'Remarkable': IBM Japan ordered to disclose AI's role in personnel evaluation: report

Landmark agreement gives union monitoring role in ensuring transparency over AI use

'Remarkable': IBM Japan ordered to disclose AI's role in personnel evaluation: report

The Labour Relations Commission in Tokyo has approved a landmark agreement mandating IBM Japan to disclose how artificial intelligence is used in its personnel evaluation process.

The agreement orders IBM Japan to inform the Japan Metal, Manufacturing, Information, and Telecommunication Workers' Union (JMITU) on the kind of data used by AI to make employee evaluations, The Asahi Shimbun reported.

This means the 40 items considered by IBM Japan's AI system in evaluating employees will now be disclosed.

The organisation will also be mandated to explain the relevance of AI to these items in the wage rules and disclose the content of the evaluations indicated by AI for employees receiving low ratings.

For JMITU, the agreement grants it a monitoring role to ensure transparency amid the lack of regulations in place on AI use on labour-management relations.

"This provides a model for a labour-management agreement to protect workers' rights and working conditions by having the union proactively monitor the use of AI and hold the company accountable for its response," the union said as quoted by The Asahi Shimbun. "The results are expected to spread to other workplaces using the same type of AI."

Yosuke Minaguchi, a lawyer representing the labour union, said the significance of the agreement is "remarkable."

"I have never heard of a labour union and a company establishing disclosure rules regarding the use of AI," Minaguchi said as quoted by the news outlet. "The significance of reaching an agreement with a company that is using AI worldwide is remarkable, and it may have an impact on other companies that have introduced AI."

AI use at IBM Japan

The agreement stems from a relief order filed by JMITU in April 2020 after IBM Japan introduced AI in its performance evaluation system in fiscal 2019.

The union claims that AI influences the decisions of the managers, which could be problematic due to privacy issues as well as discrimination and automation bias.

The AI, which is manufactured by IBM Japan, is used in evaluating employees' performance and work attitude, which is reflected in their wages.

A representative from IBM Japan confirmed the settlement but refused to explain the "circumstances that led to the disclosure."

"We will strive to build good labour-management relations in the future," the representative told The Asahi Shimbun.

Recent articles & video

Balancing data privacy and employee monitoring in Hong Kong: Legal insights

LinkedIn pauses collection of users' private data for training GenAI models

MOM, operators weeding out foreigners' misuse of food delivery accounts

Merck, Amex execs warn against corporate retreat from DEI

Most Read Articles

Singaporean Gen Zs show high preference for hybrid work: survey

NWC guidelines: Employers urged to grant LWWs 5.5% to 7.5% wage increase

Worker cries dismissal with notice was wrongful, says 'poor performance' wasn't proved