[LI] Towards AI for job seekers?

Article originally published on LinkedIn, 1 May 2017

Source: IADPP solves a random cooperative pathfinding problem, 18 June 2012, https://www.youtube.com/watch?v=Aqacl2VrgQM

On this International Labor Day, May 1, I’d like to take a moment to look at how intelligent automation or “artificial intelligence” (AI) might do more for the employment market, and especially for job seekers, and ultimately how it might transform it. Could we ultimately have intelligent agents – AI bots, in effect – working on behalf of recruiters and job hunters, which interact with each other on our behalf and help us all come up with better results for our efforts?

These days, most of the attention when discussing AI and jobs is on how the former might take over or at least dramatically change the latter. There is also some attention to how AI tools can help recruiters and even job boards. However there is next to nothing being done to put automated tools in the hands of job seekers who are basically still working with 1990s technology (manual input, eyes-on searching) plus slightly more recent social media like LinkedIn (theoretically to enhance making and maintaining person-to-person contacts) and job search services (which may provide more to process with the old technology).

A job application robot

The only documented effort to automate finding jobs and submitting employment applications that I’m aware of was undertaken by Robert Coombs and published this past March in Fast Company under the title “I Built A Bot To Apply To Thousands Of Jobs At Once–Here’s What I Learned.” Motivated in part to “turn the tables” on automated systems for processing applications, his effort points to a potential that has yet to be fully explored.

I found it interesting that Robert not only built and fine-tuned the bot he created – which by the way could actually tailor the content of cover letters, so this wasn’t just blasting applications – but he also experimented with it and documented the results. His test(s) on effectiveness of automated vs. personally written letters are very helpful to inform future development of automated systems for job seekers. I see four main conclusions:

  1. It is possible to design and use an automated system for job applications that can tailor the applications to the position and employer, and submit on a wide scale.
  2. The bot was not found to have a higher rate of success than manual applications. However this was one person at a particular level in a particular industry, so it would be interesting to try this with a sampling of people in different industries and with different levels of experience.
  3. Networking – important in job searches – was one area that the bot did not help with.
  4. Jobs that are not listed – apparently a high percentage never are announced – was another area the bot did not help with.

Even given its limitations, I think it is inherently useful to be able to automate the repetitive tasks involved in applying to positions announced online, and in the process to find and apply to more of those. This demonstration is also a step on a path towards a more equitable distribution of AI on behalf of those seeking jobs and and those hiring.

Robert’s conclusions about networking and unlisted jobs are critical. Networking is primarily a person-to-person activity, though with availability of tools (like social media) that in my opinion will certainly be further developed with AI. Could that function be united with the application bot in a more complex intelligent system?

As for unlisted jobs, one would hope that more intelligent automated processes will enable people to search outside both the job listings and their networks – and perhaps to analyze potential for openings based on other metrics, like investment or contract bids. Here again, if you had an AI, learning program integrating these functions with the above, one imagines that it could perhaps find crosscutting ways of finding positions and employers at the right moment and responding effectively.

AI & the self-driving resume

The main question I see is how increasingly-intelligent automation or intelligent agents might change the nature of job seeking up to and especially beyond the inevitable point where AI working for the applicant and AI working for the recruiter skip the online forms and formatted resumes and communicate directly with each other.

In a series of three blog posts under the title “AI & the self-driving resume,” I speculate about how this might develop, with background on what I see as anomalies in the way job hunting and hiring function today, and the roles of the resume and intermediaries (job boards and coaches):

What was started as one post a year ago became three in the process of expanding discussion. During that time, some elements had to be deleted or edited due to ongoing changes in use of technology.

I also outlined a rudimentary concept for automated job search and application as a first step. based on my (admittedly limited) understanding of currently available technologies. This “Mark 0.1” concept was obviated by Robert Coombs’ more sophisticated bot, which I had not known of when writing.

Paradigm shift in the job market?

Anomalies in the market and the rapid development of applications of AI lead me to think of a paradigm shift that will affect not only how employment is handled on all sides, but also networking, how careers develop, and perhaps other aspects of the economy.

That shift, such as I’m imagining it, would become clear when bots for automatically applying for jobs on the one hand, and automated application processing systems (i.e., ATS & descendants) on the other are replaced with some kind(s) of intelligent agents that leverage selected AI capabilities and are capable of interacting directly with each other.

That leaves us with the other, larger issue of the impact of AI on human employment as its increasing use transforms the economy. But the AI tools which can and in my opinion should be developed for individual use in the job market might conceivably help people navigate these changes.


Other blogs > LinkedIn > LinkedIn articles & posts, 2017 (Jan–Jun)

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.