The Quiet Crisis of AI Bias in Human Resource Management.

  

AI bias refers to the degree to which AI is treated unfairly. There are many reasons for this, such as unbiased training materials or potential algorithmic flaws. But it does make a special and beneficial treatment for certain groups of people. Race, skin color, age, etc.

Therefore, reducing bias in AI is something every organization should do before using AI in other areas to ensure that the technology can be used for best results.

A presentation by Haas University at Berkeley shows that AI systems have a 44% chance of gender discrimination and 26% racism. This is based on a study by Stanford University that says 16 percent of blacks knowing how to speak This question will cause incorrect content to be truncated if this AI is used in the interaction process.

Germany's Bavarian Broadcasting Company has studied the use of artificial intelligence to assist online interviews. Video recruitment AI found the tool to be full of unforgiving pitfalls, such as learning that candidates score higher when they sit in a better room. Or sitting in front of the counter. However, if the applicant wears a hat or bandana, their points will be deducted immediately. This is because the AI ​​collects statistics and finds that successful candidates or those who have been selected in the past are those without a hat. That's why I chose to find people. Not knowing the dress code does not affect the job of the candidate. Even the algorithms used to publish works from famous apps like Facebook have learned Ethical Justice. The system automatically attracts more women.

Another interesting example is an e-commerce company like Amazon (Top Talent) that uses artificial intelligence to assist in the process of reviewing top talent. 1 to 5 stars They hope the tool will help simplify and speed up the HR process.

But after all, such an artificial intelligence is not as perfect as you think. Because the real problem is that AI was created with ideas for men rather than women, allowing it to immediately assess applicants as soon as they see the word "woman" or whether the applicant has graduated from a university. school. These issues caused Amazon to immediately withdraw the device.

The cases above are statistics and examples only. It should be accepted that today's artificial intelligence tools are constantly developing and are faster than before, so many problems can be solved more easily. Therefore, it is extremely important that the HR department can choose AI to meet the needs of the organization. Remember to evaluate legacy equipment for differences to choose the best equipment and meet the needs of the organization.

What are the disadvantages of AI bias in HR or AI bias in HR management?

Nowadays, artificial intelligence is applied to HR in many aspects. The emergence of AI Bias is therefore able to create a wider range of intelligence than ever before. We can separate the issues as follows.

 Recruiting Process

1. Bias Candidate Screening: Artificial intelligence may unintentionally bias people. For algorithmic reasons, unclear keyword input or teaching too much unnecessary information This problem will prevent HR Recruiters from seeing the resumes of skilled people. But there are some details that contradict 'A good thing from the point of view of AI', it's very worrying that talented people are being brushed off unconsciously like this. Especially now that there are fewer employees in the labor market than ever before.

2. Bias Feedback: AI has a duty to feedback information to HR Recruiter whether to give the candidate an opportunity to pass through the selection process or not. Therefore, biased artificial intelligence will not be able to provide information that can be referenced. considered a waste of resources This can lead to many problems if the organization doesn't have someone to seriously oversee the use of technology.

 

Employee Benefits

1. Unequal access to welfare: if the artificial intelligence system is biased The welfare of the organization may be considered based on information to satisfy only one group of people, causing HR to be unable to choose an appropriate training course. Or design a certain culture that corresponds to the diversity of the workforce.

 

2. Reduce the efficiency of designing limited personalization (Limited Personalization): The currently accepted way of designing benefits that is most effective is designing benefits where employees can choose what they want. have privacy It's not a single welfare type, used together (One Size Fits All), which a good AI system will help us analyze the needs of each employee in detail. But on the other hand, if the AI is biased, we might overlook the needs of a group of people altogether. This means that if we trust too much in artificial intelligence The welfare of the organization will not develop as well as it should.

 

3. Diversity Under representation : Even if it happens accidentally But AI processing is often based on past experiences. and added information So if our data is focused on a single group of people and HR doesn't question or consider it thoroughly. There is a very high chance that the organization will only have people in the same way. Which, if this is the welfare that the AI processes accordingly It will be a result that is consistent with those groups. There is no welfare that allows people of another kind to get along perfectly.

 

Building corporate image (Branding)

1. Negative Reputation: Employees who know they have been disqualified for interviews because of an AI glitch may bring the issue to the public. This directly negatively affects the credibility of the organization. This makes it harder for organizations to attract skilled employees. because it shows unprofessionalism

 

2. Making existing employees think that the organization has unfair discrimination (Perception of Unfairness): As I have said many times, good use of artificial intelligence always requires someone to monitor or supervise. how organizations use the results of biased processing It reflects a failed, disorganized work environment that reduces credibility amongst employees. This makes it difficult to govern and direct other tasks in the future.

References ,

Cruiter, Vi. (2022). The legal and ethical implications of using AI in recruitment. [online] HRTech247. Available at: https://hrtech247.com/the-legal-and-ethical-implications-of-using-ai-in-recruitment/ [Accessed 12 Aug. 2023].

‌