All Posts

Using Artificial Intelligence To Improve Recruiting

This Article originally appeared in on June 17, 2020.

Hiring has much in common with selling. We want to find the ideal person to sell to (our ideal candidate) and then sell them on the company, the job and the compensation. AI is already revolutionizing ecommerce, sales and marketing; now some of those same approaches can be applied to recruitment.

As it currently stands, recruiting is people intensive. It is foolish to attempt to entirely automate it, since humans best understand the needs of the organization, the requirements for the position, and the judgment to determine culture fit and the candidate’s potential to adapt and grow as the needs of the organization evolve.   

However, AI can enhance and add efficiency to many aspects of the sourcing, screening, matching, management, onboarding, and skill development. 

Here is a table from my recent book, The AI Powered Enterprise, showing how AI improves employee productivity in three areas: 

Three broad areas of application for AI powered HR processes. (reprinted from “The AI Powered Enterprise”)  

AI Powered HR Processes chart


Lessons from eCommerce

Like sales and marketing, hiring is a lifecycle. To make AI work in any context, the organization needs to be clear about the objective – the specific problem being solved. The more we know about our prospect, the better we can target them.  

The following stages describe one model of an employee lifecycle:

Sourcing: Getting job descriptions and openings in front of the right people. Candidate sourcing is analogous to marketing. Just as AI makes marketing more effective when messages are tailored to the right targets, AI can position job openings and tailor them to the right target candidates, resulting in more appropriate candidate applications. AI can programmatically buy advertising that will reach specific demographics, industry segments and interest profiles. AI tools can also adjust phrasing and language in advertisements to optimize engagement. 

Screening: Even with the correct sourcing, candidate screening is labor intensive and tedious.  Basic text analytics tools do pattern matching identifying needed skills and education and screening out those that do not meet requirements. This requires an information architecture approach – defining terminology across skills, experience, education, and so on. More advanced technologies trained on past successful hires can identify candidates by matching resumes against more subtle patterns. The first approach This approach depends on greater amounts of training data – large numbers of resumes that represent the qualifications of ideal candidates as well as historical performance of people with those qualifications.   

Matching: Determining the fit of candidates to jobs requires another level of assessment across multiple areas. Many flavors of assessment can address different criteria such as behavior style, communication approaches, propensity to take risks, learning style and ability, and technical and cognitive skills. The value in these types of assessments is in understanding how candidates will interact with supervisors, managers and coworkers and whether their soft and hard skills will match the nuances of the role, position, and team.  AI compares the candidate’s assessment with the results of reference profiles for a type of job or with past or currently successful people in the same role or position. One would not put an introvert in an outbound sales position or place a creative problem solver who eschews formalities and rules in a position that requires rigor and adherence to process (for example, an auditor or accountant) . There are dozens of vendors that help with this type of matching; some have an industry or even a role specialization (such as assessing the abilities and style of a programmer) while others are more general.  

Management: Just as the sales organization manages a pipeline of sales prospects and opportunities, the recruiting organization manages a talent pipeline.  It’s extremely important to retain the results of interviews and assessments as well as initial and ongoing communications. A great candidate may be interested in the organization (and vice versa),  but the timing may not be right. Candidates compare opportunities in part based on how they are treated during the process. A negative experience will impact not only whether they take a position but how they communicate their experience to peers, colleagues, and the broader market. To improve the candidate experience, the HR staff must acknowledge receipt of their application and promptly communicate status updates. This includes answering routine questions and scheduling and managing interview and assessment logistics. AI tools can partially automate each of these tasks.  Chatbots can provide routine answers to questions, but they should be tested thoroughly and include features that allow escalation to a human. 

Onboarding, training and skill development: AI can help ensure greater job satisfaction and retention, even after recruitment tasks are complete and the candidate handed off to the hiring manager.  The tools that assess and predict job skills and ultimate success can also identify knowledge and skill gaps.  Customized e-learning programs will fill those gaps and AI tools can be used during on-the-job training. (For example, call centers use AI simulations to train new call center representatives). Helper bots can provide real time answers for new workers and AI powered knowledge tools that better leverage organizational knowledge and expertise will increase job satisfaction and retention. 

Tips for Selecting AI Tools

Some of these benefits are relatively easy to attain and can be part of packaged software offerings from vendors with proven track records. Others require more due diligence, are more complex, and may be less mature in the marketplace. Here are five tips to keep in mind when considering AI technologies for recruitment:  

1. AI is not magic. It needs to be understandable from a business perspective. If vendors’ explanations include “that’s proprietary” or “our algorithm does that automatically” without an explanation of the inputs and outputs, push for more concrete answers.  

2. Understand the problem you are trying to solve using AI. Don’t start with the tools and look for problems to solve. Begin with your business challenges and be as specific as you can. It helps to map out the processes and use cases to be supported with AI. You can’t automate what you don’t understand, and you can’t automate a mess. Have vendors demo your scenarios and use cases, rather than theirs. 

3. Define benchmarks for success. What part of the process will be improved and what are the baseline measures? Consider what is being measured today and make sure stakeholders understand and trust those measures as well as any promised improvements.  If people don’t trust the baselines, they will not trust the improvements.

4. Ask AI vendors how they developed their models and validated their data. You want to be aware of potential biases that could cause the AI to misinterpret candidates’ qualifications or inadvertently screen out or bias results for a particular demographic or use case. If the data sources on which the vendor built their tool do not match the diversity or demographics of your targets, that can introduce unintended bias.   

5. For advanced engagement, training, and talent management applications, good quality data is critical. Identify the sources of data and be sure to test vendor solutions using your data and use cases.  

Recruitment and talent acquisition is fundamentally a human endeavor. The nuances of human engagement depend on interactions with people. AI supports people, it doesn’t replace them. Embracing AI-powered HR tools will bring new efficiencies to the process, improve the quality of candidates and help those candidates to be satisfied and stay with the organization longer -- especially if you take a lifecycle approach and apply AI properly at each step.

Earley Information Science Team
Earley Information Science Team
We're passionate about enterprise data and love discussing industry knowledge, best practices, and insights. We look forward to hearing from you! Comment below to join the conversation.

Recent Posts

[RECORDED] Unlock the Value of Data Discovery Using Knowledge Graphs and Hybrid AI

Successful knowledge management, risk management and process automation initiatives use Knowledge Graphs and Text Analytics for data discovery to extract value from documents and transform them into actionable insights and data. Knowledge Graphs aka Semantic Networks are the bedrock of an organization’s Information Architecture - modeling an organization’s products, services and people. Such semantic approaches, leveraging Natural Language techniques, have been the backbone of Text Analytics. Recently, advances in Machine Learning (ML) are augmenting such traditional approaches to create Hybrid AI. Attend the next Earley Information Science webinar to understand the key steps to set up your next data discovery initiative for success using the latest methodology and technologies. We’ve partnered with Expert.AI, a recognized leader in document-oriented text analytics platforms to explain the technical and methodological advances that enable better data discovery.

First Party Data: The New Imperative

The need for accurate data to support digital transformation and the emergence of new restrictions on the use of third-party data have prompted many companies to focus their attention on first party data.

Knowledge Graphs, a Tool to Support Successful Digital Transformation Programs

Knowledge graphs are pretty hot these days. While this class of technology is getting a lot of market and vendor attention these days, it is not necessarily a new construct or approach. The core principles have been around for decades. Organizations are becoming more aware of the potential of knowledge graphs, but many digital leaders are puzzled as to how to take the next step and build business capabilities that leverage this technology.