While 87 percent of companies in the Benelux region will invest in GenAI in the coming year, only 6 percent have already truly deployed and integrated the technology.
That’s according to SAS research conducted by Coleman Parkes Research. 1,600 decision-makers within organizations worldwide were consulted, although only the Benelux results are currently available.
Véronique van Vlasselaer, Analytics & AI Lead South West & East Europe at SAS, sees that almost everyone has some kind of GenAI ambitions. “90 percent are already working on the technology, but we can conclude that people are still experimenting with it right now.” Given GenAI’s recent advance, she says it is not surprising that actual implementation is still some time away.
Still experimenting
The contrast between interest and actual deployment is stark. However, 18 percent of respondents use GenAI themselves on a daily basis, even if in those cases there is no immediate integration with wider business processes.
The promises of GenAI are significant, with organizations already recognizing the benefits. Among professional GenAI users, the results speak for themselves. Every respondent who uses GenAI has noticed an improvement in customer satisfaction and retention, and 82 percent see efficiency gains when processing large data sets. In addition, 71 percent of this group has spotted time savings and lower operational costs.
Obstacles
Despite GenAI’s allure, many roadblocks exist to its successful implementation. These include trouble with the effective use of public and proprietary datasets (62 percent), converting GenAI concepts to practical use (56 percent), and integrating with existing systems and processes (also 56 percent).
All of this points to a somewhat nervous lack of direction: GenAI can provide a competitive advantage, but how do we really get started with it? This is partly due to compliance concerns, something that is a critical issue in critical industries like healthcare. A narrow majority (52 percent) have already implemented an enterprise-wide GenAI policy, while only 44 percent of those surveyed are prepared for current and announced regulations surrounding AI. Internally, management through a comprehensive governance framework remains off for now, with only 9 percent having already implemented it.
Tip: AI Act receives all necessary approvals from Europe
Other pain points include technological limitations (31 percent), lack of clear guidelines or standards (22 percent) and lack of transparency and accountability (18 percent).
Asking other questions
Many companies are currently struggling with the question: what do we do with this technology? Van Vlasselaer says the question of whether a GenAI use case is possible is already not that relevant. “With GenAI both as SAS and as a broader society, we should instead start asking ourselves: should we actually do this?”
“In doing so, it is important that governance and compliance are included in the implementation and processes from the beginning. Developing a forward-thinking strategy and investing in technology that enables integration, governance and explainability of large language models are critical steps that all organizations should take before making binding decisions.”
In it, SAS talks about an “end-to-end” AI lifecycle under the banner of Trustworthy AI. With this service, the company hopes to show others how to implement GenAI the right way, taking into account data privacy, ethical considerations and talent retention from the beginning of the development phase.
Also read: SAS wants to make AI models and AI modeling as accessible as possible