Retention and reduction of positions have recently seen press and online discussions of the possible applications of “AI-style” processing to address enterprise actions.
In practice, many enterprises (governmental and private) have used intensive rule-based applications and work to determine a reduction in workforces for many years – based on union or other labor contracts. Industries with extensive union labor agreements often see contracts with work rules, talent acquisition, talent reduction, and rules to meet many situations (such as when diversity, equity, or inclusion are explicitly dealt with). [Vehicle manufacturing, railroads, metals/mining, teachers, and government civil service] Automating rule-based applications has been considered expensive and difficult, with economic benefits hard to justify as some contracts are short-lived.
Rule-based applications with a better “understanding” of contracts using AI-style processing have been an approach favored for some processing. The use for workforce retention has been complicated by the need for large amounts of data (qualifications, certification, experience, job position sequencing, performance ratings, etc.)
Integrating the larger volume of data per worker and processing it against all applicable rules with a goal (such as a financial amount to be saved or rules about minimum numbers of people/positions required for regulatory purposes) has previously been considered impractical. Considering complex stated relationships for business processes as factors in reducing the workforce have also been considered impractical. For these reasons, many ‘roles’ are often considered for reduction prior to the more difficult task of assessing, evaluating, and selecting process workers for layoff (such as customer service or field workers compared with many people in the same roles compared to marketing managers). Newer volumes of data for mining (such as logs and audit trails of transactions and usage of an ERP application) can be used as surrogate data trails for process-driven roles.
More difficult to assess are performance metrics for domain-knowledge intensive workers (is a worker making lots of source code updates highly productive, or just incompetent?) or those with extensive how-to skills (a plumber is a lot of money to be saved on a construction team – although very essential).
The assessment and evaluation of workers is a key to redeploying workers to other talent requirements inside the enterprise in the case where some units may need to reduce while others seek to expand. Having common definitions, taxonomies, and equivalences are critical knowledge bases for any application processing to assist management in redeploying talent.
Talent and workforce management applications of AI-style processing are in the early maturities of what is to come.
Possible candidates for AI-style applications must consider many factors to determine if there is a practical application to their workforce management needs.
1) Can work effort in a role be cleanly ascribed to job positions, or are process roles really separate from job positions?
2) Do performance and operational metrics really track with individuals, or just at the granularity of a group or functional designation?
3) Are workers actually capable (cross-trained and current) of filling different business process roles, or is it just an assumption of supervisors?
4) Are business process vulnerabilities to talent changes visible by metrics/costs, or will they be discovered after a RIF?
5) Does compliance (regulatory, accounting, etc.) or conformance (union contract, court order, standard) have rules visible within the talent management application, or will these have to be added to considerations manually?
This is the third post in our AI in HR processes series. You can find the first post here. The rest of the posts can be found at www.ekalore.com/bad-project-blog
Comments