My conversations with the Aderit.ai team didn’t give me rock solid better answers. It gave me a broader landscape of questions to consider. I keep finding this as I try to apply AI in my lab and in my research.
Context is elastic and interconnected. That means that the deeper you look, the more connections you see. The AI enabled automation of tasks in knowledge work rarely addresses more than a single, repeatable process.
Single repeatable processes are not really knowledge work.
Sure, you can make transactional systems intelligent. But, transactions are fundamentally administrative. The actual work is more emergent and malleable than the accounting oriented measurements associated with its measurement.
That probably bears a little parsing.
You can count the number of bottles of olive oil you’ve sold, filled, shipped, stored, lost, returned, recycled, or need to acquire. None of those transactions tells you anything about the olive oil. You can describe and measure the olive oil itself. But that won’t give you much sense of what it tastes like.
Language and measurement are one layer of abstraction removed from what’s happening. That’s great for communication, process improvement, and workflow design. It’s just not the experience itself.
Once you settle into the idea that this is an administrative jungle, it becomes clear that data always includes the bias of the person or function that curates it. As a result, every HR data set contains multiple perspectives on the same data point.
That’s where Aderit.ai shines. The core of their work is the arrangement, maintenance, tracking of all siloed HR data over time. What makes it work is that Aderit.ai never overwrites the source value. Every system keeps its own version with its lineage. Conflict resolves when the question gets asked rather than when the data is loaded
There is no real system of record in HR. There are many silo specific data sets but no single place where they are all reconciled. If that’s true, any use of AI is handicapped by data quality issues.
Aderit.ai offers a single system of record that accounts for the variation in perspective and over time. Job descriptions, for example, have multiple irreconcilable meanings and uses. It is an effective answer to standard data quality issues in HRTech.
That type of data structure is necessary for effective enterprise AI execution, governance, and implementation.
This concludes the deep dive into Aderit.ai as a part of the AI Governance series. I want to thank Bennie Reddin, Roy Altman and Darin Ries. They are the deeply seasoned brain trust at Aderit.ai. We pushed ideas back and forth. Hard. Over the course of many conversations, we took a long journey into the beating heart of the next generation of HRTech.
More to go on Data Quality, Security, Compliance, and the implementation of governance.