There is a particular kind of conversation happening around AI right now that makes me a little nervous. How much can we automate? How quickly can we adopt it? How many parts of the workflow can we put AI into? How much more can one person do now that they have it?
I understand the excitement. I use AI constantly. I build with it. I experiment with it. I have learned an incredible amount because of it. I also think the urgency to put AI everywhere can make us skip a more important question:
Because I don't think the goal should be to find as many places as possible to insert AI into recruiting. I think we should look at the work first.
Where are people carrying unnecessary cognitive load? Where are we repeatedly rebuilding context? Where does institutional knowledge disappear? What work is repetitive without becoming more valuable through repetition? Where would better access to information improve a decision? Where are people spending energy operating the process instead of thinking about the work?
Then we can ask what technology should carry. That's a very different starting point.
I want AI to carry weight.
There are things humans are carrying because, until relatively recently, there wasn't a particularly good alternative. Remember this. Find that. Summarize these things. Compare those things. Bring the context back together. Help me see what's repeating. Help me understand what changed. Help me get from a giant pile of information to the part that deserves my attention.
I am delighted to hand some of that over. Not because I want to do less. Because I want my effort going somewhere more useful.
Recruiting requires an enormous amount of context. A search develops history. Decisions get made. Calibration evolves. Candidates move. Markets respond. People learn things. Assumptions change. And humans are expected to somehow carry a surprising amount of that information around in our heads while moving between conversations, roles, stakeholders, and decisions. That is exactly where technology gets interesting to me.
Help preserve the institutional memory. Help surface the pattern. Help reduce the repetitive work. Help me interrogate what I'm seeing. Give me a better starting point. But don't confuse supporting my judgment with replacing it.
AI is allowed to be wrong.
Actually, I expect it to be. It hallucinates. It can misunderstand context. It can produce something that sounds extremely confident and is extremely incorrect. Sometimes it gives me an answer and my immediate response is essentially, Girl, absolutely not. That's useful.
Because working with AI isn't passive for me. I challenge it. I validate it. I ask another question. I give it more context. I look at what it's assuming. I make it argue the other side. I ask what I'm missing. Sometimes I throw the wildest version of an idea into the room just to see what happens when we pull it apart.
That process has become valuable to me in ways I didn't expect. AI didn't just help me produce things faster. It gave me another environment in which to think.
The thought partnership surprised me.
When I started using AI, I expected assistance. I did not expect reflection. There is something interesting that happens when you have a place where you can put down a large amount of your thinking and then interrogate it. You can ask:
What keeps repeating here? Where am I making an assumption? What would someone with a completely different perspective say? What am I actually trying to communicate? Does the evidence support what I think it supports? What don't I understand yet?
I've used AI to learn unfamiliar concepts. To understand data differently. To examine communication. To pressure-test ideas. To explore systems. To find language for things I understood intuitively but hadn't yet named. It has challenged my thinking. And occasionally it has reflected something back to me that was sitting in plain sight.
One of the strangest examples was realizing that I was a systems thinker. I genuinely didn't know. I thought I was designing better experiences. I was looking at repeated friction, identifying patterns, figuring out where something was breaking, and building ways to make the experience work better. AI essentially reflected the body of work back and said: You understand that those are systems, right? Well. Apparently. 😂
That didn't give me a new capability.
That distinction matters.
AI doesn't create the human operating it.
This is one of the reasons I get uncomfortable when we talk about AI as though access to the tool automatically produces expertise. It doesn't. AI is not calibrated to you simply because you opened a chat window. It doesn't automatically understand your judgment. Your standards. Your context. Your ethics. Your organization. Your craft. It doesn't know which answer feels wrong because of something you've learned over twelve years of doing the work.
It can help you reason. It cannot supply all of the reasoning you never developed. And I think that's where some of the conversation about replacing people starts to fall apart for me.
If an experienced recruiter and someone who has never recruited before use the same AI tool, they do not suddenly possess the same recruiting capability. The tool may be identical. The humans aren't.
One knows what to challenge. One recognizes when context is missing. One has lived through enough searches to notice when something doesn't make sense. One understands the consequence of getting the decision wrong. One knows when the technically correct answer is not the responsible one.
There are things I don't want AI deciding.
I don't want AI determining someone's worthiness for an opportunity and quietly removing them from consideration. I don't want a system producing an answer and everyone treating it as objective because a machine said it. I don't want automation to become an excuse for avoiding conversation. And I don't want recruiting to become so optimized that candidates can move through an entire hiring experience without meaningfully encountering a person.
Could technology participate in all of those things? Probably. That's not the same as saying it should.
In my own work, I am much more interested in technology helping narrow where attention is needed than making the final judgment. Help me understand alignment. Help me organize information against what we've agreed matters. Help me find what deserves a closer look. And then?
A human assesses the craft. A human understands context. A human asks the next question. A human remains accountable for the decision. Because if we're willing to accept the recommendation when it's right, we also need to know who owns it when it's wrong. “AI told me to” is not accountability.
Faster isn't automatically better.
I think this is one of the traps of this moment. There is pressure to adopt quickly because nobody wants to be left behind. But speed of adoption and quality of adoption are not the same thing.
Sometimes the responsible thing is experimentation. Try it. Learn. See where it helps. See where it breaks. Understand what people are comfortable with. Understand what the technology is actually good at. Understand what happens when it gets something wrong. Then build from what you've observed.
I would rather find a real point of friction and ask whether AI can responsibly help solve it than decide we need AI in sixty percent of a workflow and go hunting for places to put it.
The work shouldn't have to contort itself to prove we're using enough technology.
I don't want less humanity.
I want less unnecessary load around it. That's probably the simplest version of my AI philosophy.
Let technology carry the things it can carry responsibly. The repetition. The retrieval. The synthesis. The administrative weight. The context we keep losing and rebuilding. The first pass that helps a human know where to look. Give that weight somewhere else. And then give the human their attention back.
Because I know exactly where I want to spend mine. Talking to people. Understanding motivations. Recognizing hesitation. Challenging assumptions. Reading the evidence. Advocating. Creating. Solving the problem underneath the problem. Making decisions with other humans. Doing the parts of recruiting where judgment actually matters.
AI can help me do those things better. It can challenge me. Teach me. Extend me. Occasionally read me for filth. But the accountability?