The Job Candidate "Imitation Game"
Will computers eventually think like people do?
At the dawn of the information age, various far-sighted luminaries such as John Von Neumann, Alan Turing, and Vannevar Bush pondered whether computers might one day achieve human intelligence. Their discourse raised the question: "How will we recognize when a computer can think on its own?".
In 1950, Turing published a paper entitled "Computing machinery and intelligence" which defined what has become known as the Turing Test as a means to determine if a machine is "intelligent." This test, recently celebrated in the movie "The Imitation Game," states that a computer could be said to "think" if a human interrogator could not tell it apart, through conversation, from a human being.
Turing's proposed test sparked an ongoing debate about its validity. If you're interested, Walter Isaacson, in his 2014 book, "The Innovators," does a terrific job of highlighting some of the more interesting arguments and their rebuttals. For instance, he describes how, in 1980 John Searle suggested that if the computer is pre-programmed to respond in a certain way to every possible input, statement, or query from a person, then it could pass the Turing Test even if it never did anything more than consult a lookup table to obtain its response. A possible rebuttal is that, if the responses could convince a human it was real, then the "thinking" was embedded in that lookup table, which was part of the machine, and the Turing Test holds.
A new test for computer intelligence
And the debate goes on. In truth, nobody knows how to tell if a computer is really thinking. Since the philosophers have not succeeded, perhaps they need some help from the human capital sector. So, what if we slightly modified the Turing Test to:
A computer can be said to be intelligent if it can get itself hired by a company before they realize it's a computer. |
Sounds crazy, right? I mean, what recruiter would be fooled by a computer posing as a candidate?
Not as far-fetched as you may think
Well. OK. We are not there yet. But can we measure how close we are to computers demonstrating human intelligence by how much of the process they could get through before being ‘found out?' If that's the case, then I think we're getting pretty close.
For instance, let's consider the stages of the hiring process and ask yourself if a computer can pass each step:
- Locate job openings - Yes - Computers have been doing this for over a decade, scraping job postings off websites, aggregating them, classifying them, etc.
- Complete the application - Yes - This means filling out an online form. Can a computer recognize the standard form fields used in online job applications and fill in the appropriate answers based on the job description? Of course. The only barrier is the ubiquitous ‘Captcha' that you sometimes have to fill in. But it they can break the German Enigma code and understand a 5 year old boy's voice commands, they can be enabled to do that.
- Complete an online assessment - Mostly Yes - This is a partial. For simple right/wrong questions, the computer should get every question correct. For knowledge questions, a computer has the entire Internet to scan to find the answer. For some skills, like a typing test, it would have to intentionally slow itself down to avoid detection. However, for essay questions, especially those with plagiarism detection embedded, the computer may struggle to produce original, concise prose that is on topic. Also, electronic work simulations that use video and animation in branching sequences may present some challenges.
- Pass the interview - Mostly Not Yet - No way, right? Well. Let's consider that a growing number of companies are using video interviewing rather than bringing in people for in-person interviews. Recent advances in computer animation approach life-like imagery so closely that many of us can be fooled. How long will it be before our phone can have a conversation with us, or a face-time chat? Similarly, robotics continues to advance. After all, we already have the Robot Soccer World Cup - the RoboCup. The goal of the project is that: "By the middle of the 21st century, a team of fully autonomous humanoid robot soccer players shall win a soccer game, complying with the official rules of FIFA, against the winner of the most recent World Cup."
- Pass the background check - Not Yet - I suppose when computers are able to obtain social security numbers in the US, things will change. But, since they don't have birth certificates, it may be difficult for them.
So, do computers pass the test for hiring intelligence? No, not yet. But, if you're like me, you were surprised to see how far they can go within a hiring process. What if you use only video interviewing and you don't do background checks? When will a computer show up for their onboarding day? Probably about the same time as when robots beat the reigning world cup champions.
Wrapping it up
If Rene Descartes (1596-1650) was born 400 years later, do you think he he might have said something like:
"I think I can get a job, therefore I am." |
If a computer can get a job, can we deem it to be intelligent? I would say yes. Does this mean recruiters should worry they are working with a computer instead of living candidate? Probably not, at least for a year or two. But will the day come when recruiters do need to be vigilant? Absolutely. One day we will hear a news story about a company who hired a computer and put it on the payroll before realizing what happened. Hopefully that won't be at my company.
As computers begin to more closely approximate humans, the selection process will need to weed them (the non-humans) out more effectively. Captchas will not be enough. Perhaps there will be new psychometric tests that can be developed to detect the way a computer thinks. Or perhaps a simple non-invasive blood test - administered by your mobile phone, will do the trick.