An Ernst & Young report has found that 58 percent of federal respondents turn to internal data repositories as the key source used to train artificial intelligence models and that agencies are looking at various sources of data such as public datasets, third parties and internal data marketplaces to perform data analysis tasks.
EY collaborated with Market Connections to conduct an online survey of 200 federal government information technology decision-makers and influencers responsible for AI policy management in October and found that respondents are using various methods to build and train AI tools.
According to the survey, 51 percent of respondents said they are using supervised learning models and 41 percent noted that they are using natural language processing. The other methods used are hybrid and reinforcement learning models.
The report found that a lack of staff focused on data governance, data security and budget constraints are the top challenges to data discovery and access.
Sixty-five percent of federal IT leaders said their agencies have named a chief data officer, while 14 percent stated they are in the process of hiring a CDO, according to the survey.
The respondents noted that their organizations are exploring, building or using AI systems for about five use cases with data analysis tools, document analysis and process and predictive analytics as the most prevalent purposes.
The report also provided several measures agencies can take to advance AI adoption, such as developing a data management and governance plan to establish a foundation of reliable and trusted data, considering a hybrid approach and assessing and acquiring the right AI tool.
Register here to attend the Potomac Officers Club’s 5th Annual Artificial Intelligence Summit on March 21 and hear federal leaders and industry experts discuss the latest developments in the field.