When most people think of AI, they instantly think of applications like Chat GPT that work off of generative AI. With generative AI new content is created based on patterns from underlying data.
In the workplace, generative AI might be used to create first drafts of press releases or reports, brainstorms ideas, create art for a project, draft training modules or build simulations and support other creative work.
But there is another type of AI.
Definition of Predictive AI
Predictive AI, on the other hand, is focused on making predictions or forecasts based on historical data patterns. It leverages statistical algorithms and machine learning models to identify trends, correlations, and relationships within datasets, allowing it to make informed predictions about future events in broad areas like finance, stock markets, health care, and supply chains.
Predictive AI, when applied to employees at work, involves utilizing artificial intelligence systems to analyze historical data and make informed predictions about future workforce-related outcomes.
In some workplace scenarios, the use of AI can help predict trends that support strategic decision making. For example, predictive models can analyze various factors contributing to employee turnover, such as job satisfaction, work-life balance, or career development opportunities. This helps organizations implement targeted retention strategies to keep valuable employees and forecast future workforce needs. By analyzing historical data on staffing levels, skills, and market trends, organizations can strategically plan for changes in demand, skill requirements, and industry shifts.
Risks of Predictive AI
Predictive AI gets more complicated when it is applied to individuals. In some cases, AI is being used to predict the ability of an individual to perform a specific job. It’s being incorporated into selection assessments, promotional decisions, resume screening, job description writing, candidate attraction and similar human resources practices.
Using AI this way introduces risk because it is making a consequential decision about a person and that AI may or may not have been designed well.
Poor AI design can come from bad training or testing data or poor algorithms or poor modeling and decisions made while designing the AI. Sometimes that means that poorly designed AI is making decisions about who gets interviews, who gets jobs, who gets promotions, etc.
That’s a problem. And a risk.
Four Ethical & Legal Risk Areas
From a high-level perspective, there are four areas to consider from an ethical and legal risk perspective when considering the impact of predictive AI.
The good news is that some of these risks can be mitigated by following good scientific practices, conducting research, asking thorough questions, using inclusive samples and keeping the rights of the individual in mind in all situations. These tendencies come naturally to most consulting psychologists.
Area 1: Fairness and Bias
If historical data used to train the model contains biases, the AI system may inadvertently perpetuate and exacerbate these biases, leading to discriminatory outcomes. The bias in the data may or may not be visible when the AI is first used.
Area 2: Transparency and Explainability
Many AI algorithms operate as complex “black boxes,” making it challenging to understand how decisions are reached. This lack of transparency can be a significant ethical concern, especially when individuals are affected by AI-driven decisions.
Consulting psychologists should advocate for transparency in AI models and demand explanations for how predictions are made to ensure accountability and fairness.
This can be particularly challenging when working with vendors who do not want to share proprietary information.
Area 3: Privacy Concerns
Predictive AI often relies on extensive personal data, raising concerns about individual privacy. It is essential for consulting psychologists to consider the ethical implications of collecting and using sensitive information without compromising individuals’ privacy rights.
Compliance with data protection regulations and ethical guidelines becomes crucial in safeguarding individuals’ privacy in the context of predictive AI applications.
Area 4: Legal Compliance
The use of predictive AI in employment decisions must align with existing labor laws and regulations. Failing to comply with legal standards can result in severe consequences for organizations, including legal actions and reputational damage.
Consulting psychologists should stay informed about evolving legal frameworks related to AI applications in employment and ensure that their practices adhere to these standards.