Generative AI, the technology behind large language models (LLMs) like ChatGPT, is gaining significant traction among enterprises. According to a recent survey conducted by Dataiku and Cognizant, 73% of senior analytics and IT leaders at enterprise companies plan to spend more than $500,000 on generative AI in the next year, with 46% allocating over $1 million. This indicates a substantial financial commitment to exploring and implementing generative AI use cases.
However, while organizations are investing heavily in generative AI, there are challenges that come with its adoption. Many enterprises do not have a dedicated budget for generative AI initiatives, instead funding them from other sources like IT or data science budgets. It remains unclear how this allocation of funds affects other departments that could have otherwise benefited from the budget. Additionally, the return on investment (ROI) for generative AI expenditures is yet to be determined.
Despite these challenges, there is optimism surrounding the value generative AI can bring. The advances in LLMs and other generative models continue to progress, suggesting that the added value will eventually justify the costs. As more use cases and applications emerge across enterprises, IT teams will need to monitor performance and cost to maximize their investments and identify any problematic usage patterns.
One of the main challenges faced in implementing generative AI is infrastructure barriers. The survey revealed that most organizations encounter difficulties using LLMs the way they would like due to infrastructure limitations. Furthermore, they face challenges related to regulatory compliance with regional legislation and internal policy issues. Operational costs of generative models also pose a barrier, as token-based pricing models make it difficult for CIOs to manage costs at scale.
Tech stack complications further hinder generative AI adoption, with 60% of respondents reporting the use of more than five tools or pieces of software for each step in the analytics and AI lifecycle. This complexity adds to the challenges faced by organizations.
Data challenges also persist in the era of generative AI. Data quality and usability remain the biggest concerns for IT leaders, followed by data access issues. Many organizations have vast amounts of data stored in incompatible formats and silos, which need to be preprocessed, cleaned, anonymized, and consolidated before they can be used for machine learning purposes. Data engineering and data ownership management continue to be important challenges in this regard.
Despite these challenges, there are opportunities for companies that provide generative AI services. As the technology matures, there will be opportunities to simplify tech and data stacks, reducing integration complexity and enabling developers to focus on problem-solving and delivering value. Enterprises can also prepare themselves for the future wave of generative AI technologies by running pilot projects and experimenting with new technologies. This allows them to identify pain points in their data infrastructure and policies while building in-house skills to harness the technology’s full potential and drive innovation.
In conclusion, while generative AI presents challenges for enterprises, the significant investments being made and the potential value it can bring indicate a strong interest in its adoption. By addressing the hurdles and seizing the opportunities, organizations can leverage generative AI to transform their operations and products.