Featured Image

I’ll age myself with this question, but I’m going to ask anyway: Do you remember the big piles of ads that used to come in the Sunday paper? There was an actual stack — over 50 for many major metropolitans like Boston, my hometown. Everyone advertised back then — from the big box stores to mom and pop shops.

For a long time, those ads were manually sorted. There were humans who had to sort, stack, and stuff every ad to make sure they ended up in the right place. At scale, this was a huge mess. And there were a lot of advertiser refunds.

But within this chaos, I saw opportunity. I could fix it.

 

Replacing Humans to Let Humans Do the Stuff that Matters

One of my first startups helped newspapers build a system that sorted and sent ads to the correct locations. It was basic automation — using systems, rules, and networks to remove human error, accurately categorize ads, and ensure they ended up in the right place. I learned a lot from those networks about where automation made sense, where it definitely didn’t, and the nuances of local markets.

I tell this story not to let everyone know that my formative years were spent reading newspapers (and building software for them). I tell it because I learned an interesting lesson about the reality of automation and cost-savings.

The presumption seems to be that automation kills jobs — but that’s not always the goal, and it’s often not the most prominent cost savings companies see when they automate. In the example above, the majority of the cost savings came from the newspapers not having to give credits or cash refunds to advertisers because their ads didn’t go in the right newspapers.

Simply put: It wasn’t just human capital being saved. It was production capital.

 

Toeing the Line Between Hype and Real, Tangible Value

Now, I know there’s a big gap — technologically and practically — between automating newspaper ad distribution and leveraging AI to let a machine tell us who we should hire. The latter is more consequential if we get it wrong.

But as we try to build truly global recruitment marketing software, I think it’s important to consider the limitations and actual opportunities of the technology we’re building (and you’re buying).

A few questions jump to mind:

  1. Do we really want it to replace humans — or do we want it to augment them?
  2. Do we trust technology to make really critical, nuanced hiring decisions for us? And what’s the risk associated with that?
  3. Do we have any visibility into how those decisions are being made?

There’s a lesson in those questions: Often, what we think we want is different than what we actually need. And this lesson is one we have to keep in mind as we look to automate processes across the recruiting lifecycle.

Let’s face it. Tech-only experiences for recruiting — even if they’re driven by sophisticated AI — will never work. I’ve had the good fortune of working closely with some of the best companies in the world. And I’ve seen some of the smartest talent acquisition organizations (with AI and machine learning teams) try to automate their recruiting processes end-to-end.

Actually, they didn’t try. They actually succeeded. But the fruits of that labor were surprising, to say the least.

One company did such an excellent job of automating their process that it eliminated the need for sourcers and recruiters across most stages of the recruiting lifecycle. If that sounds like a panacea and a great way to save money, consider the result: More than 50% of this company’s employment offers were declined. Why? The feedback they got from candidates was clear: The whole thing felt too automated and impersonal. There was no connection to the business.

Very simply: The machines hadn’t given candidates a compelling reason to join the company. AI failed at selling the best people on the soft stuff — the stuff great recruiters and hiring managers can sell so well. So, these top-tier candidates went to competitors who had humanized their story.

 

Striking a Balance Between Artificial and Human Intelligence

This story and many like it lead me to believe we are on the cusp of a shift in how corporate TA teams evaluate technology — AI specifically.

There’s a buzz and energy around AI in HR and TA right now as companies dive into their 2019 purchasing and planning. As a technologist, I love seeing the excitement to embrace innovation. But I have a feeling that if practitioners and TA leaders don’t approach this the right way, it’s going to end in disappointment.

It has nothing to do with the quality of the tech — there are some truly innovative, incredible startups in our space doing amazing things. The issue, however, is our market’s expectations for what AI will do. It’s viewed as an all-encompassing solution — a switch we can flip to solve all our problems.

But it’s not.

It’s technology that’s dependent on well-designed processes, clean (and connected) data sets, and good human decision making. AI isn’t just inherently smart — at least not yet. And nothing “just works.” 

Let me give you an example. With Emerson, SmashFly’s intelligent recruiting assistant, we make a very intentional investment of time upfront to ensure the algorithms are properly trained and working from quality data. And even then, we give our customers the tools to supervise conversations at scale, optimize the experience over time, and ultimately make the whole system smarter. This takes work, but it’s worth it.

At the same time, we make sure Emerson gives our customers the ability to seamlessly move between machine and human. If Emerson can’t answer every question or is stumped by a candidate’s response, a recruiter or a sourcer can jump in and take over. This is rarely necessary, but it’s an excellent — and needed — safeguard to truly deliver an outstanding candidate experience.

Again, Emerson replaces the human only where the human doesn’t need to exist — answering basic questions about benefits and job openings; screening for basic or essential skills; syncing calendars to schedule pre-screen interviews. Emerson strips away the things humans shouldn’t have to worry about and creates more opportunities for recruiters and sourcers to add a human touch where it matters most.

 

A Realist’s Take on AI in 2019: Focus on What Actually Works

So, where does that leave us? Do we need to collectively redefine AI or scrap the term altogether? Or abandon the hope that it’ll alleviate the burden on talent acquisition today?

Not exactly. It’s just time to level-set. To think about what we’re really trying to solve — and why. To be realistic about the state of AI — where it is today and where it’s going. To ensure that what we build and buy is carefully crafted to solve a real problem in recruiting. To deeply consider how the technologies we use work together, so we don’t unwittingly create more work for our teams.

I think we can do all that by grounding ourselves with two simple questions:

  1. Are we buying software to eliminate humans or to free humans up to do things machines can’t — like building real relationships with the people who make our companies better?
  2. Does it actually work? And by work, I mean: Will it help you solve a real business problem and create value for your team? That might seem obvious, but it’s easy to forget when we become entranced in the potential of AI. I’m guilty of it, too. When you’re shopping for technology, don’t forget to ask vendors to let you see it in the wild — ask for real customer examples with real working product that solves a real TA problem.

My good friend Aaron Matos, founder and CEO of Paradox.ai, often uses a quote that I’ll borrow here. It comes from John McCarthy, who helped coin the term “Artificial Intelligence” and it goes something like this:

“As soon as the technology works, no one calls it AI anymore.”

And that starts with our own expectations of what “AI” needs to deliver. If we properly assess the potential of AI today, then it becomes much easier to see real value from it. We shouldn’t go into the market shopping for a miracle cure or a switch you can flip to suddenly fill your funnel with qualified candidates. We should go in knowing the capability of the tech and what we really need it to solve.

So, start with your goals and where your processes might be broken. Think about the problem you want to solve. Think about your team and your candidates — where the experience could be made better for both. And then think about out how technology — AI or otherwise — can make it a reality.

If we approach it that way, we’ll all feel a lot better about some of the amazing tech that’s coming into our industry.

Leave a Reply

X