Generative AI augmenting and accelerating human teams has dominated the news this last year.  But while AI promises to disrupt and transform work in industries ranging from software to banking to manufacturing, it can’t be allowed to race forward without some controls and boundaries. Responsible AI — the practice of designing and implementing AI while maintaining fairness and positive benefits to society — is the foundation to empower companies to innovate while maintaining compliance with regulations and ensuring their AI initiatives have a positive impact.

Responsible AI is a particularly important topic for professional services firms: the accounting, consulting, legal, and marketing firms that provide valuable insight to the world’s largest and most influential companies. Honing their skills in this domain will be important not only for their own internal transformations, but also as they help lead their clients through similar transformations. 

Workera clients in the professional services industry have actively pursued responsible AI skills transformations. Nearly 3,000 of these users have developed their Responsible AI skills using Workera’s platform. In their initial assessments, employees proved to be relatively proficient at skills like AI Fairness (45% Accomplished on first assessment) and AI Reliability (39% Accomplished on first assessment) that require an understanding of both AI technology and the ethics of it, while they lacked the same level of proficiency in technical skills like Deep Learning (8% Accomplished), Machine Learning (9% Accomplished), and AI Explainability (14% Accomplished).

Here’s how these companies have seen their capabilities leap forward in less than a quarter.

What We Know About Responsible AI

Why is Responsible AI so important for these professional services firms? First, a robust focus on these skills can ensure that the organization stays compliant at all times with relevant laws and regulations. Governance and compliance can frequently slow down tech progress. By laying the groundwork with a nuanced approach to Responsible AI, organizations can develop their solutions without hesitation.

Responsible AI also ensures that an organization’s efforts align with their values, as well as the values and expectations of society around them. Here are the four Workera domains that comprise Responsible AI:

• • AI explainability: Can you explain how the AI tool performed its work or reached its conclusions? Many view AI tools, such as generative AI chatbots, as black boxes with algorithms that are impossible to understand. AI explainability ensures that you can help your organization or client organizations come to educated decisions regarding their use of AI.
  • AI reliability: Today’s leading AI solutions are known to make high-profile mistakes, as some report hallucinations, errors, and plagiarism in their generative AI results. Someone who is proficient in AI reliability will be able to accurately determine whether they can rely on the outcomes of their AI tools. In other words, AI reliability means that you know when to trust the outcome and when you need to adjust the tool because of mistakes or inaccuracies.
  • AI accountability: This skill is fundamental to Responsible AI. When an employee is proficient in AI accountability, they understand and take responsibility for the impact of AI systems on individuals and society. Someone who is proficient in AI accountability can, for example, design ethical guidelines for diverse AI applications. Professional services employees must be proficient in AI accountability in order to help their clients safely and responsibly deploy their own AI solutions.
  • AI fairness: The algorithm that informs an AI tool depends entirely on the underlying data, and that data can sometimes be biased or unreliable. Proficiency in AI fairness means the employee is capable of recognizing and correcting for biases or sensitive variables in the decisions made after the machine learning process.

The users followed training programs prescribed by Workera’s AI-driven platform in order to improve their Responsible AI skills. The results have been impressive. Users who measure their baseline skills, learn, and then reassess have seen their skills proficiencies improve by more than 50% in AI Explainability and AI Fairness — most in less than 90 days of learning between their first and latest assessment.

Workera also tracks learning velocity: the rate at which a user is able to develop a specific skill. Among users in the professional services industry, the average score increase in Responsible AI was 46.7% per domain. This increase was obtained, on average, in 70 days of active learning. For professional services organizations, that means they can reliably increase the number of AI domain experts in their teams in around three months of upskilling.

How AI Mentorship Drives Development

Professional services companies are clearly motivated to upskill their workforces quickly in Responsible AI. These rapid improvements can only be achieved when guided by an AI mentor and a personalized, adaptive learning plan.

Why is self-guided learning inherently flawed? This approach will always be less efficient than AI-driven learning, because users will lose time by studying the wrong material, sequencing their learning in the wrong order, and misunderstanding how deeply they need to understand each skill for their role.

When conducting self-assessments, humans are highly likely to underestimate or overestimate their skill levels — both of which can cause their learning to be inefficient. According to Workera data, just 11.4% of users accurately estimate their abilities in a given skill. More than half of users (56.2%) underestimate their abilities, leading them to waste time studying something they already know, while 32.4% overestimate their abilities and fail to study the skills they need.

How Skills Signals Inform Staffing for Professional Services Firms

When an organization takes an informed approach to learning and development, they don’t just improve their overall abilities and learning velocity. An AI-backed platform with adaptive learning also allows the organization to make inferences about their workforce and to make better staffing decisions.

For example, the Workera platform data for Responsible AI indicates strong correlations between abilities in different sets of skills. Some of these pairings are obvious: abilities in Deep Learning are highly correlated with abilities in Machine Learning. However, the data also reveals other useful and informative associations. Users who have high proficiency in Machine Learning are also highly likely to be proficient in AI Explainability — the strongest correlation among the skills included. This could indicate that proficiency in Machine Learning demonstrates a highly nuanced understanding of AI as a whole, enabling these people to explain the technology to others.

These correlations allow business leaders to make educated guesses when assigning employees to projects or considering which candidates to hire. If they know a potential employee is highly skilled in Deep Learning and Machine Learning, they can expect that the candidate will have high potential in AI Explainability. 

This skills-based approach gives organizations flexibility and confidence as they put together teams to accomplish certain tasks. When the majority of your employees are Accomplished in Responsible AI, you are no longer constrained by job titles and descriptions. A data engineer or machine learning engineer might not be considered a typical contributor for certain projects based on their title; with verified skills data, a business leader will know they can rely on that person to execute the project.

Companies around the world are investing heavily in AI for their product roadmaps and business strategies. These organizations have a fundamental need for Responsible AI, and they must be able to develop these skills as quickly as possible in order to safely and effectively drive innovation. In this highly competitive market, businesses and employees cannot afford to leave their learning and development up to chance — an AI-driven approach to upskilling is the most effective way to maximize results and help employees reach their potential in this critical technology area.