The Economic Times daily newspaper is available online now.

    How artificial intelligence is reshaping recruitment, and what it means for the future of jobs

    Synopsis

    The recruitment industry is undergoing a big transformation, and artificial intelligence has been one of the biggest drivers behind this change.

    ET Bureau
    Anuj Agrawal, 35, has dabbled in the recruitment industry since 2005. With 100 employees and two offices in Noida and Bengaluru, his firm Zyoin offers recruitment and consultancy services to over 300 companies, including Amazon, Goibibo, Play Games and PayU. But there were some constant niggles. Like parsing resumes. With no universal template around which resumes are written and structured, mining and matching thousands with job positions was a huge task. Available parsing technologies were basic and didn’t sort and match well. Also, the resumes in their database would often get dated.

    Last year, Agrawal got a cold email from Anand Kumar, founder of Bengaluru-based Skillate, an artificial intelligence (AI)-based recruitment solution platform that helps companies read and match resumes. Early this year, SAP Labs picked Skillate as part of their incubator programme.

    Image article boday
    Agrawal tested Skillate’s platform for a month and figured his staff ’s efficiency rate improved sharply — while earlier they were vetting 500 resumes to get 10 right profiles, they were now averaging 50. “Their resume parsing and matching technology was very good. Our search began to throw up more relevant candidates,” says Agrawal who is among Skillate’s growing list of loyal customers. He is now looking forward to Skillate’s next feature — scanning candidates’ profiles and posts on social network platforms like Facebook, Linked-In and GitHub to automatically update old resumes. “The feature for automatic resume updation is ready. We are now trying to automate both updation and matching,” says Kumar.

    AI or artificial intelligence is the new buzzword in the corporate world. In the networked digital era amid proliferation of smart devices and surge of Big Data, AI — or smart machines that can think intuitively and make intelligent sense of the vast data — is the new battleground that is roiling multiple sectors, disrupting companies and stoking new rivalries. Serious enough for tech world’s two titans — Tesla’s Elon Musk and Facebook’s Mark Zuckerberg — to spar over it. Digital giants Google and Amazon are now duelling with their AI-based virtual assistants. Think of driverless cars and robotised assembly plants that are disrupting the automobile industry and the manufacturing sector.

    Meet the New Hiring Manager
    That disruption is now shaking up the world of work, too. The coming job losses is only part of the story. Automation (largely driven by AI) threatens 69% and 77% of jobs in India and China respectively, says a recent World Bank research. It will reportedly take away 30% of the jobs in the banking sector globally. But that’s just part of the story.“Tools and processes involving AI will proliferate and as they do, jobs will evolve. I think ‘Robot Process Automation’ is the next big thing, which means everything that can be automated will be automated,” says Jim Stroud, global head of sourcing and recruiting strategy, talent innovation centre, Randstad Sourceright, US.

    AI will also dramatically change the way candidates seek work and employers discover them. Thanks to a slew of mushrooming AI-based startups like Skillate, both
    in India and overseas. Bengaluru-based three-year-old Belong scans a range of social
    Image article boday
    networks and databases — from LinkedIn to Facebook to GitHub and ResearchGate — matching a company’s hiring history and position requirement to candidates’ profiles, many of them passive job seekers. Belong claims to save 15-20 hours of a recruiter’s work every week.

    Clients have few qualms endorsing the startup. “Belong has helped us reach out to candidates otherwise not available. Its very personalised emails going to candidates is a unique feature,” says Karthik Purushotham, India talent acquisition leader for PayPal, which has been a Belong customer for a year now. Adds Ameya Ayachit, head of talent acquisition, Directi Internet Solutions: “Belong’s engagement piece through personalisation has increased the percentage of responses we get from candidates. We have had lot of candidates write back appreciating those customised emails.”

    In another southern city, Hyderabad, Hari Krishna M set up Stockroom.io in 2015. Since then, the curated platform for developers has helped organise over 115 coding challenges and hackathons for 32 companies, including Amazon and Microsoft, to hire top-end product development talent. Last year Hari Krishna cofounded Param.ai, which automatically pre-screens resumes that land on the company’s careers’ page. “Param tells the company if the candidate is good, bad or average depending on its past hiring patterns. It automates initial screening,” he says.

    It has another product called Retarget that mines the company’s database of resumes, and sends out automatic messages asking candidates to update them; this helps keep the database fresh and current. Next up is Param 2.0, where it is developing chatbots that will automate features like prescreening candidates. The programme will ask and answer some basic queries in audio files, which will automatically be converted into text files for recruiters to take forward. Mumbai-based Klimb.io, founded in 2015, wants to tackle the problem of ‘no shows’, where candidates take the job offer and don’t join. Founder Prashanth Thiruvaipati wants to use AI to combine past data, candidate psychology and engagement analytics to predict the likelihood of a candidate dropping out.

    With AI, Pankaj Bansal, co-founder of onestop HR consultancy firm PeopleStrong, has seen dramatic improvement in productivity in his firm. Thanks to AI-based tools, last year his revenues went up 60% but the headcount just 5%. Earlier, each of his employees would take 20 minutes to sort and download relevant resumes for each position. Now they do it in two minutes. What is interesting this time is “recruitment trends in India are not very far off from what’s playing out in Silicon Valley. I have a feeling this time we are building startups that are originals,” he says.

    View from the Valley
    The US is seeing a surge in AI-based startups taking a stab at a range of issues that the job market is grappling with. In August, job platform Indeed — which claims to be the world’s No 1 jobs site — acquired a 12-employee startup called ‘Interviewed’, which helps companies with its AI-based candidate assessment tool. For example, its automated phone screening, machine learning and natural language-processing capabilities help construct a psychological profile of the candidate and judge cultural fitment. It claims to whittle down a database of 4,000 candidates within days to shortlist the top 2-3%.

    Another startup Entelo mines data and social media presence to predict which applicants are likely to switch jobs. HireVue’s AI-based platform analyses interviews on the basis of facial expressions, words being used, voice inflection and micro gestures to assess a candidate. For example, it draws inferences from candidates’ usage of active verbs such as ‘can’ and ‘will’ or negative words like ‘can’t’.

    Image article boday

    Image article boday

    Image article boday


    Talent Sonar’s algorithms help write job description that improve gender diversity. SkillSurvey, used by Adidas and Reebok, helps predict an individuals’ turnover and performance. There is Mya Systems, which has developed an AI recruiter — a chatbot — that can evaluate resumes, screen candidates and schedule meetings. Another startup Headstart uses an AIbased matching system to transform the recruitment process for graduates on campuses. Besides qualification and experience, it creates a ‘fingerprint’ for each candidate factoring in his interests, personality, skills and the like.

    The space has so much potential that Google has entered the fray with a programme called Cloud Jobs. Scanning millions of job openings to understand patterns and connections, it uses machine learning to understand both job content and intent of job-seekers, resulting in better job site engagement. Google also has an internal hiring tool called qDroid. Indentifying traits that its research shows are critical for specific positions, qDroid parses data to draft questions for interviewers.

    End of Job Portals?
    AI is shaking up the recruitment industry. But clouds have been gathering for a while now. Last year HR consultancy firm Randstad acquired online job portal Monster for $429 million, sending ripples across the recruitment industry.

    Founded in 1994, Monster heralded a new era of job-seeking and hiring just when internet usage was gaining both volume and depth globally. It spawned copycats including in India, like Naukri. Earlier this year, another job portal CareerBuilder got acquired by a group of investors led by private equity firm Apollo Global Management.

    More recently it laid off 120 employees, 4% of its staff. Sale of professional networking platform LinkedIn to Microsoft last year too made headlines. Multiple issues have bogged down job portals.

    The biggest is that they were born in the era that preceded the social network. Facebook was founded in 2004. A plethora of online social and professional networks have emerged, from Facebook to Twitter, GitHub (for software engineers) to Kaggle (for data scientists), which offer easy and accessible platforms for professionals to interact, network, collaborate and seek-and-offer jobs. A CareerBuilder report says that 70% of the employers today use social media to screen candidates before hiring, a sharp surge from 11% in 2006.

    Other changes too are afoot. Cheap handheld devices, rising internet access and falling data tariffs have democratised the digital world. Not to forget the structural shifts in the economy, which are making the employer-employee relationship complex and layered as millennials entered the workforce. Companies like Uber, Airbnb and WeWork are bringing in new pay-per-use models in both products and services. In sync, there has been a sharp rise in freelance workers who are seeking project-based work in the growing gigs economy. Projections show that 43% of the US workforce will be freelancers by 2020.

    Job portals have struggled to keep pace amid this growing complexity. In an era of interactivity, jobseekers often complained of job sites becoming black holes for their resumes. Many jobseekers complained of recruiter ghosting — no update or feedback after their interviews. This is where AI-based recruitment tools using big data and analytics are changing the game.

    Bansal of PeopleStrong says traditionally the recruitment industry had three broad verticals around which companies did business — ATS or the application tracking system (like Oracle Taleo), matchmaking and sourcing of talent. “In the era of Big Data, the three have converged,” he says. And companies from Indeed to Randstad to even PeopleStrong have been acquiring firms to prepare for the converged world even as new niche AI-based startups have mushroomed.

    Image article boday

    Image article boday

    Image article boday


    Tread with Caution
    India is still at an early stage of adoption with the IT sector, digital startups and multinational R&D centres leading the AI wave in recruitment. But expect rapid growth. And, like in other sectors, AI promises to be at the vanguard.

    There are concerns, the biggest being invasion of privacy. While most AI-based platforms assure that they only track data in the public domain, boundaries may be blurring. For instance, the political views of candidates could potentially bias an employer’s decision-making, especially in these politically-polarised times. “In India, we do not have robust privacy laws and things are still evolving. Hence, we have set our own gold standards on these issues: like, we will not scan data that is private or track political views of the candidates,” says Rishabh Kaul, cofounder, Belong.co. Increasingly, in a networked world, a candidate’s digital footprint will gain significance.

    So remember that vitriolic tweet or a picture of drunken revelry on Facebook will have deeper implications for those in the corporate world. Conversely, in the machine-dominated recruitment world, potentially-employable candidates, who are not very active online, may often fall under the radar and miss out on opportunities.

    “Our product isn’t a silver bullet. This (AI-based program) is something an employer will apply alongside a suite of other tools to make a hiring decision,” says Ben Mones, co-founder of US-based startup Fama, which helps companies spot candidates who may have a violent streak or are racists or misogynist. “It is not always a question of what is legal. The bigger question is if people will be comfortable with it.

    New technology in the wrong hands can also be used in a wrong way,” he adds. On the one hand, machine-based screening is expected to help remove human biases that often creep in the hiring process.

    But the verdict on this is fuzzy as bias mitigation tools could also worsen it. For example, because the AI algorithm factors in historical data it can easily replicate past human bias. A recent study by researchers at Carnegie Mellon University reveals that online ads for high-paying jobs are shown more often to men than women. “My hope is that you recognise the dangers of unchecked algorithms,” cautions Stroud of Randstad.

    Tesla’s Musk has warned that competition for AI superiority could trigger a World War, but those engaged in combat for talent may be harboring few such apprehensions.

    AI-based tools in recruitment will become mainstream in 5 Years: Rishabh Kaul, Belong.co
    Founded by Rishabh Kaul and Vijay Sharma in 2014, Belong.co is a predictive outbound hiring platform that uses AI to cater to recruitment needs of companies. It raised its Series A funding led by Matrix Partners in 2015 and Series B funding of $10 million led by Sequoia Capital early this year. The Bengaluru-based startup has over 100 employees and counts PayPal, Amazon, Reliance Jio and Cisco among its customers. Edited excerpts from an interview with Kaul:

    Image article boday
    On Belong.co
    With our product, we are helping companies use AI and machine learning to tackle their recruitment needs. Our platform offers predictive outbound hiring solutions (employers reaching out to candidates proactively instead of the other way around) that help companies discover, engage and hire the right talent. Talent that may not be looking out for job but may be just the right candidate for the companies.

    On how Belong.co uses AI
    The platform looks at a multitude of factors to bring to the surface the best-fit candidates for each company and role using our proprietary AI-backed algorithms. The key difference is that we go much beyond the standard keyword match approach. Each company sees a different set of candidates since we consider factors like the company’s hiring history, ideal candidates they would like to hire and match that with the available candidate pool, basis their public online presence across a range of networks like Twitter, GitHub, Stack Overflow, ResearchGate, Behance, to name a few. Beyond discovering the right candidates, it’s equally important to engage them in an effective and personalised way, and we have been able to automate that completely for our customers, saving 15-20 hours per week of effort for each recruiter.

    On how this works
    Once the platform has built a database of potential candidates matching with recruiters’ needs, the platform sends out automatic, mass emails to vet their interest. But what makes it special is that the emails are very personalised to each candidate’s profile. For example, the email could have a reference to the candidate’s favourite football club or the café he frequents or an upcoming marathon that he has been tweeting about. For recruiters, this helps make better first impressions on even passive candidates, a higher engagement level resulting in higher response rates. The platform is intelligent enough to even gauge (from his social media posts) when a candidate is more open to changing his job.

    On who are the customers
    AI-based tools are at an early stage of adoption in the recruitment world. I see them becoming mainstream in the next five years. Internet-based companies and startups like Flipkart and global MNCs — especially their R&D centres here — have been our first customers. Other sectors like manufacturing, retail, IT services companies too are promising customer segments for us. Many of these industries are getting disrupted and companies need to hire for positions that are newly created. Often, their internal team may not have capabilities to hire for these roles. While they may hire recruitment agencies to do the job, from the return-on-investment point of view, our platform is more cost-efficient. But most of our clients come to us because of superior candidate experience and engagement.

    On the flip side of AI-led hiring
    Sometimes, good potential candidates who are not very active online may miss out and find it harder to land jobs. Also, in India we do not really have robust privacy laws and things are still evolving. In that case, on principle we want to set our own gold standards on these issues. Like we will not scan data that is private or track political views of the candidates.

    We help companies identify social media red flags: Ben Mones, Fama
    California-based Fama Technologies is developing a platform that does social media background checks on potential job applicants and helps employers in their recruitment process. Fama’s AI-based technology trawls the web checking on Facebook, Twitter, Instagram, Google+ to pick up on the ‘red flags’ and ‘green flags’ of potential candidates based on their online behaviour. Set up in 2015, the startup has raised just under $3 million and landed the award of ‘most innovative use of data’ at the Wharton People Analytics Conference earlier this year. Edited excerpts from an interview:

    Image article boday
    On Fama Tech
    We screen potential employees by analysing their social media posts. An SHRM (Society for Human Resource Management) study reveals that 43% of companies use social media to screen job candidates. Fama started in January 2015 in Los Angeles with one goal in mind — to help businesses fill gaps in information on who they hire. What we do is not a thumbs up or thumbs down. Our technology helps employers red-flag candidates who could be misogynist or racist or violent and also spot those who may have a positive side to them like someone doing charity work. Employers have started using our software to fill the gap.

    On traction from employers
    The area is very new but growing rapidly. According to a survey by Careerbuilder.com, in 2006 just 5% of the recruiters were using social media screening. By 2016 it was up to 60% and in 2017 it was 70%. Many companies wanted to do it but didn’t have the money. Our thrust was to have a platform that made sense not just conceptually but also to integrate it well with the hiring process. We now have clients across the globe including western Europe, US and some in India. They include executive search firms and Fortune 500 companies.

    On the advantages of the Fama platform
    We help companies de-risk the kind of people they hire and avoid those with a background of bigotry or violence. We also make their hiring process time-efficient. Even a 10-15% speed-up in hiring is big. Since we started all our customers who signed up are with us.

    On the flip side
    This isn’t a silver bullet, an end-all-be-all solution. This is something employers will apply alongside a suite of other tools to make hiring decisions. Our product may not drive the decision but it offers an important thread for them to consider. Also, if a person has a limited social media profile then there are challenges with that. While we look at available data in the public domain and it is legal, there is still a question if people are comfortable with it. New technology in the wrong hands can be used in a wrong way.
    The Economic Times

    Stories you might be interested in