- Companies do not trust the accuracy of their AI/ML models, but it is due to the basis of the deficient data, according to the report.
- Only one in three has implemented or optimized data observability programs
- Observability must be standard throughout the data life cycle
A new Investigation of Ataccama has affirmed that a considerable proportion of companies still does not trust the production of AI models, but this could be simply because their data is not yet in order.
The study found that two out of five (42%) organizations do not trust their results of the AI/ML model, but only three out of five (58%) have implemented or optimized data observability programs.
Ataccama says this could be a problem, because traditional observability tools are not designed to monitor unstructured data, such as PDF and images.
Don’t you trust AI? The lack of adequate data could be the problem
The report also revealed the ad-hoc approach that companies often take, with the observability often implemented reactive, resulting in governance and silos fragmented throughout the organization.
Ataccama defined an effective program as proactive, automated and integrated throughout the data life cycle. The most advanced observability could also include automated data quality controls and remediation workflows, which could avoid further upstream problems.
“They have invested in tools, but they have not operationalized trust. That means integrating observability in the complete data life cycle, from the ingestion and execution of the pipe to the consumption driven by the AI, so that the problems can arise and solve before reaching production,” said CPO Jay Limburn.
However, the shortage of continuous skills and limited budgets continue to present challenges along the way. Ataccama also pointed out that unstructured entries continue to grow as a result of a greater generative adoption of AI and RAG, but currently less than one in three organizations feed unstructured data to their models.
The report continues to explain: “The most mature programs are closing that gap by integrating observability directly into their data and governance engineering frames.”
With adequate observability, companies can expect a better reliability of the data, faster decision making and a reduced operational risk.
