Recruiting and getting recruited are laborious activities that are filled with time-consuming customs and traditions. A candidate can spend hours applying for one position. Now imagine doing that dozens or hundreds of times. At the same time, an employer with a strong employer brand can get hundreds of candidates per position. Evaluating all candidates can be very laborious.
In my previous job, I was responsible for the recruitment of a trending company with a strong employer brand. The company received close to 500 applicants per each position. Spending as little as one hour per each candidate meant that there was 500 hours of wasted time and effort. Just imagine the amount of wasted time and effort on a larger scale in the global labor market and what could be done with it if there was a smarter and more controlled way to automate the application process.
We believe that recruiting and getting recruited should not be anyone’s job. Identifying and solving these flawed dynamics and understanding how recruitment influences society and the world at large are what inspired me to start this journey. If you are in a hurry, I suggest you go straight to the section “How to handle communication and timing with A.I.?”.
Applying for a position is an unproductive and ungrateful task that the employers constantly fail to recognize. Many candidates leave the recruitment process with a bad recruitment experience, which ultimately damages the employer‘s business and employer brand. Imagine what would happen if the customer service department responded to customer inquiries in 3 weeks’ or even months’ time with a generic letter. We argue that recruiting can be made more human with cleverly placed Artificial Intelligence (A.I.) that handles the communication and timing. Recruiting and getting recruited should not be anyone’s job.
There is another big reason why Artificial Intelligence (A.I.) should be taken into use in recruitment. The labor market is becoming increasingly fragmented, and the unit of labor in the digital era will be smaller than that in the industrial era (i.e. less full-time employment). This means that most of us will be more often in the state of looking for part-time jobs. Consequently, we need solutions that automatically scan the labor market and handle our employment and job portfolio for us – including applying for a job. Employers also need similar A.I. that scans workforce and matches their demand with supply. A.I. can also be of great help for the screening and pre-screening phases.
A.I. helps the recruiter communicate significantly more effectively, more frequently and to more people than it would be possible without it.
We do not think that A.I. can replace human-to-human communication, face-to-face interviews, and the final recruitment decision. In contrast, A.I. helps the recruiter communicate significantly more effectively, more frequently and to more people than it would be possible without it. In that way, the recruiter can maintain candidates’ interest towards the company and, at the same time, make higher quality and more objective hiring decisions.
Let’s ask a basic question: why do we recruit in the first place? The need for a recruitment process (Figure 1) arises from the fact that something needs to get done (1). This need is then interpreted by the hiring manager or recruiter and communicated as a job ad (2, 3). The job ad gets then published to channels that the recruiter thinks are relevant. If the employer gets lucky, relevant candidates happen to be available at the right time and see the job ad, interpret it (4), get inspired by it and actually apply for the position.
A candidate communicates her fit for the position usually with an application (written or video) and résumé or an online portfolio (5). These documents are based on candidate’s reflection on her background and competence (6). This reflection is based on candidate’s self-knowledge (7) and interpretation of the job ad (4). The recruiter then interprets applications and uses them as the primary source of information when she evaluates the candidates (8). Select few are then invited for an interview (9) whereas the recruitment process for most candidates ends here (10). Rejected candidates are typically informed months down the road that they were not selected.
We, humans, are inherently non-linear creatures. Our current state is dependent on past states, and our future is dependent on present and past. Recruitment and career follow the same principles. According to famous samurai Miyamoto Musashi in the book of five rings, there is timing in everything. He separates background timing from timing of an action and emphasizes its role as the main factor in strategy. Basically, this means that one should be aware of the background processes and know when to act depending on these processes.
In recruitment, background timing could mean that the company (or the A.I. of the company) is aware of how candidates evolve and grow relevant for the company even if they were not relevant the day they applied for the job. Employers should nurture all prospects and help them grow relevant with clever automation. From candidate’s perspective utilization of background, timing could mean that A.I. constantly helps her be aware of potential employment and prepare manual and laborious communication (application letters, résumés) for her.
As seen in figure 1, a lot biases in recruitment arise from communication – or the lack thereof. Communication can also be colored by exaggeration and utmost lies that some candidates and employers communicate. In general, candidates who are good at communicating get further in the recruitment process. Also, recruiters who are good at communicating get good candidates. The employer brand is also a form of communication that operates at the level of general awareness and associations related to the employer. We argue that employer brand can and will change as companies treat candidates with respect and care.
We deliver such communication with natural language algorithms connected to a user interface (UI) that enables employers to give fully personal feedback to each candidate.
According to Miyamoto Musashi, one should practice straight, honest, direct and immediate communication (Jiki tsû no kurai, 直通の位). We deliver such communication with natural language algorithms connected to a user interface (UI) that enables employers to give fully personal feedback to each candidate in a timeframe of seconds rather than dozens of minutes.
In near future, the candidate will receive personal status updates throughout the recruitment process and know beforehand her probability of getting the job. Also, employers will be able to automatically generate job ads and post them to relevant channels that attract right types of candidates. Employers will also be able to nurture and automatically follow-up on candidates when similar recruitment needs arise and when candidates have grown relevant to the company.
As for now, candidates are able to send a feedback request to an employer. Later we will enable candidates to apply directly – in a timeframe of seconds rather than hours – to a position that suits their backgrounds by implementing natural language algorithms that generate fully personalized application letters and résumés. This approach does not fully take away the need for human-to-human communication, but it alleviates some of the biggest time-consuming pitfalls in recruitment.
We argue that communication and timing are factors that drive recruitment and the labor market, and will always drive them, i.e. they are timeless. Also the need to get something done (the reason why we recruit) is timeless. Ultimately recruitment is a numbers game, and one can increase the probability of luck (landing a job) by taking action as much as possible. Unfortunately, fragmentation of the labor market increases the workload needed to land a (fragmented) job, but this is where A.I. can really help us focus on tasks that actually add value.
All in all, recruitment is a complex phenomenon and addressing it properly cannot be done in one post. I see how most flaws in recruitment arise from the properties of complex networks (e.g. a small number of employers get almost all candidates), game theory and information itself (especially asymmetry of information in the labor market). These properties can be modelled with neural networks and Bayesian methods, and we will do so. What puzzles me in recruitment is that in science if there is noise the scientist strives to build a model that explains the noise profile as well as possible in order to be able to extract the desired signal. In recruitment, companies typically have a low signal to noise ratio (e.g. 90% of candidates do not fit the position), but hardly anyone tries to find and analyze the noise profile. I will spare this topic for the next post and go deeper into mathematics then.