In Part 1 of this blog post series, I provided an overview of recruitment evolution since the dawn of civilization. This survey was intended to provide perspective on the dramatic technological revolution enveloping us today that dramatically alters every aspect of our lives. Our minds adapt so quickly that we are seldom conscious of how data and automation facilitate what any generation before us would have considered magic (or, perhaps, a sliver of the most recent generations might have considered science fiction).
The Historical Pace of Progress
To refresh the broader perspective, consider that for thousands of years — from the earliest-known recruiters in ancient Egypt until the emergence of printed newspapers in 16th-century Europe — work as a recruiter was near-entirely unchanged. If you’d lived at any point in that millennia, you might not have observed a single innovation; whatever tools and techniques were available when you croaked might have been the same tools and techniques as when you were born.
By the 20th century, the pace of technological change had picked up considerably. A recruiter in the United States who began by placing veterans returning from WWII could, over a multi-decade career, gradually shift from a focus on in-person canvassing and newspaper ads to telephone calls. If they enjoyed an especially lengthy career, they might also witness the emergence of emails, job boards, and digital ads.
The Contemporary Pace of Progress
In the 21st century, it is a given that recruiters’ tools, technologies, and ways of working will change dramatically over the course of a single career — likely several times over. Some of the major factors underlying these changes are quantifiable. As key examples, consider that in the two decades since 2000:
- The cost of data storage has fallen by 1000x, enabling the affordable storage of staggeringly vast datasets for training machine learning (ML) algorithms.
- The cost per transistor on a computer chip has fallen by 100x, enabling data scientists to devise increasingly intricate ML models. These algorithms can be fed vast datasets and produce staggeringly nuanced outputs such as a paragraph of text on a given topic indistinguishable from one composed by a human.
- Finally, the cost of transferring data over the Internet has fallen 500x, facilitating global, real-time, and typically free access to the latest, most sophisticated ML approaches.
In addition to the above factors, exponentially more abundant sensors (e.g., cameras, microphones, heart-rate monitors) collecting data and the 5G “internet of things” currently rolling out will accelerate this trend further by increasing sensor mobility and interconnectivity. We are also experiencing unprecedented (yet still growing) investment in data-modeling innovations from public and private sources alike.
These technological trends’ confluence has enabled ML subfields like deep learning to usher in the dawn of an artificial intelligence (A.I.) revolution. While the film industry and news media typically portray A.I. as being on the cusp of replicating human intelligence, the reality on the ground is that even the cleverest machines today have extremely narrow specializations, and it may be decades (if ever) before a machine has a learning capacity beyond an infant’s.
Fast Dumb Machines Augmenting Slow Intelligent Humans
Popular misconceptions aside, we are nevertheless demonstrably in the early stages of an A.I. revolution that is rapidly transforming all aspects of life and work. Returning to our recruitment-industry focus, the transformation is exemplified by GQR’s ML algorithms, which eliminate historical hiring biases while augmenting recruiters’ capacity, enabling them to operate over a hitherto unimaginable scale.
Each month, hundreds of thousands of new job postings from across our core markets around the world flow into GQR’s database. Our ML models evaluate the fit of any given pre-screened candidate in our database for each of those jobs. Since there are a million of these vetted, high-quality candidates, that corresponds to trillions of evaluations per month. Trillion-scale volume is far beyond the grasp of a single recruiter’s mind or even all the recruiters’ minds at a single firm, yet our algorithms have suddenly made that scale accessible. By automatically presenting the best-fitting candidates for the available job openings to a given recruiter in an intuitive, click-and-point newsfeed-like interface, the most promising handful of recruitment opportunities from amongst the trillions of possibilities can be considered individually.
Augmenting humans’ broad intelligence with the scale of very narrowly intelligent machines like this has tangible real-world benefits to society. For example, it can cut down the time it takes to place traveling nurses in regional healthcare networks responding to a local covid outbreak.
Augmenting human capacity like we’re doing with ML algorithms today is only the beginning of the A.I. revolution. Next month, in Part 3, we’ll take a peek at how the recruitment industry and broader society may be impacted by an exponential acceleration in A.I. capability over the coming decades.
Dr. Jon Krohn is GQR’s Chief Data Scientist, based out of New York.
As the Chief Data Scientist, he manages scientists and engineers in order to devise intuitive and efficient machine learning algorithms for embedding within products and processes. Dr. Krohn’s particular specialization is data modeling approaches that involve passing the natural language of billions of documents through deep neural networks.
The algorithms he has designed automate aspects of millions of job applications made worldwide each year. He accelerates hiring managers’ capacity to fill their vacancies and the speed with which recruitment consultants can identify roles that candidates are perfectly suited for.
Blue-chip corporates have done global searches across hundreds of vendors that automate recruitment and Krohn’s models placed first. Third-party investigations of his models have found they offer orders of magnitude accuracy improvements relative to their existing approaches. Large HR tech platforms trust these algorithms that lie behind their screens. Krohn has published his results and applied for a patent, with more patents to come.
Dr. Krohn’s first book, Deep Learning Illustrated, was published in 2019 and became an instant #1 bestseller that was translated into six languages. He’s renowned for his lectures at Columbia University, New York University, the NYC Data Science Academy, prestigious industry conferences, and a range of digital channels including his reliably sold-out 600-person classes in the O’Reilly learning platform. He holds a Ph.D. in neuroscience from Oxford and has published on machine learning in leading academic journals since 2010; his papers have been cited over a thousand times.
![[Video] Deep Learning + HR With Chief Data Scientist, Dr. Jon Krohn](https://no-cache.hubspot.com/cta/default/2998873/ca372e4c-07e3-43cc-bd40-d15343454cef.png)
