We’re excited to announce the beta launch of our Machine Learning (ML) solution, a product of more than two years of dedicated effort aimed at revolutionizing how matches are made. This isn’t just an update; it’s a complete reimagining of the matching process, designed to uncover the potential matches traditional algorithms might miss.
Why We Built This
Our mission was clear from the start: to eliminate bias from the matching process and ensure no opportunity is overlooked. With advanced Unsupervised Machine Learning at its core, our solution learns from data to make matches based on true compatibility, not preconceived biases.
Why We’re Not Using LLM Models for Matching
OpenAI’s GPT and similar LLMs aren’t built for specific tasks like comparing data objects. Their broad approach can lead to answers that don’t fully align with our unique data formats, highlighting the need for more targeted experimentation.
The absence of ready-made AI solutions for our recruitment process led us to develop our own. Our exploration into this space is informed by existing research and the opaque AI systems of competitors.
To uphold our commitment to unbiased results, we’ve opted for Unsupervised Machine Learning. Despite OpenAI’s support for aspects like word embeddings, it falls short on the advanced techniques we employ.
Clustering in machine learning is a method used to group similar data points together without prior knowledge of the group definitions. When applied to recruitment data to find good candidates, clustering works by analyzing the characteristics and qualifications of applicants and grouping them based on similarities in these attributes.
The inherent bias in LLMs trained via supervised methods poses challenges. Our aim is to minimize this bias, aligning with our goal for a fair matching engine.
As an example let’s imagine a company using an LLM to help write job descriptions. If the model has been trained on historical data where most engineering job ads were targeted at men and most nursing job ads were targeted at women, it might unintentionally continue this bias. As a result, job descriptions generated by the LLM could be subtly more appealing to one gender over another, even if that was not the intention.
What This Means for You
Our ML solution paves the way for less biased matches, uncovering potential pairings that a manually crafted algorithm might overlook. It merges the discernment of human judgment with the objectivity of AI to ensure a fairer, more inclusive matching process.
The Future Starts Now
As we introduce our ML solution in beta, we’re not just launching a product—we’re inviting our clients to join us at the forefront of innovation in matching technology.
Now we want to hear YOUR feedback! You can give feedback on each individual match made on our platform, in order for us to continue to improve.
Welcome to the next level of matching, where innovation meets opportunity.
