FROM Felix Anderegg & Anja Buser / HR Today

Have you been in contact with artificial intelligence (AI) today? No? Are you sure? Because for users it’s impossible to know whether AI is hiding behind a user interface or not. Firstly, AI stays in the background hiding behind chatbots, search engines, voice assistants and smartphones. For example, have you asked Siri anything today? AI. Have you taken a selfie with your smartphone? And did it recognise you? AI.

Secondly, there is often simply not a lot of knowledge about what AI actually is. People think of horror stories of computers coming to life to take over the world. “AI... could spell the end of the human race” - quotes like this one from Stephen Hawking don’t exactly help alleviate people’s fears either. And if you google AI, suggestions include terms like ‘danger’ and ‘job killer’ rather than ‘possibilities’ and ‘opportunities’. We also have Hollywood to thank for a lot of this panic, along with plenty of ignorance.

It’s no wonder then that 80 per cent of managers say that their employees need to learn more about AI in order to feel comfortable with it.² Once employees have developed an understanding of what AI is and what it can do, they get a feel for the areas in which they can benefit from AI and make the best use of it. There are a lot of opportunities to use AI particularly in HR.

What does AI do?

There is still not a general definition of AI, since there is also no scientific definition of the term ‘intelligence’. That does not make it easier for us to understand what AI is. Computer science aspires to automate intelligent behaviour with AI. Put simply, this means AI is fed with data and independently learns to recognise correlations and to generate output based on them. The more data the better. In theory, you could create a similar output using traditional programming. This would require you to know the logic behind this output. For example, being able to tell exactly why a recruiter chose a particular candidate. But the logic behind such decision making processes is too complex to be easily representable. This is where AI comes in.

What are the limits of AI?

1. Explanation:in the same way that humans cannot always give an answer to why we make certain decisions, neither can AI.
2. Empathy:when the AI system AlphaGo beat Ke Jie, one of the world’s best Go players in 2017, he was so disappointed he cried. AlphaGo on the other hand, felt neither happiness nor the need to give Ke Jie a comforting hug.
3. Creating something completely new:for example, if you input lots of pictures into AI, it transfers them into series of numbers, establishes statistical correlations and creates something new with similar numbers, patterns and distributions – a new composition of something that already exists.³
4. Identify prejudices:AI carries over the prejudices of the data it’s fed with and cannot recognise them as such. It also cannot predict the consequences of the decisions it takes. This became apparent when Amazon tried to recruit using AI and realised that female applicants were being discriminated against because the training data included more male applicants.⁴

What uses are there for AI?

Despite its limitations, there are many opportunities for using AI. But what exactly can you do with it? Put simply, there are three different fields of application for AI. These can work separately or together.

1. Categorisation: AI helps when something needs to be classified, but not everything follows the same logic. For example, it can help recognise and categorise expense receipts. Does the Starbucks coffee fall under food and drink, public transport or accommodation?
2. Calculating similarities:If you want to compare things, AI can help you make decisions. One example of this is matching CVs to job profiles. Many companies check and pre-sort their CVs this way.
3. Knowledge transfer:By allowing AI access to different data sources, it can answer questions based on the content. In some companies this is used for talent management, for example. For instance, a candidate can ask a chatbot what is required of them for a promotion.

How can HR use AI?

Based on the fields of application, HR can use AI in four different ways, depending on whether feeling and empathy or rationality and logic are more important, and whether it’s a matter of creativity and strategy or optimisation.

1 and 2: AI can take on routine tasks

HR can delegate any repetitive tasks to AI. Depending on how important empathy is, HR plays a bigger or smaller part.

1 AI relieves: when empathy is not needed or emotions are even detrimental, HR can hand over the tasks to AI – for example, in the areas of employee services, document management, HR administration, as well as travel and expense management. AI supported chatbots are already the first point of contact for employees’ questions today. In document management, AI takes on the first allocation of documents and inserts certain content. And maybe AI will be able to support HR in the monitoring of salaries in the future.

2 AI frees up your time: If routine tasks require empathy, AI operates as a tool in the background. AI can support recruiters with an initial evaluation of the CV or assess a person’s character based on video data. AI could also take on the entire planning of onboarding, and an AI supported chatbot could accompany the employee during this process.

3 and 4: AI can provide HR with support for complex issues

As tasks get more complex and require more creativity, it’s clearly a job for humans. AI can and will provide support in the future, but it’s not quite there yet in many cases.

3 Hand in hand:with complex matters, where the focus is on reason, AI already provides support in some cases. Especially in analytics, there are endless possibilities for using AI. AI calculates the probability of an employee leaving the company or provides support for personnel cost planning. AI can also be used when it comes to compensation. A simple system is already in use today for processing highly standardised bonus arrangements. In the future, AI might also assist with more complex processes.

4 AI provides selective support:as soon as empathy becomes more important and an issue becomes more complex, AI is used very selectively and in the background. The more important creativity and empathy are, the less important AI is. But particularly with issues like development, learning, as well as performance and goals, an AI supported chatbot can lead employees through the process and act as a sparring partner. For performance & goals or in succession planning, AI can make an initial suggestion for calibration or classification in the performance matrix or for a successor.

KI and HR, forms of cooperation⁵

HR-Themes and AI⁶

Do you have to disclose AI?

When so many possible uses exist and a user never knows whether AI is behind them, what do you need to consider from a legal standpoint? In Europe, the General Data Protection Regulation (GDPR) regulates to some extent how AI should be handled. Experts argue how strictly it should be interpreted, but everyone agrees that Articles 13 and 14 require companies to inform users and provide details as soon as AI makes decisions that have a significant influence on the users. There are no regulations for this yet in Switzerland. Legally speaking, it makes sense to comply with the EU regulations.

What should HR do now?

1. Educate yourself: AI does not appear to be a short-lived trend, but is here to stay. HR do not have to be AI experts, but should be aware of the possibilities and limitations of AI. This way you can assess what makes sense and where caution is needed.
2. Start with the simple matters:begin by using AI for repetitive issues that require reasoning. This is where you will also see the greatest effects at an early stage.
3. Develop with a focus:AI always hides behind another interface or a system. Before you start developing by yourself, make sure to check the market to see if a solution already exists for what you need.

Sources:

¹ Stephen Hawking, Web Summit technology conference in Lisbon, Portugal, 2017

² IBM Research: https://www.ibm.com/smarter-workforce/ai-in-hr-myths-and-facts

³ Prof. Dr. Paul Lukowicz: https://www.kulturmanagement.net/Themen/Kuenstliche-Intelligenz-und-Kreativitaet-KI-als-Kultur-Geschaeftsfuehrer-der-Zukunft,2348

⁴ Reuters, Amazon scraps secret AU recruiting tool that showed bias against women (https://www.reuters.com/article/us-amazon-com-jobs-automation-insight/amazon-scraps-secret-ai-recruiting-tool-thatshowed-bias-against-women-idUSKCN1MK08G)

⁵ Kai-Fu Lee: https://www.ted.com/talks/kai_fu_lee_how_ai_can_save_our_humanity

⁶ HR Campus in Anlehnung an Kai-Fu Lee