It’s time for some clarity around the definition of HR automation. A fundamental component in daily HR operations, ERPs, applicant tracking systems, case management software and chatbots, automation has always been a part of the transformation mix.
However in reality, automation is often a badly defined term, as the concept itself is broad ranging. So in an attempt to establish some clarity ahead of my next couple of blogs on HR automation, here's a bit of a breakdown.
1. Workflow and Business Management Systems
Familiar to many of us, these are the systems that drive a process. Putting the right work in front of the right people at the right time. These are typically well understood and most organisations will have some level of implementation of these tools.
2. Robotic Process Automation (RPA)
Capable of mimicking a human agent's interactions with computer systems, RPA removes repeatable, mundane, volume tasks from the human workload. For example, keying employee data from one system to another. Alongside improved productivity and efficiency, it improves accuracy, governance and control. Automated tasks are carried out consistently to fixed rules: an appealing trait for regulated industries.
3. Artificial Intelligence
- Speech – Converting speech to text, and text to speech
- Natural Language Processing – Translation, classification of intent & clustering, information extraction, question answering, text generation
- Planning, Scheduling & Optimising
- Expert Systems – Encoding domain-specific expert knowledge
- Machine Vision – Identifying, classifying and describing images and video
- Machine Learning – The science of getting computers to act without being explicitly programmed, inferring rules and relationships from typically large data sets
These techniques are not really new. They’ve simply become improved and refined, particularly in recent years. For example, machine vision can now operate better than human performance - at a 2% vs 5% human error rate to be exact. Speech algorithms can now accurately transcribe switchboard quality voice at (or very near) human quality at a 6% error rate.
In combination, these may be described as cognitive computing. IBM sums this up neatly in the statement: ‘Search is to data retrieval as Cognitive Computing is to decision making.’
Essentially, cognitive systems can ingest huge amounts of data compared with humans, and then accurately present recommendations, findings or the next best actions back to the end user.
Perhaps one of the most important functions of automation is converting unstructured data. 75-80% of the world’s data is unstructured, translating and transforming it into something that can be processed by systems or people opens up enormous potential usage.
Of course, this post just touches on the current types of automation out there – new approaches and tools are appearing all the time. In my next posts, I'll look at its potential impact on the workforce and what this all means for HR professionals.
Chris McKibbin is the Future Solutions Lead at Capita plc, focussing on the pragmatic application of automation and AI. Connect on LinkedIn or Twitter.
Related articles
