Why the recruitment industry needs to stop fawning over AI and embrace the RPA (Robotic Process Automation) revolution
One of our best customers – a staunch and early-day advocate for Talkpush – recently shared with me that he had told some overseas colleagues that “Talkpush does not do AI”. Holding back my outrage, I asked him as calmly as I could: "What kind of AI do you have in mind?" To which he replied, "There are competitors in the market that use AI to automate the selection process”
“You do realize that you are using the same technology when Talkpush goes through thousands of your candidates every week?” I interjected in self defense.
As it turned out, my customer had been using most of our AI capabilities, but didn’t even know it. Isn’t that how good technology is supposed to work? Sure, but I realized right then and there that it was time for me to bring some clarity on what exactly is the current state of the art of AI in the field of recruitment and why it is that Talkpush has never cared to use "AI" in most of our marketing and communications.
The rise of the AI junky
But first, how did we get there? How did the recruitment industry go from a handful of AI evangelists only 5 years ago to where it is today, where seemingly every department in every large company is begging for more AI to be injected into their operations? I think there are two macro factors at play. First, AI became much bigger in popular culture, with science fiction grabbing a bigger mindshare in movies (Her, Prometheus…) and television (Black Mirror, WestWorld). The narrative of the bots taking over humans is a thousand times overplayed, but still continues to draw strong interest, likely because it rouses our primal survival instincts.
Secondly, AI captured the minds of venture capitalists, who are always looking for the next winner-take-all category killer like Google or Amazon. To access this venture capital, tech entrepreneurs rushed to highlight how they were building the bestest, most deep-tech AI the world had ever seen. Robo-this and robo-that. Personalized digital assistants with sentiment analysis. Big data, machine-learning, conversational juke boxes with bells and whistles that play Beethoven’s fifth symphony. Anything to capture the minds of the AI junkies and the VC dollars.
Automation, not intelligence, is the true innovation engine
These lyrical promises for AI may delight some of the leaders in HR or other functions, but when the time comes to justify the budget, the business cases presented end up being much more prosaic and down to earth. Indeed, it usually turns out that 99% of the cost savings generated by this technology is in automating repetitive human tasks — not in making advanced, unbiased decisions.
Just look at some of the hottest recruitment AI tech, and the REAL reason why companies buy them.
- Video interviewing: Companies do not choose video interview platforms like Hirevue because it can read minds and emotions. They buy Hirevue to save time through asynchronous interviews. The vast majority of the users of the platform do not let Hirevue make the selection, but rather dedicate a few hours every month to watching hundreds of video interviews.
- Chatbots: Companies do not work with chatbot vendors like Talkpush because they can replace recruiters when talking to candidates. They find most of the ROI happens in automating simple tasks, like the sorting of incoming traffic or scheduling interviews. Yes, every vendor is also building NLP capabilities that can handle some general inquiries about the company, or even regarding the specific job, but the bulk of the ROI is generated in automating the pre-screening and scheduling of candidates. That's very cool automation, but it isn’t called AI.
- Voice analytics: Companies are starting to buy voice analytics services like Chorus.ai to measure the performance of their sales teams. Where is most of the value generated? Simply in distributing audio recordings for training purposes.
- AI profile matching algorithm: That's the new way to describe a resume scanner or a search engine. Although helping recruiters identify profiles with the right keywords or skill set is pretty cool, it doesn't quite feel like AI to me.
- Big data champions: The darling of the AI world, IBM Watson also offers to help the recruitment industry. But they haven't offered to help with candidate selection. Instead, in a shockingly humble user case, they concede that they can tell us ahead of time which jobs will take longer to fill. I don't know about you, but I know pretty damn well which jobs will be hard to fill. Anyway, that sounds like basic regression analysis with a smattering of AI. We are still very far from the machines taking over.
As you may have noticed, none of these changes are threatening to change recruitment as you know it. These are merely small automations of tasks that would otherwise take away some of the time recruiters spend on actual high-value functions (such as actually meeting qualified candidates and assessing their culture fit). Recruiters have been sold AI, but mostly what they got is a little bit of Automation around simple tasks.
Is this False advertisement? Should you ask for an AI refund? (or did you forget to read the fine print? )
Tech vendors promoting their technology as "AI" is as prevalent as food conglomerates calling their cereals "natural": it costs them nothing to put that label on the box and it might get you a few extra customers. Consequently, almost every tech vendor in the market is "strong on AI" . Are they lying to you? That entirely depends on your definition of AI. As it turns out, defining AI is almost as tricky as deciding what "natural" stands for.
Indeed, every software vendor is using various building blocks that could be considered AI. In the case of my company (Talkpush), we invest time and resources in building a library for our NLP (natural language processing), which helps us direct candidate inquiries better. We also use a voice to text transcription engine that can be labelled as an AI technology. And I'm sure our backend does hundreds of activities per second that rely on some form of intelligent load balancing. So, while I would have no problem making the argument that we are an "AI recruitment provider", it just feels like an empty word, just like saying cereal is natural.
Some will argue that real AI only pertains to Machine Learning being applied to vast data sets to automate complex decisions. Some purists will say AI should only be used to talk about "deep learning". If that's the real definition of AI, then recruitment is simply not at a stage of development where it’s really needed. My advice to the TA tech space: let the Amazons of the world play around with AI in recruitment at their own expense. When ML technology has matured and when politics have moved on from ostracizing AI (per the latest news: Illinois blocks AI in video interviewing) and regulating the way employers choose their employees, that will be the time to invest in software that makes decisions for you (if one could really aspire to such a thing).
Why recruiters should jump on the Recruitment Robotic Process Automation (aka RPA) bandwagon
Usual consumers of tech news know that RPA (Robot Process Automation) is the hottest field at the moment, surpassing AI over the past 12 months. Just look at the fantastic growth in headcount from RPA leaders UiPath and Automation Anywhere over the last two years:
So while RPA is taking the world by storm, recruiters are still mostly foreign to the concept. The value proposition of RPA is stated upfront: automating repetitive tasks. This gives RPA vendors a clear focus on what objectives they need to achieve with each project. It dispenses them from story-telling, constructing a narrative with a scope beyond the identified tasks. RPA came of age thanks to the improvements achieved in some of these AI building blocks — notably, the advancement of visual recognition technology that can "see" what is going on in your computer screen. But, RPA doesn’t pretend to be making decisions for us, it simply tries to automate things we don't want to do.
AI vs RPA : which is which?
Areas you can apply RPA within recruitment
It is high time the talent acquisition world embraces RPA's pragmatic approach to managing innovation, considering that the biggest cost in recruitment is the time spent handling menial tasks. To use RPA in this industry, look at the tasks that consume the most time and figure out a way to automate them, you’ll see they are almost universally the same.
- Capturing leads from various sourcing channels
- Asking candidates questions to find out what their core competence is
- Sharing key information with candidates about employer value proposition
- Scheduling interviews
- Taking interview notes
- Calling and writing to candidates to get updates
- Sharing interview notes and commentary with hiring managers
- Running assessments
- Background checks
- Preparing new hires for Day 1 on the job and onboarding
Automate the list above, and you will have reduced your cost-per-hire by more than 80%. This is a much better use of our time than fantasizing about fully autonomous hiring systems that may never materialize.
Above is an example of how Stanley, the Talkpush standard chatbot uses RPA to automate scheduling, freeing up recruiters to pursue high-value tasks — instead of sending a string of e-mails trying to nail down a convenient time.
The best part about RPA: you can roll it out step by step, process by process, in order to minimize the human impact as you redirect your workforce to more high-value tasks (like meeting qualified candidates in person, or fine tuning the employer value proposition).
To facilitate this type of endeavor, companies will need to work with tech vendors who are strong on documentation and invested in building great APIs. Those are the connectors that help the automation of a workflow. Getting systems to talk to each other is the best way to free up human operators from having to do it.
The revolution will be robotized
Not so long ago, technology was seen as the great liberator of mankind, instead of a great existential threat. In our excitement about the new possibilities unleashed by developments in NLP, Optical character recognition (OCR), image identification, and voice recognition, we got ahead of ourselves, and forgot that technology is meant to alleviate our pain points one process at a time.
Before assisting humans with decision making, which is arguably the most fun/interesting part of the job of a recruiter, technology should first help us get rid of the painful, boring parts of our job. Shifting the narrative from AI to RPA will help us gain focus on what we can accomplish next month, or maybe next year, to improve the individual lives of recruiters and candidates.
Want to learn more about Robotic Process Automation? I've listed some introductory resources below:
Talkpush is the new way employers and candidates connect all over the world. We believe hiring is about starting conversations; and are on a mission to level up the recruitment experience. Want to know more? Schedule a demo 👉👉👉 here.
If you like what you read, share using #Letthetalenttalk or #levelup and don’t be shy with your thumbs up!