New research has claimed while most (90%) enterprises are adopting artificial intelligence, outdated infrastructure and significant skills gaps are preventing them from reaping the rewards.
The Cloudera survey of 600 IT leaders from the US, EMEA and APAC highlighted the challenges and common use cases of AI in business, including the disconnect between productivity promises and the reality.
Cloudera’s report indicates many organizations are struggling to fully implement AI effectively, with many citing concerns over security and compliance risks (74%), insufficient training or talent (38%) and high costs (26%).
Companies are struggling to adopt AI effectively
The study is just one of a growing number to acknowledge the importance of a solid data foundation in the deployment of AI. An overwhelming majority (94%) of respondents trust their data, however more than half (55%) expressed extreme frustration with accessing it.
Issues such as contradictory datasets (49%), difficulty in governing data across platforms (36%) and data overload (35%) highlight the need for a more modern architecture to ensure that data is accessible.
Cloudera Chief Strategy Officer Abhas Ricky summarized: “For the majority of companies, the quality of their data is not great, it’s distributed across various infrastructures and not documented in an efficient manner, and we’re seeing the fallout from that presented in the challenges identified by the survey.”
With the right setup in place, AI is commonly used for enhancing customer experiences (60%), improving operational efficiency (57%) and speeding up analytics processes (51%).
The study also noted the wider use of artificial intelligence beyond core IT functions, alluding to its widespread acceptance and the hope that the technology could improve efficiency across teams, particularly customer service.
Looking ahead, Ricky suggests that companies reframe how they look at data in the context of AI: “Instead of bringing the data to the models, enterprises are starting to realize the advantages of bringing AI models to their data.”