Learnings from Amazon HR mishaps

Blind dependence on technology without human interventions can be dangerous.

0
36996

New York Times has found out that Amazon has been firing workers suffering medical problems or undergoing other life crises.

The story reveals that the terminations, which spanned over a period of at least one and a half years, took place when the attendance software mistakenly marked them as no-shows. Findings of the report also identify that the Company was short-changing its workers due to its flawed tech system.

“Amazon’s Worst Human Resources Problem” is what the report was dubbed as.

Is this an HR error or a technological mishap? Or does it point to a more serious issue within the Company?

“Company’s culture may be getting spoilt for the employees due to the work environment they may be operating in. Such instances are just a reflection of how bad the situation may be turning into”

Prabir Jha, prominent HR leader and consultant

Prabir Jha, prominent HR leader and consultant, believes the debacle was caused by a host of issues. He believes that although the use of HR tech is essential for large-scale organisations, their periodic refinement is still something that companies need to be vigilant about.

“For a large-scale organisation (such as Amazon), usage of such software for human resource management is a necessity because when one is looking at scales and numbers, one’d have to follow algorithms. The issue here is to refine the algorithm. Some bugs such as these do pop up from time to time and the importance of a regular fine-tuning of the system should not be overlooked,” he tells HRKatha.

Further, he believes that the responsibility of the entire episode doesn’t solely lie on the HR of the company. “HR is just an instrument of management. It is not just an HR function call. I would find it difficult to believe that HR can take the decision suo moto, to fire people at this scale,” he says.

Jha goes on to question whether the higher ups were even bothered by the mishap or concerned when the story was out in the public domain.

The question that Jha poses indicates that the Company may be dealing with more than just a technological fault. He believes the Company’s culture may be getting spoilt for the employees due to the work environment they may be operating in. Such instances are just a reflection of how bad the situation may be turning into.

“Are companies willing to be self-critical?” questions Jha. According to him, such employee patterns must push a smart HR official to admit to the company’s internal faults and start rectifying them.

“One must understand what to automate and what not to automate. When performance management is automated like payroll, it is a very serious matter and organisations expose themselves to such issues”

Pankaj Lochan, CHRO, Jindal Steel and Power

While Jha believes it to be primarily a company culture issue, Pankaj Lochan, CHRO, Jindal Steel and Power, believes the flaw lies with the utilisation of AI human resource management tools. He clarifies that even though automation can be utilised for a majority of HR processes, “One must understand what to automate and what not to automate. When performance management is automated like payroll, it is a very serious matter and organisations expose themselves to such issues,” Lochan tells HRKatha.

Lochan further says that there cannot be 100 per cent automation of performance evaluation —which ultimately leads to the determination of whether an employee should be terminated or not. Some machine involvement is fine, but complete reliance on it may lead to shortcomings. “There are things that are ever changing and are not part of the algorithm codes. Performance management in changing times, especially the pandemic, is a very different issue which is focused more on the KPI and measurement rather than the number of hours clocked. Had the operator interfaced with the system while it deduced concerned people for termination, such a disaster may have been averted,” Lochan asserts.

Another senior HR leader and practitioner, Adil Malia refers this to as more of a human inadequacy and points out three concerns from the Amazon crisis – programme integrity because there could be chances of human errors in programming, employee sensitivity because it was a case of human insensitivity and lethargy in investigating and redressing people concerns at a faster pace, and last abdication of judgment and blind veneration to systems.

“Sometimes, it makes one wonder if the people values these corporations proudly talk about are empty ‘Employer Value Propositions’ for brand positioning or are they authentic people values the company authentically subscribes to”

Adil Malia, Senior HR leader and practitioner

“If the program ‘specs’ have not been duly audited, errors are bound to crawl in through the loops. One cannot also be venerating so much before digital tools that we literally abdicate human judgment, he says.

Malia’s biggest concern was the way people managers and leaders were insensitive in quick redressal of concerns after they had been highlighted. The skill of the soldier and the sharpness of sword have both to be in unison.

“It is sad that a company of the image and reputation of Amazon, allowed such an issue to fester for such a long time before it realised the mountain of resentment it was facing from its employees,” he quips.

“Sometimes, it makes one wonder if the people values these corporations proudly talk about are empty ‘Employer Value Propositions’ for brand positioning or are they authentic people values the company authentically subscribes to,” Malia questions.

Lochan ascertains that at a tally level, a manual override should always be provided, to rectify the shortcomings of the machine. Such an issue may not come up in other HR processes, but in certain matters as sensitive as an employee’s review, it may make some bad calls. He also says that such issues will be common for any company relying on AI for performance evaluation.

Comment on the Article

Please enter your comment!
Please enter your name here

9 − seven =