Understanding AI for HR: A journey through 6 myths

To debunk the mystery behind AI for HR professionals, this article will cover the 6 most common myths to shed light on what is true, what is not, and a key insight revealed by each myth.

This is a guest post by Pooja Khandelwal. She is a Product Marketing Manager at impress.ai – AI Powered Chatbot for Recruiters. Interested in all things tech, from AI to IoT and machine learning. She believes in healthy living, lifelong learning, and having an open mind.

According to a recent tweet from Josh Bersin, founder of Bersin & Associates, there has been a 10% increase in HR Tech spending over the past year.

Feeding this trend is a wave of new HR platforms powered by artificial intelligence (AI). It’s no doubt that we’re living through the golden age of AI and now it’s disrupting the way we approach human resources.

With the increasing impact that AI will have on human resources, HR professionals need to gain a better understanding of the technology, before implementing it in their department.

To debunk the mystery behind AI for HR professionals, this article will cover the 6 most common myths to shed light on what is true, what is not, and a key insight revealed by each myth.


Myth 1: Deep learning = Machine Learning = Artificial Intelligence

While these terms are often used interchangeably, they have different meanings. A common misconception is that deep learning implies a “deeper” understanding of data.

In reality, deep learning algorithms learn from patterns of data over time. Deep Learning is a subfield of machine learning concerned with algorithms inspired by the structure and function of the brain called artificial neural networks.

Machine learning is a field where machines “learn” from data through a series of tests. Whereas AI refers to technology that exhibits a level of intelligence normally 

displayed by humans, like the ability to process text-based conversations or speak. An AI agent is one that acts autonomously towards its goal by sensing its environment and executing different actions.

Insight: Intelligence is about solving the problem, not just machine learning.


Myth 2: AI algorithms learn on their own based on interactions

What actually happens is that AI algorithms use historical data to create and train the model.  This model starts making decisions and goes through multiple rounds of iterations which improves the accuracy level.

Teaching the system is a human intensive task which takes place offline as opposed to online.

For example, the average data scientist spends up to 80% of their time data cleaning, which plays a role in improving the model.

Insight: Systems need to be built to minimise human effort while maximising AI learning.


Myth 3: Neural networks are exactly how the human brain is built

Computer scientists were inspired by biological systems and used them to create analogies.

For example, long-short term memory systems are referred to as bugs, viruses, and worms.

Neural networks are a class of algorithms whose idea was inspired by how biological neurons are structured. Similar to the human brain, neural networks are complex structures.

Insight: AI systems don’t need to mimic human behaviour to be useful.


Myth 4: AI requires a huge amount of data

Machine Learning needs some data to get started.

This amount of data depends on the level of creativity and the problem statement. Transfer Learning makes it possible to leverage on domains where there is a lot of data and use it in areas where there is limited data.

Simulations and inference systems can be used to increase knowledge with limited amounts of data.

For example, impress.ai’s platform can learn from as few as 10-20 data points as opposed to 1000’s or 100’s of thousands of data points.

Insight: AI systems can still work with limited amounts of data.


Myth 5: AI is a black box that auto-magically gives people insights

While certain approaches are black box methods, computer scientists don’t

always know what is going on when a prediction is made.

Even in traditional black box methods, the system can be understood without knowing exactly what is going on inside.

A relatable example would be that a person does not need to know exactly how a teammate thinks in order to be a good manager.

Insight: It’s ok not to know the math behind how a system works as long as it produces verifiable behaviour.


Myth 6: AI will take over HR jobs

To put your fear at ease, AI cannot take over your entire job, it can only take over parts of your job, specifically the repetitive and mundane tasks.

The AI revolution is just like the industrial revolution. AI, enhanced by machine learning, helps us accomplish more in less time.

Technology will cause a shift in the workforce and allow us to focus on higher value tasks such as creative thinking, strategy, and problem solving.

HR professionals will be equipped with better tools compared to traditional ATS’s and Excel sheets.

Insight: As opposed to making a decision about an HR tool on whether it uses AI, rephrase the question to “Does this tool make my work easier or better?”


Key takeaways about AI that HR professionals should know

AI for HR is designed to reduce, or even remove low-level activities like manually screening resumes or scheduling interviews.

This allows HR professionals to focus their time on higher-value tasks such as growing relationships with candidates and employees.

Regardless of the area within human resources, the future of the industry will involve the cooperation of humans and AI, because AI is here to stay for good.

Please note: I reserve the right to delete comments that are offensive or off-topic.