference given to personal characteristics or job status. All users that have made themselves available as buddies, will be put in the waiting list in the order they registered. While a buddy is participating in an onboarding process, they will not appear in the waiting list. Once the onboarding process is finished, the buddy will be placed at the back of the line once again.
10. Regular check and exit option
For us, good data security is the result of good collaboration – both within Tandemploy as well as with our external data security officer, with whom we execute routine risk assessments. Should we have the slightest doubt about the safe and ethical use of the data, we can stop and improve it at any time. In addition, we log all transactions in a digital processing directory. An interview with Marco Tessendorf, director of Procado and our external data security officer, and with our internal data security expert, Silja, about the implementation of the GDPR at Tandemploy can be found here.
And now to you: How do you build the bridge between people and technology, between ethics and automation, between good data and New Work? How do you prevent discrimination? What does your data check look like? We look forward to hearing your ideas and thoughts on the matter!