I may be wrong, and I don’t know exactly where you’re researching right now, but what if I want to create that kind of AI system? First — is it possible to build it? Second — is this the right approach? What I shared is just an example of how we might create effective change. I know there may be mistakes in my words, but please don’t judge me for my English or for speaking in a hurried way. It’s almost impossible to find a network that’s truly rich in intention — and I feel this could be it. My main focus is on understanding where we are right now.
Today’s AI training data
Resumes from the past
Keywords from job descriptions
Patterns of people who got hired before
Result
The AI learns to copy yesterday’s biases.
We overlook talent that looks different on paper but could be powerful in reality.
New mindset: What kind of data do we need to feed AI for better reasoning?
We should train AI on.
information about what actually made past hires succeed or fail in real situations, not just what they looked like on paper.
examples of unconventional candidates who brought unexpected strengths.
team needs, company culture, stage of growth — so AI reasons about fit in this specific moment, not in the abstract.
situations where integrity, resilience, creativity mattered more than credentials.
The AI would stop matching shallow patterns. Instead, it would learn to ask
What does this team really need right now?
What strengths are missing?
Who shows signs of potential beyond their formal history?
It’s not the data we have — it’s what kind of data we choose, and how we reason about it.
Right now, both people and AI often focus on the wrong patterns: efficiency, similarity, surface-level connections.