Vivek Ahuja, VP-IT at rSTARleading business and IT transformation with a focus on manufacturing, energy/utilities and construction.
While generative AI (GenAI) has revolutionized business performance across industries, there are many considerations to take into account before embarking on a GenAI project. Establishing a strong GenAI governance framework is an important strategic move to ensure its ethical and responsible use.
My experience in creating AI governance frameworks includes collaborating with cross-functional teams to create comprehensive policies and procedures that ensure the ethical, safe and compliant use of AI technologies. This includes defining governance structures, developing ethical guidelines, implementing data management practices, and establishing ongoing monitoring and auditing mechanisms.
In running some of the pilots for customers, we encountered several challenges with GenAI that demonstrate the importance of governance:
• The issue of bias and justice: One of the main challenges with GenAI is the unintentional and persistent introduction of biases present in the training data.
• Lack of explainability: GenAI models, especially those based on deep learning, often operate as “black boxes”, making it difficult to understand how they generate results.
• Data privacy concerns: GenAI systems often require large amounts of data, raising significant privacy concerns especially for healthcare or utility client projects, which can lead to legal ramifications.
Let’s look at what it takes to build a successful governance framework.
Be aware of offers, restrictions and regulations
Understanding the capabilities and limitations of specific GenAI initiatives is essential. To do this, you must embrace a culture of continuous learning and development to stay abreast of AI technology advances, opportunities, limitations, risks, regulatory changes and industry best practices.
Therefore, companies embarking on their AI journey should have an AI ethics committee or review board that oversees AI projects to ensure they adhere to ethical standards. Organizations such as Microsoft, Google, IBM, and many state and federal governments have created similar review boards to oversee and govern AI activities to maintain trust and transparency.
These governance frameworks must meet emerging standards and regulations, ensuring the ethical, safe and effective use of GenAI. Additionally, try to choose the right GenAI platform with robust data infrastructure, giving teams more control over GenAI models, data and management.
Bridging the gap between business and technology
All GenAI initiatives should have a business angle, such as solving specific pain points, maximizing cost, or optimizing defined processes. An example might be a customer care center where you are improving the lives of agents and customers by bringing in GenAI.
Start by defining the business problems GenAI can solve, ensuring technical teams understand them, and providing all the resources and guidance needed to develop such solutions.
Hold regular cross-functional meetings and workshops throughout the implementation process to facilitate alignment between teams and ensure AI solutions are technically sound and strategically relevant.
Responsible AU
Responsive AI is the cornerstone of any GenAI governance framework. This means implementing AI solutions that increase productivity while respecting ethical standards and implementing AI systems that are impartial, transparent, accountable and fair.
Create and integrate a clear ethical code of conduct into the AI development lifecycle. My previous article discusses many ethical considerations when working with GenAI.
Human Supervision
As GenAI improves operations, oversight is essential for ethical use. Teams must monitor AI systems, review results for biases, and intervene when necessary. Establishing a board of stakeholders for oversight, employee training, and intervention protocols can help ensure that GenAI complements rather than replaces human expertise.
A good example of how GenAI complements, rather than replaces humans, is through a review board. GenAI is widely known to hallucinate. This can be caused by incomplete or incorrect data or poorly worded questions. However, if you have someone with expertise in the subject entering the requests and carefully reviewing the responses, you can avoid the odd error. Together, the person and AI can produce better results.
Regulatory compliance
The regulatory landscape is evolving, so business leaders must stay up-to-date and ensure compliance. Create a dedicated compliance team, define roles and data access levels, and consult with legal experts to navigate AI regulations and avoid legal issues.
Regulatory compliance for GenAI includes adherence to data protection laws (eg GDPR, CCPA), ethical guidelines, transparency requirements and industry-specific regulations to ensure ethical, safe and fair use of AI.
Risk management
Effective risk management in GenAI governance involves identifying and mitigating AI deployment risks.
Address data privacy, strengthen security and prepare for disruption. Develop a comprehensive risk strategy with continuous monitoring, routine audits and contingency plans to protect the organization and increase stakeholder confidence.
Remember that if your company handles sensitive data such as personally identifiable information (PII) and health data protected by privacy laws (HIPAA), your AI models must follow established security guidelines and protocols to protect such data. Never enter sensitive personal information into a public AI model (such as ChatGPT), where it is added to the information repository used to program the language model. A good rule of thumb to manage risk around the use of AI is to keep anything proprietary and confidential from AI models.
Transparent
IT leaders must make the operation of AI systems understandable and clearly explain AI decision-making processes. The principle of transparency in GenAI ensures that these systems are understandable, well-documented and auditable. Examples include Google’s AI Principles, IBM’s Watson, the European Commission’s guidelines, OpenAI’s model charters, and Singapore’s governance framework, promoting trust and the ethical use of AI.
Use explainable AI techniques, maintain detailed communication and regularly communicate AI policies and updates to all relevant stakeholders.
Technical Infrastructure
A strong technical infrastructure is essential for the success of GenAI. Use scalable cloud platforms, high-performance computing and advanced data storage. Invest in cybersecurity measures to protect against attacks, ensuring the integrity and reliability of AI operations. This improves deployment and management of AI systems effectively.
Establishing AI governance is critical to success
For optimal results, organizations should adopt a phased approach to governance framework development. They must define business objectives, access current AI capabilities, engage relevant stakeholders, and establish clear policies and guidelines. When necessary, they should collaborate with AI governance experts to provide tailored support and insights.
With good governance frameworks, organizations can maximize their GenAI output while significantly reducing constraints.
The Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology leaders. Do I qualify?