How AI Is Transforming Talent Acquisition Teams?
Truth and myths in implementing artificial intelligence for recruitment
Posted on 06-17-2019, Read Time: - Min
Share:
The industry is buzzing with talk about how Artificial Intelligence (AI) and Machine Learning (ML) are going to revolutionize talent sourcing.
HR professionals are divided between those who seek ways to make talent sourcing more efficient and those who fear that AI and ML will take away their jobs.
Some recruitment professionals believe that technology will never be able to replace humans completely and will always be limited, especially when it comes to hiring where subjectivity and human intuition are essential. Yet, more and more recruitment professionals are waking up to the potential benefits of AI and ML to their industry and to their own job.
According to Wikipedia, Artificial intelligence (AI), sometimes called machine intelligence, is “intelligence demonstrated by machines, in contrast to the natural intelligence displayed by humans and other animals”. The term “artificial intelligence” is applied when a machine mimics “cognitive” functions that humans associate with other human minds, such as “learning” and “problem solving”.
In the talent sourcing realm, Artificial Intelligence is used to “mimic” manual operations by a recruiter to generate results that would otherwise require significant amount of manual work. For example, AI could examine a job requirement and rapidly formulate a search on a resume database, using keywords extracted from the job description. This contrasts with the much longer time that it would take a human recruiter to manually formulate and run the optimal search query, especially since this requires specialized training of the recruiter on how to perform these sophisticated searches.
AI can also “read” a large number of resumes and grade them according to how well they qualify for the job. A human recruiter can perform these tasks as well, but the machine can do it much faster and in a standardized way (removing recruiter subjectivity from the process).
Machine learning (ML) is a subset of artificial intelligence in the field of computer science that often uses analytical and statistical techniques to give computers the ability to “learn” (i.e., progressively improve performance on a specific task) from data, without being explicitly programmed for each possible input.
In the realm of talent sourcing, ML can be used to improve resume search by learning from the recruiters’ selections of the candidates they prefer. For example, ML algorithms can analyze a recruiter’s selection of several candidates (a cognitively intense and highly skilled operation) from a list of candidates that came up in their search results to improve the selection search criteria. This can identify a more qualified candidate list for the job (finding ‘more-talents-like-these’). Given selections by multiple recruiters, the algorithms can be further refined and optimized beyond the individual selections made by one recruiter, now based on many searches.
ML is used in applications that incorporate data prediction. For example, if the recruiter is looking for a “Software Developer” that has specific skills (e.g. Java, C++, etc.), then machine learning algorithms will use the job titles used in the searches performed by many recruiters to determine the kind of candidates the recruiter may be interested in, even if they have other job titles such as “Programmer”, “Software Engineer” or “Front End Developer”. Over time, the software can “learn” search patterns and make decisions about the kind of candidates that recruiters are likely to be interested in, even if the recruiters did not specify them in their keyword search.
ML is also making inroads in the important area of predicting the candidates’ likelihood to be interested in moving to a new job. Talent sourcing software can use data from the candidate’s career patterns and social media activity to provide a prediction score of the “candidate’s likelihood to move”. The software can further use feedback from the recruiters after they have communicated with the candidate to further refine the prediction algorithms.
Quality data is key. When searching for talent we must make sure that we are searching on full and qualified data. Talent information may be found in many sources. People leave digital fingerprints on thousands of public pages. They may leave information in their LinkedIn account, Github, Stack Overflow, Indeed.com, Facebook, patent databases, academic publications, meetups, Twitter and many more.
In order to provide a comprehensive search on candidate skills and experience, talent sourcing software must merge candidate data from multiple sources into a single profile (a process called “Data Fusion”).
There are talent sourcing solutions that aggregate, interpret, correlate, and merge data from multiple sources into a fully searchable candidate profile. Obviously, such solutions result in a more comprehensive and qualified database to search on.
Another key attribute of the data is its freshness. Most talent sourcing solutions, other than LinkedIn, do not use fresh information.
Data freshness is critical to the ability of identifying top talent that is ready to change jobs. Candidates (on LinkedIn) update their skills and current positions regularly, and the difference between a current profile to that of 3-12 months ago, could be critical.
Regardless of the variety of the sources for data and their freshness, talent profiles often contain only partial data. This is because people tend not to update their profiles very often, and when they do, they often omit details that they may have momentarily forgotten, or feel are not relevant, or leave them out for a myriad of other reasons. Here, AI can come to the rescue of recruiters by identifying skills that are likely to be missing from the talents’ profiles. The process includes the analysis of thousands of most-similar profiles (“look-alikes”) and identifying correlated common skills. Such skills are then added to the profiles of the candidates (enrichment) making their profiles more complete. This is a giant step in overcoming a major challenge of talent sourcing – dealing with partial data.
Although AI and ML can improve talent search, they are greatly dependent on the quality of the search criteria that are entered by the recruiter. A new generation of talent search tools is eliminating the need for the recruiter to enter search keywords. The ability of AI and ML to analyze similarities (look-alikes) provide an unprecedented capability to formulate a search around a profile as opposed to a job description (which is, primarily, a marketing and a legal document). Imagine a search that begins with a profile of an ideal candidate for a job. Perhaps, someone who is already working in a similar capacity. The AI-powered software will “look” at that profile and extract all the relevant information that this profile contains for the search: skills, career history, education, etc. Then, that information will be used to formulate a search that will result in many more similar profiles.
The next step will involve Machine Learning. The hiring manager will be asked to look at additional profiles and rate them either thumb up or down. The ratings will be used to calibrate and improve the search algorithm and bring them more in line with the specific hiring manager’s preferences and priorities.
AI and ML-powered searching generate much higher quality results that save talent acquisition and hiring manager valuable time. Such searches produce higher quality profiles and increase the available talent pool.
Traditional Boolean search is as good as the keywords selected by the recruiter. But we cannot expect a sourcer to know all possible job titles that are relevant to the role, all synonyms and permutations that skills may have. However, we can expect the AI-powered software to manage and use all the possible keywords by analyzing millions of profiles.
Thus, AI-powered search is likely to use hundreds of keywords and present many more candidates who may use a variety of job titles and skills on their profiles.
AI and ML-based search technologies can largely eliminate the need to train recruiters on how to be search experts. The recruiters just need to specify what kind of candidates they want, review results, provide feedback and then get a targeted list of candidates to contact.
One of the essential capabilities of talent search tools is the ability to grade candidates according to how well they qualify for the job, based on the search criteria. Highly qualified candidates get the highest scores. When you search on LinkedIn, candidates are not graded - you need to sift through many profiles in order to find the most qualified ones.
AI-powered technology can grade, rank and prioritize candidates. The software can even “know” how proficient the candidate is in specific skills is. For example, programmers post projects on Github and such projects are graded by their peers. That information can be collected and used to grade candidates according to their public ratings.
Finally, AI and ML-powered software can help companies minimize bias and improve employee diversity. The software can ignore parameters such as gender and ethnicity, but it can also reflect to the company the specific biases that talent acquisition staff and hiring managers have. It can identify company patterns regarding the candidates that are hired, and the ones rejected and then help them overcome such biases.
Most importantly, AI and ML powered software can save companies much time and costs in the hiring process and make talent acquisition teams more efficient. Talent acquisition time is better utilized when performing critical tasks such as interviewing and qualifying candidates as well as coordinating the processes. Sitting by computers, entering keywords and sifting through thousands of profiles are bad use of time. Machines do a better job at that.
HR professionals are divided between those who seek ways to make talent sourcing more efficient and those who fear that AI and ML will take away their jobs.
Some recruitment professionals believe that technology will never be able to replace humans completely and will always be limited, especially when it comes to hiring where subjectivity and human intuition are essential. Yet, more and more recruitment professionals are waking up to the potential benefits of AI and ML to their industry and to their own job.
According to Wikipedia, Artificial intelligence (AI), sometimes called machine intelligence, is “intelligence demonstrated by machines, in contrast to the natural intelligence displayed by humans and other animals”. The term “artificial intelligence” is applied when a machine mimics “cognitive” functions that humans associate with other human minds, such as “learning” and “problem solving”.
In the talent sourcing realm, Artificial Intelligence is used to “mimic” manual operations by a recruiter to generate results that would otherwise require significant amount of manual work. For example, AI could examine a job requirement and rapidly formulate a search on a resume database, using keywords extracted from the job description. This contrasts with the much longer time that it would take a human recruiter to manually formulate and run the optimal search query, especially since this requires specialized training of the recruiter on how to perform these sophisticated searches.
AI can also “read” a large number of resumes and grade them according to how well they qualify for the job. A human recruiter can perform these tasks as well, but the machine can do it much faster and in a standardized way (removing recruiter subjectivity from the process).
Machine learning (ML) is a subset of artificial intelligence in the field of computer science that often uses analytical and statistical techniques to give computers the ability to “learn” (i.e., progressively improve performance on a specific task) from data, without being explicitly programmed for each possible input.
In the realm of talent sourcing, ML can be used to improve resume search by learning from the recruiters’ selections of the candidates they prefer. For example, ML algorithms can analyze a recruiter’s selection of several candidates (a cognitively intense and highly skilled operation) from a list of candidates that came up in their search results to improve the selection search criteria. This can identify a more qualified candidate list for the job (finding ‘more-talents-like-these’). Given selections by multiple recruiters, the algorithms can be further refined and optimized beyond the individual selections made by one recruiter, now based on many searches.
ML is used in applications that incorporate data prediction. For example, if the recruiter is looking for a “Software Developer” that has specific skills (e.g. Java, C++, etc.), then machine learning algorithms will use the job titles used in the searches performed by many recruiters to determine the kind of candidates the recruiter may be interested in, even if they have other job titles such as “Programmer”, “Software Engineer” or “Front End Developer”. Over time, the software can “learn” search patterns and make decisions about the kind of candidates that recruiters are likely to be interested in, even if the recruiters did not specify them in their keyword search.
ML is also making inroads in the important area of predicting the candidates’ likelihood to be interested in moving to a new job. Talent sourcing software can use data from the candidate’s career patterns and social media activity to provide a prediction score of the “candidate’s likelihood to move”. The software can further use feedback from the recruiters after they have communicated with the candidate to further refine the prediction algorithms.
Quality data is key. When searching for talent we must make sure that we are searching on full and qualified data. Talent information may be found in many sources. People leave digital fingerprints on thousands of public pages. They may leave information in their LinkedIn account, Github, Stack Overflow, Indeed.com, Facebook, patent databases, academic publications, meetups, Twitter and many more.
In order to provide a comprehensive search on candidate skills and experience, talent sourcing software must merge candidate data from multiple sources into a single profile (a process called “Data Fusion”).
There are talent sourcing solutions that aggregate, interpret, correlate, and merge data from multiple sources into a fully searchable candidate profile. Obviously, such solutions result in a more comprehensive and qualified database to search on.
Another key attribute of the data is its freshness. Most talent sourcing solutions, other than LinkedIn, do not use fresh information.
Data freshness is critical to the ability of identifying top talent that is ready to change jobs. Candidates (on LinkedIn) update their skills and current positions regularly, and the difference between a current profile to that of 3-12 months ago, could be critical.
Regardless of the variety of the sources for data and their freshness, talent profiles often contain only partial data. This is because people tend not to update their profiles very often, and when they do, they often omit details that they may have momentarily forgotten, or feel are not relevant, or leave them out for a myriad of other reasons. Here, AI can come to the rescue of recruiters by identifying skills that are likely to be missing from the talents’ profiles. The process includes the analysis of thousands of most-similar profiles (“look-alikes”) and identifying correlated common skills. Such skills are then added to the profiles of the candidates (enrichment) making their profiles more complete. This is a giant step in overcoming a major challenge of talent sourcing – dealing with partial data.
Although AI and ML can improve talent search, they are greatly dependent on the quality of the search criteria that are entered by the recruiter. A new generation of talent search tools is eliminating the need for the recruiter to enter search keywords. The ability of AI and ML to analyze similarities (look-alikes) provide an unprecedented capability to formulate a search around a profile as opposed to a job description (which is, primarily, a marketing and a legal document). Imagine a search that begins with a profile of an ideal candidate for a job. Perhaps, someone who is already working in a similar capacity. The AI-powered software will “look” at that profile and extract all the relevant information that this profile contains for the search: skills, career history, education, etc. Then, that information will be used to formulate a search that will result in many more similar profiles.
The next step will involve Machine Learning. The hiring manager will be asked to look at additional profiles and rate them either thumb up or down. The ratings will be used to calibrate and improve the search algorithm and bring them more in line with the specific hiring manager’s preferences and priorities.
AI and ML-powered searching generate much higher quality results that save talent acquisition and hiring manager valuable time. Such searches produce higher quality profiles and increase the available talent pool.
Traditional Boolean search is as good as the keywords selected by the recruiter. But we cannot expect a sourcer to know all possible job titles that are relevant to the role, all synonyms and permutations that skills may have. However, we can expect the AI-powered software to manage and use all the possible keywords by analyzing millions of profiles.
Thus, AI-powered search is likely to use hundreds of keywords and present many more candidates who may use a variety of job titles and skills on their profiles.
AI and ML-based search technologies can largely eliminate the need to train recruiters on how to be search experts. The recruiters just need to specify what kind of candidates they want, review results, provide feedback and then get a targeted list of candidates to contact.
One of the essential capabilities of talent search tools is the ability to grade candidates according to how well they qualify for the job, based on the search criteria. Highly qualified candidates get the highest scores. When you search on LinkedIn, candidates are not graded - you need to sift through many profiles in order to find the most qualified ones.
AI-powered technology can grade, rank and prioritize candidates. The software can even “know” how proficient the candidate is in specific skills is. For example, programmers post projects on Github and such projects are graded by their peers. That information can be collected and used to grade candidates according to their public ratings.
Finally, AI and ML-powered software can help companies minimize bias and improve employee diversity. The software can ignore parameters such as gender and ethnicity, but it can also reflect to the company the specific biases that talent acquisition staff and hiring managers have. It can identify company patterns regarding the candidates that are hired, and the ones rejected and then help them overcome such biases.
Most importantly, AI and ML powered software can save companies much time and costs in the hiring process and make talent acquisition teams more efficient. Talent acquisition time is better utilized when performing critical tasks such as interviewing and qualifying candidates as well as coordinating the processes. Sitting by computers, entering keywords and sifting through thousands of profiles are bad use of time. Machines do a better job at that.
Author Bio
| Gal Almog is Co-Founder and CEO of Talenya. He spent the last 20 years turning ideas into market leading products. Prior to Talenya he founded PandoLogic. Visit www.talenya.com Connect Gal Almog |
Error: No such template "/CustomCode/topleader/category"!
