AI in LinkedIn: The Best Way to Connect Recruiters and Job Seekers
We all have used LinkedIn, I believe, regardless of who we are: fresh graduates searching for a job, working professionals willing to switch careers, recruiters searching for talents to join our teams, or even scientists willing to stay up to date with what’s happening around. We can all agree that LinkedIn has always been the greatest platform that helps us showcase who we are, what we are good at, and what we are searching for, no matter what professional goals we have.
The ultimate beauty of LinkedIn resides in connecting professionals from all over the world, allowing every new opportunity to be discoverable by every job seeker and every applicant to be contactable by every recruiter. What makes things even greater is, without a doubt, the huge number of users that are active on the platform. In fact, the number of LinkedIn users has been continuously increasing ever since the platform was created in 2002 and it reached more than 740 million users by the end of 2021. As interesting as this sounds, the abundance of users will have 2 slight drawbacks:
- A lot of jobs might be relevant for one job seeker, and this is because most companies are relying on LinkedIn rather than other places to post their available positions.
- A lot of profiles might fit one particular job offering, and this is because remember: LinkedIn is connecting the WHOLE world!
So the problem is that we will have a lot of available choices when we are making a decision! And this can be more challenging than having little or no choices at all, trust me! It is like being at a road intersection where you have several paths you can take, but you don’t know which one to pick, because all of them look similar, yet some roads might lead to dead ends.
So in the professional world, a person might apply for an “ML Engineering” position in 2 different companies, but get accepted in one and rejected in the other one. Ideally, if this person knew beforehand that he would be rejected in a company, he wouldn’t have applied in the first place, right? He would have focused instead on the companies that are most likely to accept him.
Yeah, it makes sense!
With the same logic, a recruiter might contact 2 candidates for a “Software Engineering” position let’s say, but one of them will be interested in the offer and the other won’t. Again here, if the recruiter knew beforehand that a candidate would not be interested, he wouldn’t have contacted him in the first place, right? He would have focused instead on the candidates that are more likely to respond.
Umm… I see, I see… But Fouad! How can someone possibly know this?! 😲
Well, he can’t know for sure! And for this reason, LinkedIn uses Artificial Intelligence (AI) to help us overcome these kinds of issues, and this is what we will be clarifying through our blog. In the upcoming sections, we will be answering the following questions:
- How does AI help a recruiter find the right candidate for a particular job?
- How does AI help a job seeker find his dream job?
- What are other AI use cases in LinkedIn?
- What are the benefits of empowering LinkedIn with AI?
As you have seen, we’ve got a lot to cover. So let’s dive in!
Let’s start with Recruiters
Let’s imagine that one day, you opened your LinkedIn, and suddenly, out of nowhere, you find such a message in your inbox.
“Hello Mr. Trad! We came across your profile, and we thought you’d be a great addition to our team! If interested, we would like to set up an interview the soonest possible.”
I’m sure some of you have seen someday such a message, and for those who haven’t yet, I hope you see something similar very very soon! But let’s focus for now on what’s more important. When you receive such a message, how do you think the recruiter was able to reach your profile in the first place?! And why was he interested in contacting you?!
Don’t get me wrong, I’m not questioning the quality of your profile at all 😅 But don’t you think that there are other candidates who might be equally important? Why did they choose you? Well, to know the answer to this, you should understand what happens on the recruiter’s end. LinkedIn offers recruiters a tool called LinkedIn Recruiter that they can use in order to search for talents, and yes, you’re thinking right! This tool is powered by AI to help recruiters find candidates who, not only match the job requirements but are also more likely to be interested in the offer. Below, you can find a sample screenshot from a recruiter’s search.
So basically, a recruiter is usually searching for someone located in a specific area around the world, having a specific job title and a specific skill set among other requirements, hoping that this person will fill a particular position. The question is: how would AI help this recruiter find great talents — like you — that are suitable for the corresponding position?
First, it helps with the skills
After the recruiter specifies the title he is willing to hire for, he will start specifying the skills that a potential candidate should have. At this stage, AI will help the recruiter by suggesting relevant skills automatically. Not only that, but the profiles that the recruiter will receive will not only include talents having the exact skills that he entered, but they will also contain candidates having related/similar skills. So, for example, when a recruiter is searching for someone having Machine Learning as a skill, he would also get results for people having skills such as AI, Data Science, Deep Learning, etc.
Nice! But how does this happen exactly?
Well, in simple terms, LinkedIn is not only looking for the specified skills. It looks for skills that are related/similar to each other and to the title or description of the job you have provided.
Wait a minute, let me explain a bit more. In LinkedIn, each entity is defined by what we call an “embedding”. This embedding is just a way of representing an entity in the space. Therefore, each entity would be characterized by “coordinates” in this space, and as a result, similar or related entities would be close to each other. Below, you can find an image highlighting these facts, where we have skills represented via a 2-dimensional embedding space. You can see how skills related to AI are close to each other, yet far from other skills like the ones related to Microsoft Office Products for example.
We specified the skills but … we have a problem!
Too many possibilities, let’s narrow them down
After entering the requirements needed, the recruiter will get a list of profiles that are suitable for the characteristics he specified. But as you know, the list is long! And keep in mind that some of them might lead us nowhere! For this reason, AI will help at this stage narrow down the results to people that are not only suitable for the job but also likely to respond to the recruiter’s message. In other words, results will include people who will most likely be interested in the opportunity once provided!
But how does LinkedIn know this information beforehand?
They simply know it from the behavior of the users. A user might have shown interest in particular positions in the past by clicking on them, applying for them, etc. LinkedIn uses this information to know what opportunities should be more interesting for a particular user.
Okay, so we removed some profiles, what can we do furthermore?!
Relevant users, let’s rank them to speed up the recruiting process
Even if the above process narrows things down for the recruiter, the results are still many, because a lot of people might have the right skills and be interested in the provided position. For this reason, at this stage, AI will help even more by ranking candidates, not only according to how likely they are to respond but also according to what the recruiter wants! When the recruiter gets a list of potential candidates, he will interact with this list by saving, hiding, viewing, or contacting certain profiles. This information will be used to update the order of the candidates in real-time so that the profiles that are most interesting for the recruiter become placed at the top of the search.
But why would the recruiter be interested in someone more than the others, given that he has specified the skills he’s looking for?!
Excellent Question! Although all provided profiles will have some common traits, definitely they will have dissimilarities. No two people can be alike, right? A recruiter might be interested in a profile more than the other for different reasons including the way the profile is organized, the additional skills that a candidate might have, the educational background, previous working locations, and many more. So inside the search results, profiles can be clustered according to similarities as you can see in the figure below.
The more the recruiter interacts with the results, the more AI would learn what the recruiter is really searching for and would recommend the most relevant profiles at the top accordingly.
Hmm… Okay, one last question …
How does LinkedIn know that these recommendations are relevant?!
Another excellent question. Extra points for this one! So by the end of the day, what the recruiter is getting is just a list of candidates that he is contacting. How can we make sure that the job done by AI played its role correctly? What you should note here is that LinkedIn wants to provide recruiters with the best experience, and for this reason, they have to make sure that the recommendations they provide are the most relevant ones, and this information is gonna be used to improve their AI recommendation system!
One metric that LinkedIn looks at in order to improve its talent recommendation engine is calculated through the number of accepted InMail requests. Normally, when the recruiter wants to hire a candidate, he sends him an InMail, which is a message sent to the candidate along with a connection request. When the candidate accepts this InMail and replies in a positive way, LinkedIn considers this as an indication of interest in the provided position, and this would mean that the recommendation of this particular candidate was beneficial for the recruiter.
The more LinkedIn sees accepted InMails, the more they are sure that their AI recommendation engine is doing its job perfectly.
Now, let’s switch Places
Suppose, you are searching for your dream job, and suddenly, out of nowhere, you see a job opening on LinkedIn that matches your ambition.
You can call it “fate” if you want, but AI definitely helped with this fate. How so?
Well, when you are a job seeker, you will normally search for a particular job title available within a particular region/country along with other specifications, and then LinkedIn will provide you with a list of job recommendations. These recommendations will definitely hold the title that you are searching for, but will also include titles that are similar or related. So, if you were searching for a job title like “AI Engineer”, the recommendations might include things like “Data Scientist”, “Data Engineer”, etc. And if you’re searching for titles like “Software Engineer”, you will receive recommendations for additional titles like “Backend Developer”, “Full-Stack Developer”, “DevOps Engineer”, etc.
Yes, you’re right. It’s also about embeddings, where similar titles will be close to each other in the embedding space.
Also, note that LinkedIn will give you the jobs that are supposed to be interesting to you, and that suit your skillset at the same time, and it will rank them accordingly.
So, after all, LinkedIn for a recruiter is not much different from LinkedIn for a job seeker!
That’s the point! The same process happens on both ends. You are searching for the best job that matches your taste and skillset, and they are searching for the best candidates that match the job requirements and are likely to be interested in the offer!
But that’s not all!
Besides the fact that AI helps with what is most important for us on LinkedIn (job search, or candidate search), AI is used in other areas as well.
1. Detecting & Removing Fake Accounts
Like any other platform, anyone can create an account on LinkedIn. The problem is that someone might create a fake account to do illegal acts like scraping, fraud, phishing, etc. AI helps take immediate actions against these fake accounts in order to make sure that all profiles are legit and represent real people that are responsibly using LinkedIn.
2. Ranking the news feed according to your taste!
AI will rank the news feed that you see on LinkedIn, where you get to see things that are most important to you at the very top. How does LinkedIn know this information? It’s of course based on your previous interactions. So, if you showed interest in our Zaka posts for example by liking, commenting, or sharing, LinkedIn would make sure you never miss a thing about us! Even sometimes, you would get notifications about these kinds of posts.
3. More Recommendations
LinkedIn does not only recommend candidates for recruiters or jobs for job seekers. LinkedIn takes things a step further by recommending people and organizations you should follow based on what interests you. They even recommend for you the right courses that will help you gain more skills and advance your career according to the profile that you have. How fascinating!
AI Value… By numbers
AI helped LinkedIn the same way it was able to help any other Tech company. According to LinkedIn, when personalized job recommendations started, they noticed an increase of 30% in the number of job applications, and this number kept increasing year after year. When they introduced AI in recruiters’ tools, the InMail response rates increased to around 45%. All of this makes LinkedIn a great platform that connects recruiters and job seekers more efficiently, thanks to AI.
Conclusion
We’ve seen how important AI is to LinkedIn. According to Deepak Agarwal, LinkedIn’s Vice President of Engineering: “AI is like oxygen at LinkedIn — it powers everything that we do”. The power of AI comes from the massive amount of data generated by a huge amount of users. This “Big Data” is considered as the fuel for AI-driven solutions inside LinkedIn. But definitely, the process is not challenge-free. Implementing these solutions for such a large scale of users is something very challenging on many levels and the most significant challenge is that the solutions might not be always super accurate, and this requires massive testing from LinkedIn’s side to make sure that the algorithms are functioning the way they’re intended to be.
Don’t forget to support with a clap!
You can join our efforts at Zaka and help democratize AI in your city! Reach out and let us know.
To discover Zaka, visit www.zaka.ai
Subscribe to our newsletter and follow us on our social media accounts to stay up to date with our news and activities:
