Generative Artificial Intelligence continues to advance with countless use cases yet to be discovered. Most companies are beginning to leverage its capabilities along with your customers, employees, and competitors. The benefits can include increased productivity, ability to solve complex problems, reduced human errors, enhanced customer engagement and so much more.

According to McKinsey & Company’s “The State of AI in early 2024”, 65% of respondents report that their organizations are regularly using generative AI, nearly doubling from ten months prior. This rapid adoption is expected to continue driving significant business value, particularly in human resources, supply chain management, and marketing.

Welcome to the Wild West

In the rush to embrace AI, it feels a bit like the wild west right now. Poorly designed AI projects can bring new risks, especially with increasing data and legal sensitivities, a more complex regulatory environment, and high market demands. To mitigate these risks and make the most of valuable resources, companies need to adopt a robust enterprise AI governance strategy to fully leverage the potential of AI.

By designing an agile and results-driven enterprise governance approach, you can make better decisions about developing or buying AI tools, prioritizing and managing your AI portfolio, minimizing enterprise risks, and engaging customers and employees more effectively. This approach also leads to wiser investment decisions. Let’s dive into the pillars of effective AI governance and explore how to build an AI governance structure to manage this new level of corporate complexity.

The Pillars of Effective AI Governance

Corporate AI governance is all about setting the rules and frameworks that guide the development, deployment, and use of AI within a company. It ensures that AI products and services are not only innovative and valuable but also ethical, transparent, and accountable.

Photo by **Sebastian Pichler on Unsplash**

Ethics and Bias Mitigation

Ensuring fairness and mitigating bias in AI products and services not only helps avoid legal issues but also is the right thing to do. AI should be a force for good, not a tool that perpetuates biases.

Amazon encountered a significant issue while developing a recruiting engine designed to filter job candidates. Unfortunately, the tool exhibited a built-in gender bias that favored male candidates over female ones. The algorithm was trained on a decade’s worth of successful candidate data, who were predominantly men. Consequently, it penalized resumes containing the word “women,” such as “Women Chess Club Captain.” Due to this evident bias, the tool lost favor from management and was ultimately scrapped. Dr. Joy Buolamwini, author of Unmasking AI, calls these data challenges “power shadows” which occur when societal bias or systemic exclusion are built into the data.

Timely testing and reviews with input from Legal, HR, Marketing, Compliance and other key team members should be built into your governance process to ensure ethical product development and minimized ethical risks which can be a PR nightmare.

Transparency and Explainability

Transparency isn’t just a buzzword — it’s the bedrock of trust. AI choices and approaches need to be explainable, so stakeholders understand and trust the outcomes especially when personal data is utilized.

The loan approval process has historically faced discrimination challenges against protected classes. To build customer confidence and avoid costly legal battles, it’s crucial to communicate how AI tools utilize collected data, the decision-making process, built-in precautions, employee training, and overall results. By highlighting that your tools are designed with a robust governance framework to prevent discrimination, you can assure customers and employees of your commitment to fairness and transparency.

Leadership set the transparency tone on new technologies. With the game-changing opportunities of AI, effective sponsorship, communications, employee engagement and management can build the right culture for AI success or not. Transparency and trust begins internally and is reflected externally.

Value-Based AI Investments

There are numerous AI investment opportunities to explore across your organization. Incorporating smart selection criteria into your governance process will facilitate quicker decision-making and yield stronger financial results. To invest wisely, periodically review market and regulatory expectations, prioritize your current AI projects, review new opportunities, and measure overall outcomes. Your governance strategy should include a proof-of-concept process to ensure customer interest, market readiness, and the capability to effectively develop and launch new AI tools.

In November 2021, Zillow shut down its Zillow Offers program, it’s home-flipping business, and eliminated 25% of its staff. Errors in the machine learning algorithm used to predict home prices had Zillow purchase homes at prices higher than what they could sell them for in the future, leading to a $304 million loss in inventory in Q3 2021. Although they considered tweaking the tool, it was ultimately seen as too risky to continue using.

AI Privacy Protection

Photo by **Bernard Hermant** on Unsplash

Balancing data utilization and innovation with user privacy is like walking a tightrope especially with data breaches and facial recognition concerns. But with the right measures, it’s entirely doable. This is where engaging the right experts on your AI Governance leadership team matters which can include cybersecurity, IT, legal, HR and others.

Microsoft launched a new feature named CoPilot+Recall, which regularly took screenshots of the user’s desktop, and archived all the data. The feature was to be implemented automatically. The idea behind this was to create a searchable database of information for a later date, but in practice, many people were squeamish about having their every move recorded — who’d have thought? As part of the backlash, numerous cybersecurity experts came forward and pointed out that having a searchable archive of a person’s every movement, including pages they’ve visited, forms they’ve filled in, and so on, is a treasure trove to a hacker. So, Microsoft backed down, and announced that when the feature launched, it would be opt-in, meaning users must give their consent before the feature is activated. With governance demanding the right market research ahead of time, this may have discovered sooner.

Practical Steps for Developing AI Governance Strategy

Define Your Enterprise Governance Organizational Structure

Creating an AI governance organizational structure for your company isn’t simply delegating the responsibility to a department. I recommend a two layer governance approach: an AI enterprise portfolio steering committee and a product-specific governance approach that can be built into all your AI products and services.

Your AI enterprise steering committee establishes the core AI guidelines, playbook, accountability and cultural AI mindset for your organization. This enterprise leadership team will oversee the AI portfolio to ensure the core pillars are built into each of your AI product and service offerings and the overall success of the portfolio. Selecting the right leadership team (leaders from the Business, IT, Legal, HR, Marketing, Finance and other key leaders) ensures your AI portfolio is strategic, financial viable and safe.

For your AI product-specific approach, your established AI frameworks built at the enterprise-level will drive alignment and manage risks. Additionally, IT may lead development but you will also need specific governance and sponsorship from the appropriate teams, i.e., HR leadership for any AI recruiting or HR tools. These AI product-specific leadership teams will have established processes to follow (see AI Governance Playbook below) that feed information to the enterprise governance structure for ongoing strategy, adhere to standards and progress measurements. Starting on a solid governance foundation streamlines the decision-making process and moves new and current AI products through a more robust proof-of-concept to sunsetting lifecycle.

Build Your AI Governance Playbook

An AI Governance Playbook will provide the AI Portfolio framework to manage your AI governance stakeholders, systems, processes, and accountabilities throughout the entire AI lifecycle, from product ideation to retirement. With the right processes and safeguards in place, your AI decisions will be faster, value-driven, and balance innovation and risk. While developing this playbook requires an investment of time and money, it will drive alignment, accountability, and results within your teams both in the short and long term.

As you develop your AI lifecycle playbook, consider the following high-level processes:

Stand Up Your Enterprise and Product-Specific AI Governance Structures by aligning on your core leadership team, mission, guidelines, meeting cadences and decision-rights
Map Out Your AI Lifecycle Processes from selection, prioritization, proof-of concept, team development, requirements, testing, pilot, launch, change management, etc.
Develop Your AI Enterprise Governance Review process from meeting cadence, KPIs, dashboards, report-outs, success criteria and more.
Define your Enterprise Change Management approach to communicate and educate your customers, employees, regulatory parties and others about your AI story. This includes your enterprise upskilling plans to develop your workforce AI capabilities.

For each of the processes above, I recommend building this into your AI Governance Playbook and determining stakeholder roles and responsbilities, decision-making rights, guidelines, templates and tools to support each step in the process. Internally publishing this and holding all accountable to this approach will strengthen the process, adoption and overall success.

Engage Key Leadership and Build Your Change Management Approach

Photo by Elissa Garcia on Unsplash

As you build your enterprise AI approach, select the right leaders early from the Business, IT, Legal, HR, Marketing and others to have a seat at the AI enterprise table. Their input will ensure your pillars are covered from all angles. You’ll also need your key product sponsor(s) such as IT leaders for AI cybersecurity tools or Marketing for AI customer engagement tools. They will be the champions to drive engagement and results. Their role can easily be outlined in the playbook above.

Additionally, define your AI customer and employee engagement strategy to prepare for market changes and inform your employees on your plans for the future. Your employees want to be part of the solution and they are nervous about how AI will impact their roles. They are the ones that can bring you great use case ideas. If you educate them on AI opportunities and upskill as needed, then they will have smart AI ideas that could truly benefit the organization in the long-run.

77% of people expressed their apprehension that AI could bring about job losses in the imminent future, indicating widespread concern about the potential impact of technology on employment opportunities. Haan, Katherine, Forbes Advisor ,“24 Top AI Statistics And Trends In 2024”

Microsoft’s “AI at Work Report” reveals that only 39% of employees using AI at work have received training from their companies, and only 25% of companies plan to offer AI training this year.

A formal change management approach should be part of your AI product development. This includes leader engagement, kickoffs, impact assessments, training, communications and much more. Again, this can be built into your AI governance playbook to ensure it’s getting done well.

Monitoring AI Results and Managing Opportunities and Challenges

Photo by Lukas Blazek on Unsplash

Your governance structure will need a robust review process, which should be an integral part of your playbook. Identify your core KPIs, establish product-specific and enterprise-wide governance progress reviews, and build dashboards to track progress. Additionally, ongoing portfolio and product-specific oversight will help keep your AI portfolio compliant with regulatory demands, market changes, and customer needs, ensuring a comprehensive AI lifecycle process.

Conclusion

Taming the wild west of corporate AI investment can seem overwhelming, but with the right AI governance safeguards, you can minimize risks, maximize gains, and achieve peace of mind. Developing governance shouldn’t be taken lightly. Doing it well will save time and money, leading to more fruitful results both in the short and long term. Neglecting governance can lead to significant problems and confusion for your customers and employees. It’s a worthwhile investment. So, dear reader, it’s time to adopt an AI governance approach that will benefit you, your clients, employees, and stakeholders to truly harness the power of AI.

Strategic Governance to Fully Harness the Potential of AI

The Pillars of Effective AI Governance

Practical Steps for Developing AI Governance Strategy

AI Product Adoption Checklist