responsible AI: insights into implementing it at your company.

The modern artificial intelligence (AI) revolution has quietly simmered for over a decade, but it’s only recently that highly advanced AI tools have been readily available to non-technical users. 

Generative AI tools, capable of using simple prompts to generate impressive text or image content, have really made a breakthrough. Already, they’re a standard part of the modern professional’s toolkit for brainstorming ideas, creating quick visualizations and getting instant feedback from a digital co-worker.

Non-generative AI has also increased the power of existing digital tools, allowing professionals to use its capacity for spotting patterns and understanding unformatted data to enhance their workflows.

AI is a powerful tool — but great power comes with great responsibility. The limits and risks of AI are well-documented, from concerns over intellectual property issues with generative AI tools, to flawed reasoning patterns that enforce and enhance common human biases and prejudices.

For that reason, companies that use and develop AI tools are striving towards responsible, ethical AI. It’s a mindset that aims to develop the positive uses of AI and make them available to more people, while taking steps to mitigate the serious risks it creates.

AI tools aren’t going away, and in the future, they’ll be even more advanced and widely used than today. So it’s time for every company to start thinking about how responsible, ethical AI can be ensured in the organization. This article shouldn’t be seen as legal or compliance advice, but it will help you understand what responsible AI means in practice and how to encourage responsible AI use in your company.

what is responsible AI?

Although the basic principles are the same, the definition of responsible AI varies. According to Google, which is investing heavily in its own AI tools:

“Responsible AI considers the societal impact of the development and scale of [AI] technologies, including potential harms and benefits.”

For us we use AI according to our AI principles. Randstad uses AI when it's beneficial outcomes for all, human oversight and respect for human autonomy, can give transparency and explainable, fair and inclusive by design, used with security and safety in mind and lastly the those who deploy any AI system is accountable.

In all their AI work, they aim to stay responsible by always focusing on fairness, accountability, safety and privacy. Microsoft, another major driver of the AI transformation, defines responsible AI systems as fair, reliable and safe, private and secure, inclusive, transparent and accountable.

Your company’s exact definition of responsible AI use might be slightly different. But the basic meaning will be the same – AI that is used in an ethical, lawful, positive, inclusive way. It means using AI to complement and enhance human capabilities while reducing the influence of common human biases, rather than using it to replace humans entirely.

In the world of HR, a potential example of responsible use could be using an AI chatbot to scan and adjust your job descriptions to avoid non-inclusive language. However, Amazon’s flawed AI recruiting tool, which the company stopped using after discovering it was biased against female candidates, is not a good example.

how to ensure responsible AI use at your company

Depending on your business and your company’s previous experience with AI, there’s multiple ways to create a common understanding of what responsible AI use is and put it into practice across the company. However, most companies that are already succeeding do it in one or more of these ways:

training in responsible AI use

It’s unlikely you would introduce an important new tool like a CRM or ATS system without providing employees with proper training. So when a new technology like AI starts seeping into all parts of your business, it’s just as important to make sure your team knows how to use it properly and responsibly. 

Since user-friendly generative AI tools went mainstream, the hype has only escalated. Inexperienced AI users may see these tools as completely positive and risk-free, overlooking the potential risks – such as data security issues, intellectual property breaches and unfairness and bias caused among others by poor-quality training data.

Training is an opportunity to dispel the myths around AI, and inform colleagues of the technology’s drawbacks and limitations. When they have all the facts, they’ll be able to take advantage of AI’s benefits, while avoiding its most serious dangers.

Training is also in demand from today’s employees. In Randstad’s Workmonitor survey of over 27,000 workers worldwide, AI was at the top of the list of skills respondents wanted to develop. A PwC survey of over 54,000 workers had similar results, finding a largely positive attitude to AI in the workplace among respondents. Educating employees in key AI skills and how to use the technology as a helpful tool can make them confident, effective AI users and meet their demand for knowledge.

monitor AI use and engage employees

Once again, you wouldn’t introduce a major new tool like a CRM system into the organization and then never check up on how it’s being used. To ensure success, the teams working with the tool would continually adjust their processes and explore new features, figuring out what works and making sure the tool isn’t used incorrectly. Your organization should do the same with its AI tools to ensure responsible use.

On an operational level, designating specific employees as ‘owners’ of the AI tools helps build expertise and create accountability. Ensuring these owners are familiar with responsible AI principles also helps them assess new use cases and prevents the unsupervised use of powerful AI tools in new areas of the company.

On a higher level, regular surveys can help company leadership understand how AI is currently used in the organization and how it spreads over time – ensuring you aren’t taken by surprise if potential AI issues should crop up.

In short, new AI tools shouldn’t be implemented in a ‘set and forget’ manner. Instead, employees should understand the importance of thinking critically about their AI use, and work to improve and develop their use of the tools over time. 

clear responsible AI governance and policies

Having clear AI governance documents in place and a common definition of responsible AI is one of the most important measures. Each person’s idea of ‘responsible’ can differ, and a defined AI policy, endorsed by company leadership and distributed to all colleagues, gets everyone on the same page. It gives team members something to fall back on if they’re not 100% sure that their planned AI task is permissible and responsible.

Our templates for creating your own AI principles and policies are great resources if you’re starting to design your company’s official position on AI. But if you’re getting started and want to make sure you’re asking the right questions about AI, download our flowchart to identifying responsible AI use in your organization. By posing a series of questions, it’ll help you assess specific AI use cases and tell you whether they can be considered responsible – or if you need to do some more work before implementing them in the company.