Artificial Intelligence: Can it read a resume as well as a human?

 The simple answer is no, but AI has definitely been making strides when analyzing resumes. In recent years, recruiters and hiring managers alike are putting the first line of major shortlisting in the hands of the robots.


Let’s get something very clear – AI has been around for some time, but as our expectations of what machines should be able to do grow, we grow skeptical of whether what we’ve been calling AI is really AI at all.



The reality is, AI has come a long way, but it’s still not perfect. There is an array of resume reading bot services available from a number of innovative vendors in the HR and recruitment space and improvements to these technologies happen often.


The biggest challenge for the machines is understanding the way humans write, which sounds a lot simpler than it actually is, – but when words change meaning and context when used in different sentences it becomes far more difficult for bots scanning resumes to make the right matching choices.


This is where NLP (Natural Language Process) comes in; this is a form of computer intelligence that enables computers to understand human language in its most natural form.  Coupled with good parsing (pulling text from a file) and a large taxonomy (the dictionary the machine has been taught), this enables bots to scan resumes at a high enough level of accuracy to sufficiently replace humans.


I have witnessed firsthand AI match 3 million jobs to 600 resumes per second.  Consider how long and how many humans that would take to achieve those results?  Thanks to technology, we are able to place people in the right job role faster than ever before.


But it’s not all sunshine. Technology still gets it wrong often and even some of the greatest household names in the online recruitment space have delivered highly inaccurate candidate and job matches.  But how do we overcome this and improve on where we are?


Simple… it’s called ML (Machine Learning).


The best bots are ones that have the ability to learn and what better way to teach the machine than monitor the behavior of the user it works for.  In this case, it’s the job seeker and the recruiter.  Every time a job or resume is rejected, the machine takes into account the content of the rejected file and notes it in relations to the file it matched it too.  Make sense?  Or to put it another way, every time the machine delivers a job or resumes that is accepted, the machine takes into account the content of the file it got right when matching, using that as a way of further refining results later on.  This sort of technology already exists and has done for some time.


But where are we going? 


The thing about AI is that it will never be finished – this technology will constantly learn and grow as long as we give it the right inputs to do so.  Today, three major AI improvements are in development:


Sentiment analysis

Graph Database technology

Predictive matching


Sentiment analysis can determine the true meaning and feeling behind the words on a resume.  But why does that matter when words can be manipulated?  While this is true, not everyone goes to those lengths or thoughts to fool ai, especially during the interview stage, computers could even tell your future employer if you are lying.  I kid you not.


Graph Database technology is being used to create employment universes (hear me out), imagine putting all the words ever written into a text cloud and associating all words, strings, and various related words that are seen before and after each word (confusing right).  Well, the outcome is simple, it allows fast profiling and matching of a job seekers resume to an open position.  The market leader of online recruitment in Latin America invested in this technology some time ago and from the looks of it, it’s working very well.


Predictive matching is one of my favorite offerings from AI – this allows machines to predict your career future with a level of percentage accuracy.  But how?  Imagine a database of millions of job seekers resumes, and then imagine how many of them are similar to your own resume –  ten, one hundred, one thousand?  Well in short, based on matching other resumes and then day dating those documents by pulling out significant dates, the machines can work out whether someone is older or younger than you and use the career path for the ones who are older that match your timeline and then make predictive matching career path suggestions to you, with a fairly decent level of accuracy.  For those of you who may struggle with what next in your career path, this is the AI tech for you.


What we can conclude is AI is and will continue its prominent place within the hiring process.  As it grows in sophistication, you will begin to see more use cases beyond just reading resumes.  In the end, AI may even become the one that actually conducts the interview, it might make sense to become fluent in AI sooner rather than later.

Author: Editorial Team

Share This Post On