Artificial intelligence (AI) in HR remains a mystery to many of us. The application of AI technology in other scenarios is more straight-forward because the goal of AI is clear. For example, AI is used in autonomous vehicles, constantly learning and adjusting to navigate the road better. AI in HR is more complex because its data are very specific and every organisation is unique, going through constant change.
HR AI is tricky and can’t be too scientific
As a Data Scientist I find HR to be one of the most challenging fields in AI for many reasons. One of the main reasons is that HR data is mostly composed of “natural language”, which is prone to interpretation. A difficult task for computers because of localisation, synonyms and homonyms issues.
On top of that HR data is subjective: people describe themselves and their team members differently based on their own agenda and framework. The language used will differ from person to person, manager to manager. In contrast, data in the legal domain is also natural language, however meant to be clear and objective.
HR concepts themselves are quite blurry. Calling myself a “data scientist” is using a label which is far from giving a clear view of what I’m doing. Using skills to understand a person’s role is a far more precise way getting accurate data, but even skills must be interpreted with care. For example, two people with “intercultural management” skills could have opposite views of what it means and what the resulting actions are.
In addition, applying science to human behaviours is hard. The human mind is infinitely more complex than any transport route optimisation or hotel pricing strategy. We will never have enough data to really understand humans.
Let’s be clear, HR AI can’t be too scientific. A common saying in the AI community is that AI is usually beaten by a human expert. It's especially true in HR, automating entire HR processes is unrealistic.
However, it can lead to ground-breaking features in HRIS
From my experience, AI in HR is not intended to make the decisions for the HR teams, as it is too complex. Instead, AI can bring an understanding to some HR concepts (skills, jobs, resumes, profiles, careers, departments, learning and development...) and their relationships (skills on a CV and the level of mastery, skills by roles, links between learning and career paths, careers in certain departments).
This understanding will allow software to have an automatic level of understanding on every data point available across the whole company and:
· improve the search/browsing to display appropriate content
· add recommendation for the different HR use cases
· to better represent the data despite its complexity
The opportunity differs for each user, so let’s look into the details and find out how it could impact employees, HR and candidates:
AI can help users to find the right objects (jobs, mentor, learning, career advice, development path...). It can also allow users to express themselves in their own language (e.g. not to choose skills from closed nested lists). Thanks to this, employees will naturally update their data regularly and this data will open new many use cases, such as transversal team composed based on skills.
Beyond having more autonomous employees, AI can optimise certain processes by accelerating access to the right data (finding who can do x in BU y, who are the best candidates for z?). It can also help to reduce the need for administration as there is less custom framework to maintain.
AI can improve analytics and open the door to the "real" Stategic Workforce Planning: What is the company current situation, what are our business needs, what trainings are useful or not, what action should I take to reduce my skill gap? It also opens the door to advanced analytics in order to respond with data to the company's strategic issues (transformation, new acquisition, downsizing, etc).
Thanks to AI, candidates will give more information to the company in less time, for example what they can and want to do. But it is also true in the other direction; candidates will know what is missing from their skills set to get the job they want. Also it can surface other opportunities, for example, with your profile, you would be among the 10% of the best candidates for our job x, would you be interested? We could also imagine giving a candidate more visibility on a future career in a company, based on examples of similar profiles in the past (and why not have a chat with such employees)?
These are just some of the examples that I have had the chance to work on. Most of the future possibilities of AI are still to be defined. Overall, the HR challenge in the field of AI masks amazing potential. Contrary to traditional sectors where many scientific approaches are already present and quite usual (for example, nobody expected data science to apply statistics to hotel prices), HR is discovering the potential of their data and the use cases that can be made of it. This is a real leap forward for HR, and we can expect the HRIS to be completely redesigned in the near future.
About the AuthorFollow on Twitter Follow on Linkedin More Content by Cyril Le Mat