We’re all hearing about Artificial Intelligence and how it will impact business, but what really is it and what is everybody really talking about when it comes to AI?
And of course, how will AI impact us as recruiters?
Here at Undercover Recruiter, we are conducting an ongoing investigation into AI so we can understand it all ourselves and help our subscribers benefit from our findings. From that we can see that the definition of AI is different depending on who you talk to and what their individual frame of reference is.
Is AI the big scary thing that will take over everyone’s jobs, or something a bit more niche that will help us do what we need to do in a more efficient, scalable, and creative way?
In this article we will talk about what AI seems to be, and how it is all around us already and we haven’t really noticed, and why it isn’t necessarily as ground-shaking as we think it is.
The AI effect
The thing about AI is that we don’t really notice it. That’s because it is generally embedded into something else and just becomes part of the “technology”.
Big names in AI like IBM’s Watson or Google’s DeepMind make the headlines when they take on challenges that previously it was thought only humans could do – but AI is much more pervasive than is at first apparent.
— john pearce (@techpearce3) May 4, 2017
Take a moment to think about how you interact with technology every day. How much of it has an AI component helping it to help you?
AI is in your phone, in your favourite search engine, your favourite e-commerce site, and your games console. It controls traffic lights , is in building control systems , and will soon be in your car if it isn’t already.
So when AI is working and functioning and helping us out in an everyday way, it becomes part of the background and no longer feels like AI. It seems that “AI is whatever hasn’t been done yet“, and “As soon as AI successfully solves a problem, the problem is no longer a part of AI”.
So given that AI is something that becomes part of the technological background as it is implemented, what is all the fuss about?
Let’s first of all categorise AI so that we can break it down into manageable chunks and move away from the big hairy, scary, overarching concept of doom-laden AI that totally disappears when it becomes useful.
Then we can figure out how it can help recruiters in the future.
The categories of AI
AI is, as you know, “Artificial Intelligence” and labels the concept of creating a machine that exhibits intelligence.
But did you know that AI as a concept in the form of “mechanical reasoning” has been something philosophers and scientists have been toying with since ancient times? You can check the extensive definitions and references in the Wikipedia article on the subject.
But let’s look at the categories of AI and after we’ve explored those a bit, we’ll look at the types of functions AI provides.
AI as a broad definition can be broken down into three categories:
- Artificial Super Intelligence, or “ASI”
- Artificial General Intelligence, or “AGI”
- Artificial Narrow Intelligence, or “ANI”.
1. Artificial Super Intelligence
ASI is the AI of science fiction – the artificial intelligence that surpasses human intelligence in every respect.
Most of the doom-laden hype about AI concerns itself with this kind of AI and how it will take over the world and rule all humanity. In fact, what purpose would we humans serve in the presence of such mighty intelligence?
We’re certainly not there yet and this kind of AI does reside purely in speculative fiction.
So for our lowly recruiting purposes, we can ignore this one.
2. Artificial General Intelligence
Also known as “Strong AI”, AGI is ASI-lite and is characterised as being as smart as a human in all respects, across the board.
This kind of AI can reason, plan, solve problems, think, abstractly, comprehend complex ideas and learn quickly from experience and demonstrates what can be seen as an equivalent intelligence to human beings.
The famed Turing Test was proposed in 1950 to test, though a text-based interaction, if an AI is indistinguishable from a human being.
Although strides are being taken in this direction, we are also still a long way from this and we will not have AGI to help us out in the near future.
Now on to something more useful …
3. Artificial Narrow Intelligence
ANI is also known as “Weak AI”.
ANI covers the Ads we see today in every AI application. They are “narrow” in the sense that they each specialise in performing a single function, for example stock trading, responding to voice requests on a smart phone or device, or performing image recognition services.
These are the AI solutions used in all around us today, and we will be seeing more and more ANI implementations as we move through the next couple of years.
For example, here at Undercover Recruiter we have an Amazon Alexa to play our music in the office via voice control. That is ANI in action.
For recruiters ANI is the most useful category of AI for our business.
Let’s now explore types of ANI functionality and how they might be applied to recruitment.
Now that we know that it is ANI we are talking about, and not AI in general, we can take a look at the types of functions that ANI can perform.
Broadly speaking, these can be divided into two categories of services:
- Cognitive services, where the ANI understands something and based on that, does something. IBM Watson is an example of a cognitive ANI.
- Decision-making services, where the ANI compares things, and depending on its confidence level in the data presented to it, can make a decision. Rainbird.ai is an example of a decision-making ANI.
Combining cognitive and decision-making ANIs can give us a way to understand something, and based on that understanding, make a confident decision based on that understanding.
Are you starting to see how this may apply to recruiting?
Using ANI in recruitment
If we think about the work that is done in the recruitment life-cycle, we can start to consider how ANI can be used to fulfil some of the tasks.
Looking broadly at a recruitment business, the activities that generate revenue and provide value to clients can be categorised as follows:
“Sell”, “Source” and “Match” are hopefully obvious tasks, with “Manage” including managing the entire recruitment process, and relationships with clients and candidates.
Here at Undercover Recruiter, we can foresee that all of the tasks which are administration-based, repetitive, cyclical, and generally feel like a chore to recruiters can be automated.
And all of the relationship-based tasks which require interaction and negotiation with people, at this stage at least, cannot be automated.
Let’s look at a more detailed list of tasks and see which ones can be automated today. There are undoubtedly more tasks that should be on the list, but let’s run with the following:
|Sell||Identify appropriate clients||Automatable|
|Sell||Sell recruitment services to the identified clients||Not Automatable|
|Sell||Negotiate rates and business terms with new clients||Not Automatable|
|Sell||Manage client relationships and up-sell work||Not Automatable|
|Manage||Understand a clients’ business, culture, and employer brand||Not Automatable|
|Manage||Understand role requirements and assist clients in creating appropriate role descriptions which will attract the right candidates||Not Automatable|
|Source||Source candidates as appropriate for each role||Automatable|
|Match||Identify suitable candidates within the pool of sourced candidates||Automatable|
|Source||Schedule calls with the candidates that you have identified||Automatable|
|Manage||Sell the company and the role to the identified candidates||Not Automatable|
|Manage||Understand the requirements and expectations of the candidates||Not Automatable|
|Manage||Manage the candidates’ expectations and keep them appraised of the current status of the role||Not Automatable|
|Manage||Send a few test candidates to the client to solicit feedback, repeating the previous steps as necessary to adjust the scope of the work||Not Automatable|
|Match||Create a shortlist of candidates||Automatable|
|Manage||Meet the shortlist of candidates face-to-face to make sure they fit with the expectations of the role and to identify any red flags missed during the process so far||Not Automatable|
|Manage||Present the shortlist of candidates to the client and understand their feedback, adjusting and repeating the preceding steps until a suitable shortlist has been created||Not Automatable|
|Manage||Feed back as appropriate to the candidates to keep them appraised of the status of the role once more||Not Automatable|
|Manage||Schedule and manage interviews between the selected candidates and the client||Automatable|
|Manage||Manage the feedback from the client from the interviews to the candidates, and repeat any of the previous steps as necessary||Not Automatable|
|Manage||Schedule further interviews with the shortlisted candidates as required by the client||Automatable|
|Manage||Manage the final candidate selection with both client and candidate||Not Automatable|
|Manage||Assist in negotiation between client and final candidates||Not Automatable|
|Manage||Close the job and bill the client||Automatable|
If we then take the types of work (Sell, Source, Match, Manage) and group them according to whether they can currently be automated or not, we get the following:
|Work Type||Automatable||Not Automatable||Total # Tasks||Automation Ratio|
It’s interesting to see that 100% of the Source and Match tasks are automatable whereas only 20% and 25% respectively of Manage and Sell are automatable. This is where the human touch lies which is so necessary in recruiting.
Overall, 35% of tasks can be automated, so if a recruitment organisation moves towards automation, 35% more of every recruiter’s day can be spent on those essential human-oriented tasks with the repeatable, administration and process-oriented tasks being automated.
Sounds like freedom, eh?