6 ways to turn GenAI adoption into real business value

China leads the world in business use of generative AI, with 83% of organizations saying they are actively using the technology. However, using the technology and using it for specific purposes are two different things. And the US leads the world in full implementation, with 24% (compared to 19% in China).

Researchers from SAS and Coleman Parkes Research came to this conclusion after surveying 1,600 decision makers worldwide who are responsible for strategy surrounding GenAI or data analysis.

The respondents came from a wide range of industries, including healthcare. The companies surveyed employed between at least 500 and over 10,000 people.

SAS will publish the results and an analysis on July 9, and offer tips on how to embed the technology meaningfully into existing or evolving operations. Here are six of them.

1. Use data management tools to ensure that large language models (LLMs) are fed with the highest quality data and prompts – data that is both auditable and traceable.

These tools can ensure user privacy and security by providing robust data protection measures such as data minimization, anonymization and encryption, the report’s authors said, ensuring that sensitive information remains protected. “In addition, workflows can be automated to find the shortest and most direct path to building or optimizing an LLM.” More:

“Organizations should adhere to governance and compliance policies to provide a basic framework for applying data management tools.”

2. Make sure key decision makers have AI knowledge before developing your comprehensive GenAI strategy.

“This takes time and most often requires engaging outside experts to advise your team,” the authors write. Bryan Harris, Executive VP and Chief Technology Officer of SAS, adds:

“With any new technology, companies must go through a discovery phase where they separate hype from reality to understand the complexity of real-world implementations in the enterprise. With generative AI, we have reached that point.”

3. Identify your best GenAI use case to achieve a quick return on investment.

The first step to successfully implementing GenAI is to identify effective use cases for the technology that will help achieve a measurable return on investment as quickly as possible, the authors suggest. Marinela Profi, strategic AI consultant at SAS, explains this point:

“LLMs alone don’t solve business problems. GenAI is nothing more than a capability that can extend your existing processes, but you need tools that enable them to be integrated, controlled and orchestrated. And most importantly, you need people who can use tools to ensure the appropriate level of orchestration.”

4. Make sure your GenAI software vendors enable integration with existing workflow and decision-making platforms.

GenAI is an ideal contribution to hyperautomation, which facilitates the automation of all possible tasks within an organization, the authors explain, adding:

“GenAI excels at aggregating massive amounts of data to support decision-making and enables real-time interactions tailored to your preferred business processes.”

5. To achieve measurable results, use a decision workflow system to integrate GenAI into existing business processes.

LLMs can only perform a few tasks of a use case, the authors emphasize. More:

“Organizations continue to need an end-to-end process that orchestrates the AI ​​lifecycle while improving transparency and governance of LLMs.”

6. Prepare for obstacles.

“Across organizations, the use of GenAI can lead to fears about privacy, security, and lack of governance – along with concerns about technology dependency and its potential to increase bias,” the authors write. “Many of these organizations have not fully prepared to comply with regulations and do not have GenAI governance or ways to monitor the technology.” More:

“Our research shows that companies are rushing into GenAI before they have established appropriate governance systems, which could lead to serious quality and compliance problems later.”

For more information, see SAS’s full research report (contact information required for access) and interactive data dashboard.