Source: bagi1998/iStock/Getty images.
A third of surveyed organizations have adopted generative AI for data science and analytics projects, according to a recent S&P Global Market Intelligence 451 Research survey focused on generative AI for data and analytics. Moreover, GenAI-driven capabilities are proving valuable across a broad range of use cases. These include automated data-driven recommendations and data visualizations; natural language summaries and analysis; and natural language queries and automated code generation. Nonetheless, obstacles remain, with data privacy, security risks and costs as major concerns.
The Take
Generative AI has moved from experimentation to production, and it is providing value across a range of data science and analysis tasks. Many of these use cases already existed but were underpinned by classical machine learning models that lacked the ability to support natural language interactions or to generate text, code and visuals. These differences illustrate why a generative AI-driven approach, underpinned by large language models, promises a superior experience.
However, the GenAI use cases identified remain relatively rudimentary, suggesting the technology has an important role in assisting expert and non-expert personas without replacing human roles. For all its capabilities, GenAI lacks the critical thinking, creativity, nuanced understanding and organizational context possessed by data scientists and analysts, as well as innate human qualities such as common sense. Moreover, data privacy and security concerns must be navigated as obstacles to success, as organizations grapple with unintentional exposure of private data through model outputs, unauthorized use of data collected for other purposes, and challenges with data retention and traceability.
Summary of findings
Generative AI has taken off for data science and analytics, and adoption is set to continue at a healthy pace. One-third of respondents say their organization is using generative AI for data science and analytics use cases. Furthermore, a quarter of respondents say generative AI for data science and analytics is in discovery/proof of concept, and 15% say their organization plans to implement it in the next year.
Organizations find value in using generative AI for many data science and analytics tasks, with no single standout use case. More than half of respondents (54%) cite automated recommendations based on data as one of their organization’s most highly valued generative AI use cases, while 52% cite auto-generated data visualizations. Around half of respondents include automated natural language explanations (50%), natural language queries (49%) and automated code generation (48%) among their most valued use cases.
Automated visualizations are the top-cited high-value use case for the next 12 months, but only by a small margin. Half (50%) of respondents expect that producing automated data visualizations will be one of their organization’s most valuable GenAI use cases in the next year. Almost as many (48%) expect automated recommendations or suggestions based on data to be among the most valuable uses.
Data privacy, security risks and cost are the top three barriers to adoption. Nearly half (46%) of respondents say data privacy is the greatest challenge their organization faces when using generative AI for data science or analytics projects, while 43% say security risks are the greatest concern, and 38% cite cost as the greatest barrier.
Generative AI policies and guidelines exist, but not universally. Slightly less than half of respondents say their organization has a formal policy regarding the use of generative AI tools for data science and analytics. A third of respondents say there is no policy, but guidelines have been communicated. However, a fifth of respondents say there is no policy or guidelines in place.
Strategic data-driven decision-making is increasing again, but still down from 2023. Nearly one in five respondents (17%) say nearly all strategic decisions are data-driven within their organization, and 49% of respondents say most such decisions are data-driven. These findings represent a slight increase from 15% and 45%, respectively, in 2024’s survey. Nonetheless, the latest figures remain below those reported in 2023, when 27% of respondents said nearly all strategic decisions were data-driven and 53% said most were data-driven in another study conducted by S&P Global Market Intelligence 451 Research focused on data science and decision intelligence platforms within data and analytics.
Can generative AI modernize legacy code bases?
Want insights on data platforms and analytics trends delivered to your inbox? Join the 451 Alliance.
