Artificial Intelligence affects almost everything these days.
So what does AI in HR mean to you now?
With the arrival of ChatGPT, Generative AI that is based on algorithms known as Large Language Models (LLM) firmly established itself in the public consciousness as the ultimate AI. It is widely believed that it will propel technological progress to the next level (if used wisely). However, some will be surprised to learn that Generative AI is not the only game in town. And most of us are already using it.
We’ve known about the revolutionary potential of AI to transform our lives for decades, and we’ve been working on it for just as long. Here is a short video from almost 10 years ago where Xconomy asked me to predict the future of technology. What I was referring to in this video is AI. But not Generative AI, rather its more tenured colleague Narrow AI.
Narrow AI is hardly a new technology. For years, it has been powering countless business, commerce, production and science processes. For example, PepsiCo started embracing AI for all its business processes in 2020, and they are not alone.
So let’s unpack the difference between Generative AI (i.e. ChatGPT), which has captured the public imagination recently, and Narrow AI, which has already quietly permeated many aspects of business and life.
Why AI in HR matters
AI has the potential to benefit HR in numerous ways if applied properly — from decreasing time to hire by predicting candidate success to building diversity through the removal of bias from the hiring process or improving the candidate experience with chatbots that answer any applicant question.
But in the world of talent management, Narrow AI will most likely play a more significant role than its more spectacular counterpart — Generative AI. This is because Narrow AI is much better tuned to your specific goals and objectives and is easier to control for biases.
Understanding the differences between these two types of AI will help you identify all the ways AI could benefit your hiring process, where each type of AI works best, and how to evaluate an AI technology for your organization.
The differences between generative and narrow AI
Scope of knowledge
The very name Narrow AI implies that the subject matter that this type of AI deals with is limited. This limited subject matter could be maximizing a financial return or recognizing a face. When applied to HR, Narrow AI can understand and predict what makes a good employee in any given organization. Regardless of the type of machine learning or predictive algorithm, Narrow AI is usually highly specialized in one specific thing.
Generative AI can have some areas of specialization too, but these areas are typically far broader — for example, answering questions, creating content or providing companionship. In talent management, you might see Generative AI as a chatbot that answers candidate questions.
Learning data
Generative AI learns from a vast amount of media forms that convey information about the world: words, images, sound bites and videos. This allows Generative AI to connect words with other words, or words with images and sound bites in a way that creates the illusion that it knows what it’s talking about. This is why we see job applicants using Generative AI to write or enhance resumes and cover letters.
It can understand the job requirements listed in a publicly available job description and string together content that fulfills those requirements. The ability to learn from this vast amount of media is also why we observe AI that starts acting like a sentient being, having a personality, emotions and feelings.
Narrow AI, on the other hand, usually learns from specific data that represents the very nature of an outcome that it’s designed to predict. For example, to predict weather patterns it learns from historical data on conditions like winds and precipitation.
To make a new drug, pharmaceutical AI learns from tens of millions of data points about the components and effects of other drugs. To assess a candidate for a job, AI learns from personality data and success metrics data to predict which candidates will be successful.
Relationship with the truth
Both AI types can be wrong but in very different ways.
Narrow AI usually deals with predicting future outcomes and, as such, it’s sometimes wrong about its predictions. No matter how accurate a model’s understanding of a subject is, no predictive tool is accurate 100% of the time.
Generative AI, however, is often wrong not about the future but about the past and the present. It does not have any understanding of a subject, but rather it has a model of its subjective descriptions. As such, it can learn all the inaccuracies, biases and falsehoods in the broad data set it learns from.
Countless uses, potential risks
Generative AI rightfully created enormous excitement, as well as anxiety, because of how much it reminds us not just of ourselves, but of an infinitely more capable version of ourselves. We immediately found countless uses for it, and just as quickly identified its potential risks, from massive disinformation to job loss to AI terrorism.
Narrow AI is less flashy, but it will continue to be as important of a tool as its Generative sibling. In the field of HR, Narrow AI will continue to be the dominant technology for a few reasons:
- bias mitigation
- reliable focus on specific outcomes, and
- explainability
That is not to say that there is no place in HR for Generative AI. We all could use a recruitment chatbot that is genuinely helpful.