An AI-Based Shortlisting Model for Sustainability of Human Resource Management

Round 1

Reviewer 1 Report

I appreciate the authors engagement with this important and increasingly relevant topic. Unfortunately, I do not understand the author's intent and the introduction introduces many elements of the HR process which adds substantial fuzziness to the ultimate focus of the article. It is difficult to discern the authors overall objective and so to their own analysis suffers. The writing lists the contributions of many, but does not coalesce into an analysis that seems to be going somewhere.Elements about payroll and employee development are not related to the focus of the subsequent analysis (recruitment efficiency). A characteristic sentence "One of the hottest topics is Artificial Intelligence (AI) and AI-based technologies, such as virtual reality, natural language processing, computer vision, and virtual agents. " reflects both a casual approach to the subject and introduces a variety of concepts that are not related to adjacent context or referred to again in the article (computer vision and virtual agents [relabeled chatbots, which is also sort of erroneous].  Lacking focus, it was difficult for me to see where the article was going. If the article is to be successfully revised, the introduction should only briefly touch on these elements and emphasize the area where the study focuses (recruitment, and specifically, shortlist creation). 

The literature review does has several paragraphs that start very broadly (e.g. HRM is x) and then turn to a specific technological function without framing that context for recruitment (e.g. chatbots, natural language processing - two very different things - one a public facing software element that uses NLP, and another an algorithmic procedure. The authors acknowledge this, but it is not clear why these are the sections and how they ultimately relate to to the analysis, which seems mostly to be based on more efficiently identifying a subset of applicants (a goal of AI Systems generally). The quick turn from a broad HR focus to AI functions misses a more thorough and detailed justification for recruitment as the area of focus.  There is substantial literature about recruitment challenges as a time concern (As the abstract implies) and an equity concern (which is implied unevenly but without focus in the paper). 

When reviewing the method, it is not clear the procedure that is undertaken and the result. It needs substantial crosswalk to be better understandable - how do MDL and SVM actually process a complex text document like a resume? Are we using successful applicants or templated successful application packages to train? Figure 2 seems to be used again from somewhere - it could be adjusted to explain what is going on in better detail. 

The authors emphasize that large data is needed for AI To be successful, yet only select 100 applicants for analysis (though this is also somewhat unclear because it mentions 100 data points, which might be the various variables included in training, but clarifies them as applicants). These appear to be, though it is unclear, for various different types of jobs (figure 3). Is this the result of natural language processing identifying those elements and grading them or the authors. 

There is an uneven explanation of the data cleaning employed, but it is not clear the ultimate purpose of the exercise. The algorithm used (SVM) is introduced, but it is unclear what that method is intended to gain and how that applies to the complex data elements (codified resumes). The tables that supply results are not sufficiently explained. 

There is a key missing element that I might just not be able to parse - but how are the 85 elements identified? It implies a human removing those 22 pieces, but this is unclear.

The authors imply in their abstract that this type of analysis technique will save time, but provide no comparative context - how much does processing 100 resumes take for a human versus the analysis? The author implies that the data is necessarily messy and required substantial cleaning and contextualization in order to use the AI - does this negate the time savings?  

The conclusion does not add further insight to these questions and seems to acknowledge a lack of value so far. This could be substantially improved by a more rigorous literature review focused on recruitment, applied elements anticipated to be addressed in the model, and specific results for those elements. Introduction of the many different AI possibilities as well as the many HRM responsibilities does not help understand why the model runs are useful but the authors seem to assume that a reader will understand the significance of their results, but unfortunately for me they seem to be running on noisy data and have some sort of result - but that result does not look like a shortlist. Based on the abstract (less clear in the introduction) I might expect some sort of result like this -  from 100 or 140 resumes/applicants, this method was able to shortlist 20 for consideration with .91% confidence and this was verified by independent HR professionals who identified 21 for consideration (19 from the 20 identified by AI) but took 3 hours to complete the task compared to 5 minutes for both model runs verifying the value of the AI tool but highlighting the potential for different imperatives. 

Author Response

Thank you for your time, your effort and guidance.

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The manuscript is interesting to read. However, the purpose and significance of the study should be clearly discussed. The methodology should be discussed elaborately. In its current form, the theoretical and practical contributions are very narrow. The author(s) should discuss the present study's limitations and the directions for future research. The manuscript requires English proofreading. 

Comments for author File: Comments.pdf

Author Response

Thank you for your time, your effort and guidance.

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Dear Authors

The topic is interesting. Artificial intelligence has fascinated researchers in various scientific fields for many years. In human resource management, including recruitment, it is already used by many companies. The article has potential, but the Authors need to make some changes and corrections.

1. The purpose of the article should be specified. It should be included both in the abstract and in the methodology.

2. The authors write " but also can eliminate unconscious

human bias and enables HR to focus on attracting the best candidates to keep competitive advantage for the company [14]" It should be noted here that artificial intelligence can also be discriminatory and unfair (see A Systematic Review on Ethical Concerns and Related Strategies for Designing with AI in

healthcare. AI 2023, 4, 28–52. https://doi.org/10.3390/ai4010003)

3. The authors write: "It looks like AI will be here for a long time, but still for now there are some obstacles and challenges regarding AI developments, such as training data shortage [9], regulations, and lack of standards [10] In addition to that, ethical issues and transparency concerns have been started to discuss [11]. It is worth extending these considerations to the problem of technological unemployment (see.Technological Unemployment in the Perspective of Industry 4.0. Development. Virtual Economics 2022, 5(1) 7-23. https://doi.org/10.34021/ve.2022.05.01(1) and the social perception of artificial intelligence (see Cognitive Technologies and Artificial Intelligence in Social Perception. Management Systems in Production Engineering, 2022, 30( 2), pp. 109–115. https://doi.org/10.2478/mspe-2022-0014)

4. Conclusion and Discussions needs to be supplemented. In the discussion, the results of the Authors' own research should be analyzed and compared with previously conducted research. It should be determined what is the novelty of this article, i.e. what the authors have brought with their research to the development of science and practice on the use of artificial intelligence in human resource management.

5. Some drawings are of poor quality. You can't see much of them. Need to improve.

6. Specify when the Authors' own research was conducted.

7. In the title, the authors announce that it will be "... Model for Sustainability of Human Resource Management" - unfortunately, there is no reference to sustainability anywhere in the text. How is the model expected to contribute to sustainability?

8. Please carefully analyze the article by P. Grünwald again - it seems to me that there is some problem of language in understanding.

Good luck!

Kind regards

Reviewer

Author Response

Thank you for your time, your effort and guidance.

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

I appreciate the author(s) for carefully addressing my comments. I have no further comments for the author(s). I wish them the best in their research endeavor.

Author Response

Thank you very much for your valuable comments. We've made improvements in English and added more detailed explanations to the manuscript as your guided.

We appreciate your efforts to our manuscript.

Reviewer 3 Report

Dear Authors

Thanks for the work you put into improving the article.

There are still some minor faults that need to be fixed.

1. Change Aleksandra to Kuzior:

"When automation is mentioned in every sector, there is a fear of unemployment. As Aleksandra  Kuzior stated, every technological development shift human labor and needed skills which are needed because of redesigned job descriptions. She stated that new technologies may create technological unemployment, which is considered as a short-term unemployment and can be overcome by proper education to gain the required specialized competencies"

2. Change Aleksandra and Aleksy to Kuzior and Kwilinski:

Also, Aleksandra  Kuzior and Aleksy  Kwilinski say, in their article, that the social aspect of AI subcategories are not fully comprehended by every individual. Especially depending of age categories, the awareness of AI interaction changes. But they noted that since the industry needs specialists in new technologies, it is crucial to get education"

3. In references, add the name of the journal, now there is twice the title of the article:

Kuzior, A., & Kwilinski, A. (2022). Cognitive Technologies and Artificial Intelligence in Social Perception. Cognitive Technologies and Artificial Intelligence in Social Perception, Management Systems in Production Engineering, 30(2), 109 - 115.

Good luck

Reviewer

Author Response

Dear Editor,

 

   We would like to thank you for your careful review and guidance. We have done the corrections as you requested. We've made improvements in English and added more detailed explanations to the manuscript as you guided.

Thank you for everyting