11 hours ago
Launching GenAI Successfully: From Hype To High-Value
Blake Gibson is both GC and COO of a leading fintech. Blake's legal and technical background give him a unique insight into innovation.
The arrival of GenAI presents a transformative opportunity for industries, promising significant improvements in productivity, innovation and customer engagement. However, GenAI has become like the new 'January gym membership' for some: The money is paid. The desire is there. But the use fades by February.
How can you avoid that issue? Navigating the GenAI hype to achieve tangible value requires a clear strategy, robust governance and a focus on business outcomes to avoid costly missteps. This article provides a road map for successfully implementing GenAI.
Step 1: Architecting A GenAI Governance Framework
The use of GenAI must be rooted in a strong governance framework. Effective governance ensures GenAI is used responsibly, ethically and legally, underpinning the reliability and safety of GenAI systems and fostering trust. Neglecting governance can lead to data leakage, compliance violations and reputational damage. It's a concern that keeps CEOs up at night. True governance also involves cultivating an organizational culture of responsible AI use through continuous education.
A cross-functional steering committee is vital, including legal, compliance, data, cybersecurity and business stakeholders. Clear mandates and regular conversations are crucial. While decentralization is one of the most powerful tools to move initiatives forward, AI adoption is a rare exception. The business impacts of AI are like no other and require careful governance and top-down sponsorship of actual workstreams. Governance must be agile to adapt to rapid GenAI evolution while upholding foundational principles like data security and ethical conduct.
The quality and integrity of data are table stakes ("garbage in, garbage out"). Many organizations overlook data readiness. Key practices include data filtering and anonymization (especially for NPI/PII), access controls, data lineage and metadata management and clear data security principles. Existing data governance frameworks likely need strengthening, with particular attention to unstructured data.
A comprehensive approach to risk management is nonnegotiable, addressing data privacy, cybersecurity, model accuracy, IP infringement, costs and bias. Strategies include bias mitigation through diverse datasets and audits, robust cybersecurity measures, adherence to regulations and IP rights and establishing clear ethical guidelines.
Step 2: Choosing Your GenAI Partner
The booming GenAI market requires careful vendor vetting. Vetting should focus on more than a standard technology vendor assessment. GenAI tools require extra care around use restrictions, data handling and intellectual property ownership. Also, ask about their approach to responsible AI, including bias prevention and ethical standards. A lack of transparency is a major red flag.
Understand precisely where your data goes, who accesses it and who owns AI-generated outputs. Preference should be given to private models. Review service agreements meticulously to avoid issues with data sovereignty and intellectual property. Be mindful of vendor lock-in due to proprietary ecosystems: inquire about data portability and exit strategies. Given the upheaval in copyright laws around appropriate use of pre-existing intellectual property by GenAI vendors, a warranty from the AI vendor (such as Microsoft's Customer Copyright Commitment or OpenAI's Copyright Shield) should be a fundamental expectation until the courts weigh in on copyright law.
Step 3: Deploying GenAI For Maximum Impact
This phase translates strategy into operational reality. As mentioned above, address data quality issues before deployment. GenAI's value is exponentially increased by pairing the GenAI tool with your enterprise's data. The enterprise data needs to be properly tagged and sequestered to avoid sensitive company or client data from being exposed by the GenAI tool to the wrong employees.
Start with small, well-defined pilot projects, gather feedback and incrementally deploy features. Prioritize high-value use cases with measurable impact. GenAI tools can be 'horizontal' in nature (providing support for many functions) or 'vertical' in nature (providing support for discrete use cases or functions). Focus on enhancing existing processes initially ("small t" transformations) before attempting wholesale redesigns.
GenAI augments, not replaces, human judgment. Invest in training on prompt engineering (crafting effective inputs), ethical and responsible use (data privacy, IP rights, bias awareness) and fact-checking/validation of AI outputs to counter "hallucinations."
Address employee concerns and foster a positive adoption environment. The key is to communicate GenAI's goal as augmenting human work by involving employees, soliciting feedback and articulating the "why" behind adoption. Some employees still feel that management is trying to supplant employees with GenAI. That should not be the case. Based on return on investment (ROI), the argument should be that we should hire more, not less, when we deploy GenAI. Our employees' impact with GenAI increases their value across the board.
Step 4: Measuring GenAI Adoption And ROI
The final step is to assess real business value and user adoption.
Define clear objectives to select relevant key performance indicators (KPI). The KPIs should focus on ROI. The ROI can be measured through employee surveys, tracking increases in closed tickets and employee satisfaction. If used properly, GenAI should also enhance client satisfaction survey scores. Avoid over-reliance on purely technical metrics by focusing on business outcomes. Like any software deployment, user adoption metrics are leading indicators of successful integration.
While many report positive ROI, achieving sustained returns takes time and effort. Not all projects succeed, and some initial positive figures may come from early adopters. However, sustained ROI will depend on overcoming challenges like data readiness and skills gaps. The last thing you want is the GenAI tool sitting unused like a gym membership.
Conclusion: Generating Value, Not Just Text
A successful GenAI launch requires diligent planning, robust governance and continuous adaptation. It's about thoughtfully integrating intelligence to augment human capabilities. The future with GenAI is human-guided, ethically grounded and training-driven.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
