Companies are interested in different ways in which they can benefit from generative AI for both business and workers productivity. While the absence of proper planning and the lack of experienced human resource are the biggest hurdles in the way of unleashing the full potential of this asset.

According to a 2024 study by Coleman Parkes Research which was commissioned by SAS, an analytics firm, 300 top decision-makers of US companies who either manage GenAI strategy or data analytics were surveyed. The researcher focused on determining the key investment areas and the difficulties organizations have.

Marinela Profi, a strategic AI advisor at SAS, noted: "Firms are realizing that LLMs by themselves cannot effectuate business solutions." Rather than the AI GenAI only as an additional tool for business owners to reach their goals, they should see it as a very important part for improving and speeding up the existing systems as well as processes. It is fundamental to come up with a visionary plan and integrate technology that simplifies governance, integration, and understandable guidance of LLMs before using them.

Organizations are encountering problems in four main areas of GenAI implementation:Organizations are encountering problems in four main areas of GenAI implementation:

1. Reliability of data utilization and compliance achievement. On the other hand, only 10% of companies have a solid method of identifying and quantifying bias and privacy risks in LLM. In addition, 93% of US businesses do not govern GenAI with a strong structure, thus increasing the chance of breaking regulations.

2. The integration of GenAI with existing systems and processes will enable better workflow efficiency. Many companies state that they have some troubles working with GenAI in their systems right now.

3. Talent and skills. It is obvious that the GenAI does not own any of the in-house expertise. HR departments are having a hard time to find enough good candidates and this makes leaders uneasy because they do not have the skills to fully realize the potential of GenAI.

4. Predicting costs. Leaders find the direct and indirect costs of using LLMs too high. The estimates given by model creators have turned out to be too costly, and the expenses related to private data preparation, training, and managing ModelOps are substantial and complicated.

Profi further stated: "The focus should be on identifying practical use cases that offer great value and meet human needs effectively and on a large scale. With this study, we continue to support organizations in staying up-to-date, investing smartly, and remaining tough. In today's fast-evolving AI landscape, staying competitive heavily relies on adapting to changes."

The findings of the study were revealed at SAS Innovate in Las Vegas, a conference by SAS Software focusing on AI and analytics for business leaders, tech users, and SAS partners.