A FRAMEWORK TO IMPLEMENT EFFECTIVE, SUSTAINABLE HUMAN-CENTERED ARTIFICIAL INTELLIGENCE SOLUTIONS

INTRODUCTION

Humanity has continually developed new innovations and technologies throughout its history from the discovery of fire one million years ago, Gutenberg’s invention of the printing press in 1440, and the rise of the industrial world in the late 19th century to the advent of the internet in the late 20th century.

Each innovation carried benefits but also significant risks to its users and to the community. However, societies have figured out ways to balance the potential to improve our lives and productivity against the need to mitigate societal risks.

Similarly, artificial intelligence (AI) comes with significant benefits and risks to individuals, organizations, and society. According to a research study published in Forbes (Haan, 2023), 400 million workers could be displaced as a result of AI transforming work and workplace from routine, repetitive tasks to more intellectual, white-collar types of jobs. With the help of AI, many jobs will evolve from knowledge to relationship, from data crunching to consultation, from administration to relationships. As a result, the World Economic Forum estimates that 97 million robotization- or AI-related jobs could be created (Russo, 2020).

However, 400 minus 97 equals 303. Does this mean that AI could potentially result in a net job loss of 303 million employees? With approximately 3.4 billion people employed in the world in 2023 (Dyvik, 2023), this is equivalent to one out of 11 jobs potentially displaced by AI.

AI also increases inherent risks to individuals from the invasion of privacy, dangerous mistakes in the medical field (Tracy & Armour, 2023), identity theft, data security breaches, corporate espionage, surveillance, or societal issues such as the perpetuation of bias, discrimination, and decision-making inaccuracies.

AI VERSUS UNIQUE HUMAN SKILLS

According to the International Monetary Fund, occupations in which AI is capable of autonomously completing tasks might face decreased demand for human labor and lower wages (Cazzaniga et al., 2024). Other types of jobs will be less affected by AI as they rely on uniquely human skills and abilities: jobs requiring a particular craftsmanship, dexterity, emotional intelligence, creativity, cultural awareness, critical thinking, relationship building, education, and ethical decision-making—capacities that AI has not yet matched. Whereas AI may automate certain tasks, these positions that draw upon distinctly human qualities are likely to remain in demand.

As AI continues to advance and become more integrated into various industries, regulatory frameworks are emerging to ensure its responsible and ethical use. The European Union defined seven key requirements to assess the risk of AI applications (European Commission, 2021). The “Assessment List for Trustworthy AI” covers seven areas: human agency and oversight; technical robustness and safety; privacy and data governance; transparency; diversity, nondiscrimination, and fairness; environmental and societal well-being; and accountability (High-Level Expert Group on AI, 2020).

Similarly, a U.S. White House 2023 executive order directs 50 federal entities to engage in more than 100 specific actions to implement new AI guidance set forth across eight policy areas (The White House, 2023). Those include AI safety and security, AI innovation and competition, worker support, consideration of AI bias and civil rights, consumer protection, data privacy, the federal use of AI, and international leadership.

The International Organization for Standardization (ISO) and International Electrotechnical Commission (IEC) are two leading global organizations that develop and publish international standards. The ISO covers a wide range of technology and manufacturing, whereas the IEC focuses on electrical, electronic, and related technologies.

Together, they form a strategic partnership to align their standards and improve efficiency. The ISO/IEC 42001 is a joint effort, providing a comprehensive framework of 308 standards for AI management systems. It helps organizations responsibly integrate AI systems, addressing areas such as risk management, trustworthiness, security, fairness, and more.

PREPARING PERFORMANCE IMPROVEMENT (PI) PROFESSIONALS FOR THE FUTURE OF WORK

The rapid proliferation of AI will require us to guide our clients in becoming critical consumers.—Judy Hale, January 2024

The International Society for Performance Improvement (ISPI) carries out periodic job studies to stay abreast of current trends and demands in the PI field. The latest job study, presented at ISPI’s 2022 annual conference, highlighted how critical technology has become in the realm of PI.

In response to the question, “What trends or changes do you expect your consultants to know?” PI professionals from 27 countries selected “increasing dependence on technology (AI, virtual learning, etc.)” as the most significant trend (ISPI Certification and Credentialing Committee, 2022). AI is top of mind for many PI and learning professionals.

To address this market demand, organizations such as ISPI are working to provide their members with the most current tools, evidence-based strategies, and research. This ensures that PI practitioners are well-prepared to navigate the ever-changing landscape of workplace learning and performance. One emerging area at the intersection of these disciplines is AI for PI (AI4PI), the application of AI techniques to PI challenges. Whereas not all PI interventions require an AI component, AI4PI solutions leverage the power of AI to enhance traditional PI methods and drive greater organizational impact.

Figure 1 illustrates how AI4PI solutions are positioned at the intersection of the AI and PI disciplines.

View Figure

• Download as Powerpoint
• Download Figure

By equipping its members with the latest AI4PI knowledge and best practices, ISPI aims to ensure that PI practitioners are well-positioned to lead their organizations through the ever-evolving landscape of workplace learning and PI.

COMMON THEMES AND CONCEPTS AS KEY DESIGN CRITERIA

We analyzed topics, guidelines, and standards at the intersection of ISO/IEC (International Organization for Standardization and International, Electrotechnical Commission, 2023), the White House executive order on AI, and the EU AI key requirements. This analysis resulted in the identification of 10 common themes aligned with ISPI’s performance standards. They are summarized in Table 1.

TABLE 1 Common Themes and Concepts

View Fullscreen

The first column in Table 1 lists the common themes that provide a foundation for guiding the application of PI interventions that include an AI component. These themes establish an initial framework for defining the criteria of effective, sustainable, human-centered AI4PI solutions.

PROPOSED AI4PI FRAMEWORK

Using the common themes as a foundation, we defined criteria and key concepts to guide the design and implementation of AI4PI solutions:

• The first four are overarching principles: They serve as guiding principles for the integration of AI into PI initiatives.

• The next six are actionable steps designed to help PI professionals analyze, design, develop, test, implement, and evaluate AI4PI solutions.

Table 2 summarizes the 10 criteria that form the proposed outline of the AI4PI framework.

TABLE 2 AI4PI Framework Overview

View Fullscreen

The framework applies ISPI’s PI standards to help address the specific challenges of AI4PI solutions. Similar to other technologies in the past, such as the rise of the internet or virtual reality, ISPI helps their members and organizations make a difference by providing tools and strategies for effective and universal improvement. The AI4PI framework aims to provide guidance to performance and learning professionals to navigate the challenges, risks, benefits, and discussions with their clients about the potential application of AI and how to make it work.

The AI4PI framework is intended for use by any professionals who aim to implement AI in a PI or learning initiative. This includes, but is not limited to the following:

• PI or learning consultants working for a commercial, military, or government organization

• Students or academic staff in colleges/universities

• Solution providers or creators who design and offer products using AI and intended to support human performance, such as performance support tools

CONCLUSION

The proposed framework follows ISPI’s established performance standards: This approach aims to equip PI practitioners with the knowledge and tools to responsibly integrate AI into their interventions. The framework covers key considerations, such as maintaining human oversight, protecting data privacy, and mitigating algorithmic biases. In doing so, it assists PI professionals in navigating the evolving landscape of workplace learning and performance improvement.

The AI4PI framework offers a practical guide for PI professionals to effectively utilize AI in their work while maintaining a human-centered approach. AI presents considerable potential advantages; however, it also poses substantial risks, such as the displacement of millions of jobs. Nonetheless, AI will not replace all human skills, especially those that require creativity, emotional intelligence, and ethical decision making.