Many companies still do not trust their AI systems – and that can be a big problem
- Advertisement -
Related Posts
- Advertisement -
- Companies do not rely on the accuracy of their AI/ML models, but this is due to poor data foundations, report claims
- Only one in three has implemented or optimized programs for data observation
- Observability must be standard over the entire life cycle of data
New research by Ataccama has claimed that a considerable part of the companies still does not trust the output of AI models – but this can be simple because their data is not yet in order.
The study showed that two in five (42%) organizations do not trust their AI/ML model output, but only three in five (58%) have implemented or implemented or.
Ataccama says that this can be a problem, because traditional observability tools are not designed to check unstructured data, such as PDFs and images.
Do you not trust AI? A lack of suitable data can be the problem
The report also revealed the ad-hoc approach that companies often follow, often with perceptibility implemented reactively implemented, which resulted in fragmented administration and silos throughout the organization.
Ataccama defined an effective program such as proactive, automated and embedded in the data life. More advanced perceptibility can also include automated control quality controls and remediation workflows, which can ultimately prevent further problems upstream.
“They have invested in tools, but they have not operationalized trust. That means embedding perceptibility in the full data living cycle, from intake and pipeline version to AI-driven consumption, so that problems can come up and resolved before they reach production,” explained CPO Jay Limburn.
However, current shortages of skills and limited budgets still present challenges along the way. Ataccama also noted that unstructured inputs continue to grow as a result of increased generative AI and rags acceptance, but currently feed less than one in three organizations unstructured data in their models.
The report further explains: “The most mature programs close that gap by directly integrating the perceptibility into their data engineering and governance frameworks.”
With the right perceptibility, companies can expect improved data reliability, faster decision -making and reduced operational risk.
Maybe you like it too
- Advertisement -
