Roulette or Rigor? Don’t Rely on Luck With Generative AI

By Brice Jaggars, Head of Technology, North America, at Avanade

A global engineering company reduced its RFP response time from 50 to 7.5 hours. A multi-national managed healthcare and insurance company equipped its contact center with pre- and post-call summaries and in-call conversation assistance. A leading global bank empowered business unit "citizen developers" to generate programming code, freeing up IT for more strategic initiatives.

As the generative AI hype subsides, some early adopters have already seen compelling business value in process automation, customer support, and IT operations. However, as with any groundbreaking technology, the implementation of generative AI is fraught with challenges. So, what have successful organizations done differently? Instead of playing roulette with AI, they managed their deployments thoughtfully and thoroughly, steering clear of costly errors and missed opportunities.

Here are the most common mistakes organizations make that prevent AI initiatives from delivering on their promise and how you can avoid them:

1. Lack of clear objectives

One of the most common mistakes companies make is diving into generative AI without clearly defined objectives. The excitement around AI’s potential can lead to focusing on technology for technology’s sake rather than as a means to solve specific business challenges. Without clear goals, AI initiatives often result in efforts that don’t meet expectations. Organizations that don’t know where to begin can work with a partner to create an AI strategy tied to their business strategy. It’s the North Star that will identify the most impactful use cases across the enterprise and ensure that your AI projects are innovative, strategically sound, and poised for success.

Related:Employees Beginning to See Benefits of AI Autonomy in the Workplace

2. Poor data quality

Data is the foundation of any AI system, and generative AI is no exception. However, many organizations underestimate the importance of data quality, leading to AI models that are inaccurate, biased, or simply ineffective. Poor data management can also severely undermine the potential of generative AI. Organizations must build, modernize, and maintain robust data pipelines to support generative AI initiatives. Start with a comprehensive data assessment to identify gaps and ensures that your data is clean, accurate, and well-organized. Adopt a data governance framework for training your AI models on the highest quality data, leading to better outcomes and more reliable AI systems.

3. Unrealistic expectations

While generative AI is a powerful tool, it’s not a cure-all for every business challenge. Many organizations make the mistake of overestimating what generative AI can do, expecting it to deliver results beyond its current capabilities. This creates frustration and, ultimately, project failure. Look for iterative transformation opportunities with small, manageable projects that enable quick wins and gradual scaling. Set achievable outcomes and realistic timelines to build stakeholder confidence in your AI initiatives and ensure they deliver measurable value to your organization. 

Related:AI Quiz 2024: Test Your AI Knowledge

4. Insufficient change management program

Speaking of stakeholder confidence, nothing will derail your initiative faster than end users who don’t understand the end goal and their role in achieving it. Even the most advanced generative AI systems require skilled users to realize their full potential. Without proper training, AI initiatives will likely encounter resistance and fail to deliver the expected benefits. The right partner will bundle a comprehensive program with your solution tailored to your organization’s needs. It should be focused not only on teaching employees how to use the tools effectively but also on fostering a culture of innovation where AI is embraced as an asset. 

Related:Master AI Cybersecurity: Protect and Enhance Your Network

5. Weak security posture

As organizations rapidly adopt emerging technologies like generative AI, it’s essential to consider the immediate, mid-term, and long-term impacts on their cyber resilience and embed security by design throughout their generative AI journey. Regardless of how generative AI is implemented, ensuring it adheres to security, privacy, and compliance regulations is critical. I recommend conducting a comprehensive security assessment to understand the security and compliance issues related to AI and organizational data, then implementing data loss prevention strategies.

Yes, games of chance can be fun, but the successful implementation of generative AI requires a thoughtful, strategic approach. When you put in the work to avoid pitfalls, your organization will unlock its immense potential – from cost optimization opportunities to employee efficiencies – and spark revenue growth in an increasingly competitive landscape. 

About the Author

Brice Jaggars is Head of Technology, North America, at Avanade.