Artificial intelligence (AI) is unquestionably a powerful tool. Its economic value is increasing tremendously and transforming numerous industries such as manufacturing, fintech, healthcare and automobile. Workers in finance and marketing have much success using AI technologies, whereas HR practitioners find it rather hard to integrate these into their daily practices.
Prasanna Tambe, Peter Cappelli and Valery Yakubovich state in their research: “there are systemic and structural differences for HR that do make it harder, when you are building an AI-based system.” . Due to the fact that the quality and explanatory power of big data and AI are limited they are still considered unconventional in the fields of HR and employment. To have a better understanding of this matter, we need to consider the AI problems in terms of data science in human resource management (HRM).
There are three main challenges in HR practices when it comes to data science. The first issue is the lack of consistency in HR process measurement throughout the employee life cycle. For example, when determining which candidate to hire or choosing whom to promote, it is crucial to consistently record and analyze which criteria and skills were the decisive factors in the previous hiring process.
The second problem with HR practices is the limitation of data sets produced in HRM. Unlike some fields such as marketing and finance, where a lot of data are generated and easily gathered, data collection in HRM faces big challenges in terms of quantity and quality. Moreover, data in HRM is oftentimes unstructured (on paper, in excel or PDF) and consequently, difficult for a computer to process.
The last difficulty concerns ethical issues related to data processing. The results of HR decisions can have a significant impact on someone’s career. Therefore, it is imperative to think about how fairness and transparency can be achieved. Furthermore, it is also crucial to know how employees react to the results that are solely based on data-driven algorithms. As Morgan Hampton from Tesla declared, “recruitment should be automated as much as possible, hiring should remain human.”
Taking these three issues into consideration when searching for AI solutions, HR managers should focus on the following aspects in order to utilize AI more effectively. Firstly, HR managers need to create the right HR process that is ready both for the digital age and AI technology.
Currently, AI technologies are separately, for example in recruiting and talent acquisition, payroll management and self-service transactions. However, they lack a mechanism to generate data that can assist the whole AI process in HR practices.
HR managers often merely keep the applications that they are interested in and don’t retain those that are screened out. This leads to a one-dimensional analysis and conclusion . All these criteria should be gathered in data collection and eventually, be evaluated to facilitate the development of big data models and AI processes.
Furthermore, it is also imperative to generate data in a sustainable way. For instance, there are AI applications that can predict which workers are about to quit their jobs, and some even track data points from employees’ social media or emails . If employees were aware of such a system, they would probably change their behavior and deliberately produce misleading data.
Last year, the story about the Amazon AI recruiting tool being biased against women was proof that machine learning can mimic human attitudes. Gender, nonetheless, is not the only aspect that is reason for discrimination. Others such as age, nationality or ethnicity could also have a negative impact, keeping companies from inclusive and diverse hiring. HR managers should carefully collect data samples that are representative and look for explainable AI solutions. The complex neural networks in deep learning is far from self-explanatory.
Up until today, the standard data criteria that HR managers should respect throughout the HR practice cycle does not exist yet. This means that HR managers have to team up with their company’s internal IT department or with external AI vendors to determine what data to trace and how to measure those data, in order to establish the best practices for AI within their companies.
At JANZZ.technology we believe that collecting and structuring data is fundamental for creating smart data. Our parsing tool extracts the right entities from paper, Excel or PDF, ensuring a fair end-to-end data processing from the very beginning. Do you want to know more about our parser and how we can assist you in your AI transformation journey? Please write now to firstname.lastname@example.org
Prasanna Tambe, Peter Cappellli and Valery Yakubovich. 2019. Artificial intelligence in human resources management: Challenges and a path forward. URL:https://www.researchgate.net/profile/Peter_Cappelli/publication/328798021_Artificial_Intelligence_in_Human_Resources_Management_Challenges_and_a_Path_Forward/links/5c5edc7f299bf1d14cb7dc5f/Artificial-Intelligence-in-Human-Resources-Management-Challenges-and-a-Path-Forward.pdf [2019.10.20]
 Samantha Mclaren. 2019. Here’s how IBM predicts 95% of its turnover using data. URL:https://business.linkedin.com/talent-solutions/blog/artificial-intelligence/2019/IBM-predicts-95-percent-of-turnover-using-AI-and-data[2019.10.20]