The term AI has become somewhat of a buzzword in the business world. Just Google the terms “artificial intelligence” or “machine learning” and you’ll get hundreds of pages of results, many of which make lofty claims about how this technology is the magical solution to just about any problem an organization may have.
Unfortunately, this simply isn’t the case – at least not for the time being. That’s not to say AI is useless. To the contrary, AI does, indeed, work well in some instances and industries.
Take, for example, use cases like AI in terms of security products. With cybercrime at an all-time high, and online criminals becoming more sophisticated by the day, there is a plethora of data that can be used to train intelligent automation to counteract hacking attempts and help organizations keep their sensitive information secure.
Likewise, consumer-facing businesses gather a ton of transactional information on a daily basis. This information can in turn be used to enhance the customer experience. For instance, prior behavior can be tracked and used to automatically match a particular customer with promos that are personalized to his or her preferences.
So, what about recruiting? It would seem logical and straightforward that a predictive tool could be used to better match candidates with open positions. Unfortunately, the science simply isn’t quite there yet.
The key differentiator here is data.
That is, the true benefits of AI can only be realized when there is sufficient base data to build dynamic, predictive models. And not just any data will do, either. In addition to being plentiful and readily available, in order for AI to deliver value, the information it consumes must be of good quality and relevant to the situation at hand. Otherwise, it’s simply garbage in, garbage out.
Making Some Headway
What’s the old saying? You’ve got to start somewhere. Well, the good news is, there has been a decent amount of headway made in terms of leveraging artificial intelligence as a tool for HR.
For instance, predictive analytics can be useful in identifying which individuals are most likely to change jobs. This can be accomplished because, while there are many variables involved, companies are able to track at least one of the major parameters, which is historical employee behavior. This provides enough quality and relevant data for AI to use.
Another example of where intelligent automation can add value to the recruitment process is through the use of chatbots. Essentially, virtual agents can be leveraged to engage with applicants, going through qualifying criteria and even answering commonly asked questions that candidates may have. This enables recruiters to effectively pre-screen a large pool of applicants without having to do so manually.
A Long Way to Go
Where AI for recruiting currently falls short, however, is in the in-depth process of actually hiring – the “nitty gritty,” if you will. And, sadly, there are many companies out there making big promises that they cannot realistically deliver on. For instance, one recruiting platform claims to help clients “Find the best candidates 10X faster with AI sourcing.” In reality, talent intelligence hasn’t yet reached that capability.
Think about it. Recruiters routinely break down job requirements into mandatory and preferred skills. The problem is, the chances of a person having all of those mandatory requirements actually listed on their resume or business social profile is highly unlikely. In fact, the likelihood of a job seeker including every single skill, product or technology they have experience with is slim to none.
As a result of this, the data being fed to AI is not of sufficient quality because it’s incomplete. And because this core data is flawed, there isn’t enough information for the predictive models to generate valuable insight. Remember – without good base data, AI is essentially useless.
To drive this point home, we performed our own experimental search on LinkedIn. We entered just 5 mandatory requirements, which returned multiple pages of results. Yet, we didn’t locate a single candidate who had all of our mandatory requirements until we reached page 6 of the results. Not only was this a tremendous waste of precious time that busy recruiters simply don’t have, but because of flawed or missing information, many candidates who were qualified may not have even shown up in the search results at all.
Unfortunately, the companies promising to automate and improve the sourcing process through artificial intelligence are using the same incomplete data that our LinkedIn experiment produced. They are essentially building predictive models on missing data, which means their results are equally flawed.
“If it seems too good to be true, it probably is”
A word of caution: don’t believe everything you see or hear. If a company tries to woo you with claims of AI being able to source candidates 10 times faster or generate perfect candidate matches in just seconds, don’t be afraid to ask questions. Specifically, we recommend inquiring on the volume, source and relevancy of the data they’re using. This will enable you to better determine whether the platform will truly help you evaluate candidates’ qualifications, interest, or availability.
Also, beware of buzzwords like AI, machine learning and big data. Anyone can use these terms, but not everyone who does is doing so ethically. Before you spend money on another tool, assess the value carefully. And remember, if it seems too good to be true, it probably is.