I’m often asked to explain what our Artificial Intelligence actually does. My first reflex is to jump to the whiteboard and gush about the magic of vector space embeddings over messy diagrams of deep neural network architectures. Usually, my audience politely waits until I pause to catch my breath, before gently explaining that they wanted to understand what it means for an end-user.
It’s a great question, and it’s easy to explain. Without any knowledge of vector space embeddings.
Here’s the current state of affairs: if you use job boards and other search tools to comb the internet for candidates or jobs, you’re likely using a keyword-based approach.
For example, you could try searching for positions requiring proficiency with the programming language Python in New York. I get about 5,000 results using a typical recruiting site.
Here’s the problem: Python can be used in many different ways. It can mean very different things in different working environments. And there are different types of engineers that use Python every day.
There are data scientists who write Python code to create models.
There are software developers who write Python code to build applications.
There are DevOps engineers who write Python code to manage infrastructure.
To distinguish between these, you need to understand the natural language in which resumes and job descriptions are written. You need to look at the context in which words are used. You need to understand nuanced terms within these documents.
When you take the time to appreciate and account for these distinctions, you improve the chances of matching the right kind of engineer to the right kind of position.
That’s what our Artificial Intelligence actually does.