Predictive algorithms, machine learning and Alexa-like ‘intelligent assistants’ will revolutionise recruitment, says Dr Achim Preuss.
We now have more insights at our fingertips than ever to find the perfect candidate. Predictive algorithms can trawl through masses of data and examine multiple variables to provide detailed and specific insights on candidates that human recruiters might overlook.
HR and recruitment data
This has the potential to revolutionise recruitment, because it resolves the problem of information overload. These systems process an unwieldy amount of information into something tangible and useful.
For example, if you search online for “What is gold?”, you’ll see millions of results, including images of gold as well as links to what gold is used for and how it is made.
But if you ask the same question to an intelligent personal assistant (such as Amazon Echo), it will crystallise all of the available information to give you one relevant answer: “Gold is a yellow precious metal …”
Recruiters can now have their own intelligent assistants that will increasingly provide manageable and meaningful information about candidates. This information will be drawn from reactive and non-reactive data.
Reactive versus non-reactive data
Reactive data occurs when an individual reacts or responds to something. Assessments are a good example.
You set a psychometric test, the candidate responds to it and you gain data as a result. Non-reactive data is information that you can obtain about a candidate, without them taking any action.
Much of this data is publicly available online, for example on social media sites, and via the digital footprints that we all leave behind us when we conduct online activity. Marketers have been quick to capitalise on this data.
That’s why you’ll often see targeted internet adverts for products that you’ve looked at previously online.
Now, recruiters can tap into non-reactive data too. Through artificial intelligence and machine learning, algorithms can analyse everything from a candidate’s choice of words, their gestures and the emotional tone of their social media posts.
All of these details can be compiled and combined with reactive assessment data – by your intelligent assistant – to form a psychological profile of each candidate.
In other words, you gain a more holistic view of your candidates – and their suitability for your roles. By reviewing these profiles in the early stages of your selection process, you can narrow down your candidate pool.
Google is seeking to connect job seekers with suitable employers with its Google For Jobs feature, which uses machine learning to match jobs to the preferences of jobseekers.
It’s essentially a search engine that collects job listings from across the web and eliminates duplicates.
But for this to work effectively there needs to be a standardised way of classifying and describing job roles that is internationally accepted. That’s not available yet but no doubt it will come.
By searching non-reactive data, your intelligent assistants will effectively be able to act as “e-headhunters”.
For example, you could ask your intelligent assistant to find you a good systems engineer.
The algorithm would understand the skills and traits that are required and it would source a shortlist of possible candidates who could be approached. It would provide you with an initial report of the insights that it has obtained about each candidate from their publicly available data.
This won’t include their browser history. A person’s browser history is linked to their IP address (the ‘address’ assigned to the device on which they access the internet). Viewing history is stored within users’ internet browsers.
This means it isn’t currently possible to know exactly who is browsing your careers site. However, you’ll know what other sites that individual has visited before they came to you.
With these insights you should be able to make some assumptions about them. The benefit of this is that you could highlight different jobs to different people when they view your careers page, based on what you’ve deduced about their interests.
What about privacy?
There will undoubtedly be concerns about privacy issues and the legal aspects of uncovering publicly available data about candidates. However, companies already conduct pre-hire background screening checks on prospective employees.
These often cover sensitive details such as education credentials, employment verification, credit/financial history checks and criminal record checks.
Employers justify and defend these actions by saying they are protecting their work environment, their brand and their reputation. This is no different to trying to improve the quality of your hires by analysing a candidate’s publicly available data.
It’s important to emphasise that non-reactive data should not be used in isolation. It should only be used in conjunction with reactive data to provide a clearer picture of your candidates. Direct measurement of candidates through assessment should still be undertaken.
For example, if I want to know the weight of a person, I’d put them on some bathroom scales. That measurement would be accurate.
But if I supplement this with additional data – such as the fact that they regularly order a large pizza for home delivery – I can then start to make assumptions about what their weight might be in the future. The more measurement points you have, the more accurately you’ll be able to predict future scenarios.
When recruiting talent, you want as much information as you can get, to help you make the right decisions. Already, too much information is publicly available for any human to process.
And the quantity of available data is rapidly increasing. This has created a market for intelligent assistants that can collate and interpret relevant insights into manageable reports, so that a “human” decision can be made about which candidate is best-suited to the role and the organisation.
As yet, no one has developed an assistant who can do all of this. When they do, the world will beat a path to their door.