This horizon scan from the Agile Research Network (ARN) covers insights from the Fifth AAAI/ACM Artificial Intelligence, Ethics, and Society (AIES) conference 2022 in Oxford (United Kingdom). It draws on the conference proceedings and discussions with attendees to identify potential game-changing trends of high relevance to business agility researchers and practitioners. The conference attendance was funded by the Open University’s School of Computing & Communications.
1. Important takeaways
-
The use of Artificial Intelligence (AI)/Machine Learning (ML)-led algorithms by third-party auditors is expected to grow in relation to financial, human resource, and information security audits.
-
The use of incorrect, misleading, or mis-labelled data in algorithmic audits poses risks to businesses. Business agility practitioners need to devise mitigation strategies and innovate existing practices pertaining to data collection, management, and reporting practices related to third-party AI/ML algorithmic audits.
-
ML software is also expected to be used in a range of Human Resource (HR) functions including recruitment, salary negotiations, performance appraisals, promotions, and employee retention (including planned and unplanned employee exits).
-
Incorrectly supervised use of ML in HR functions raises the prospect of increased litigation and adverse publicity.
-
Business agility practitioners tasked with HR and cultural transformations will need to prepare to understand and engage with the role of ML in their or client organisation’s decision-making.
2. Detailed insights
The AIES conference 2022 highlighted two emerging changes which hold potentially game-changing relevance of AI/ML for business agility researchers and practitioners:
-
The growing role of algorithmic audits; and
-
The prospect of increased use of AI/ML tools and techniques for Human Resource (HR) functions.
2.1 AI-based algorithmic audits
AI-based algorithmic audit could herald wide-ranging changes to businesses in terms of how their financial, human resources, and information security audits are conducted. An AI-based algorithmic audit would not only automate the audit process but also collect the data which would be used to perform the audit process. For businesses, this poses a challenge in ensuring that their data is available in a format which the AI/ML auditor can understand. Given the wide variety of data formats, varying degrees of information still held on paper, and the extent to which data is likely to be unstructured (in an electronic sense), businesses will need to be highly agile to adapt the demands of an AI/ML-led audit. This agility appears crucial since third-party auditors are reportedly increasing their use of AI/ML for cost and efficiency reasons as the complexity of electronic data storage, maintenance, and retrieval increases, particularly in the knowledge services sector.
The other challenge posed by the emergence of AI/ML led audit of businesses is the data on which the auditing algorithm has trained. The research shows that the existing algorithmic audits do not appear to have sufficiently representative data to cover the wide range of finance, HR, and information security audit scenarios. This raises the risk that the algorithms rely too much on historical precedent, use incorrect or mis-labelled data, may be too lenient in the assessment, draw incorrect conclusions, or worse yet, make the auditing process highly opaque instead of ensuring rigour and transparency.
2.2 The use of AI/ML in HR functions
Apart from algorithmic audits, the increased use of AI/ML tools and techniques in human resource management functions is also likely to be of interest to business agility practitioners. ML is now increasingly used in some sectors for decision-making related to recruitment, salary negotiations, performance appraisals, promotions, and employee retention (including planned and unplanned employee exits). The susceptibility of ML tools and techniques to bias, historically skewed datasets vis-à-vis gender and ethnicity, and the risk of increased alienation of entry- and mid-level staff from c-suite executives, were just some of the concerns highlighted by the research.
The use of ML without proper oversight is likely to imperil the work of those currently employed in human resource functions. Incorrectly supervised use of ML also raises the prospect of increased litigation, adverse publicity, and risks over-emphasis on cultural conformity due to biased data sets. However, despite the advances in AI and ML, the technology currently lacks the ability to replicate human empathy in any form, a characteristic which is essential to modern HR function. As a result, the research suggests businesses are likely to rely on ML for mainly data crunching in the near-term. HR professionals with proven soft skills and leadership are likely to be in high-demand as the ‘rote’ aspects of HR functions are taken over by ML.
3. Why does this matter to business agility practitioners and researchers?
Increasing reach and use of AI-based algorithmic audits and AI/ML in HR functions hold implications for all businesses, not just business agility practitioners and researchers.
Despite the challenges of algorithmic audits, increased role of AI/ML algorithms in business audits is foreseen by some experts. Business agility practitioners need to consider how specific sectors and agile practices would adapt to the increased norms of transactional transparency, changes to the role of human oversight of audits, and legal compliances in a hybrid machine-human audit system. For business agility researchers, this is an opportunity to examine how agile processes function when human workers rely on machine-led decision-making, specific adaptations to business agility frameworks vis-à-vis the function of hybrid human-machine audit, and post-audit evaluations.
For business agility practitioners tracking changes in HR functions and overseeing cultural transformations, understanding the role and function of ML-led decision-making in any organisational set-up can be expected to become an increasingly crucial part of their jobs. For business agility researchers, ML-led transformation of the HR function is an opportunity to examine the extent to which AI/ML tools aid the pragmatism and flexibility that is at the core of agile frameworks.
Visit their website for more information:
