Survey: Legacy Data Systems Hinder AI Agent Scaling
According to a recent study by MIT Technology Review Insights, which surveyed 300 executives in data and technology, outdated data systems hinder the efficiency of AI agents in numerous organizations. The report, titled "Scaling AI agents with trustworthy data," indicates that these agents have access to only 45% of company data on average, with "data laggards" offering access to 30% or less. Conversely, "data leaders" provide over 70% access and report higher success rates. Trust in AI decisions correlates with data readiness; approximately half of the organizations trust their agents, while all data leaders do. Additionally, only 8% of data leaders face scaling and speed issues, compared to two-thirds of laggards. Within two years, all respondents aim to adopt agentic AI, with 69% anticipating widespread implementation. The report emphasizes the need for improved access to both structured and unstructured data and better governance. It can be downloaded from Google Cloud's website. This content was created by Insights, the custom content division of MIT Technology Review, and not authored by its editorial team.
Key facts
- Survey of 300 data and technology executives
- AI agents have access to average of 45% of company data
- Data laggards provide access to 30% or less of data
- Data leaders provide access to over 70% of data
- Only about half of organizations trust AI agent decisions
- 100% of data leaders trust their agents' decisions
- 66% of data laggards say legacy systems limit scaling, 68% say they prevent speed
- Only 8% of data leaders report scaling or speed constraints
- 100% of respondents plan to use agentic AI within two years, 69% expect widespread use
- Improving access to structured and unstructured data is top priority
Entities
Institutions
- MIT Technology Review
- Google Cloud
- Gartner
- Deloitte