Tuesday, April 21, 2026
There’s a pattern that’s becoming harder to ignore, and it shows up in places that don’t seem connected at first.
In driving, we’re starting to rely on systems that can take over enough of the task that you stop thinking about them. Mile after mile, everything works exactly as it should. And then, when something goes wrong, the expectation is that the human was paying close attention the entire time; and is responsible when they weren’t. The system performs well enough to change behavior…but not well enough to take responsibility for it.
In the legal world, there have now been multiple cases where attorneys relied on AI tools that produced convincing case citations that didn’t actually exist. The work got easier, the output looked convincing. But the responsibility for accuracy never moved, the consequences of failure fall on the attorney.
Different industries. Same shape.
The system gets very good at doing the work. Good enough that you begin to rely on it. Good enough that it starts to feel like the nature of the work itself has changed. Change that is shaped, in part, by the work moving through it. And by the people training the system.
What also happens is that control, responsibility, and ownership, once tightly coupled, begin to separate. This isn’t really about adapting to change; we’ve always done that well. But if you’ve been in the title and settlement space for a while, this conversation may feel familiar.
There was a time when agents were very deliberate, and extremely cautious, about who owned the systems we used…and why.
The concern then wasn’t about functionality; it was about visibility. If someone else owned the system, they could see our transactions. Over time, they could accumulate our work products and knowledge. We asked ourselves questions like “how long will they need us?”. Today, it seems nearly quaint that the “they” in those scenarios were our underwriters.
At the center of that hesitation was ownership. Ownership of assets: of our title plant, of the data, of the work product. There was a recognition that those things were tied to independence and long-term value.
But we don’t talk about it the same way now. We talk about efficiency. Integration. Speed. And the systems available today deliver on those things in ways that are hard to argue with. No question about it.
Yet alongside that, something else has been happening. We’ve become more comfortable placing our work inside systems we don’t control. More comfortable letting those systems shape how the work gets done. More comfortable contributing data (transaction after transaction) into other environments where it accumulates.
The data that once defined the value of the business, how work is done, how issues are resolved, and how decisions are made, is increasingly sitting somewhere else; and that’s not something we used to be casual about.
The responsibility hasn’t moved; neither has the liability. The agent still signs, the policy is still issued, and the liability still sits where it always has. But the relationship to the work has started to shift.
There’s less discovery and more confirmation. Less interpretation, more alignment with what the system surfaces. And as that happens, expectations start to move as well; turn times compress, variability becomes less acceptable, less value is placed on the upfront quality of the work.
It’s less “how did you find that?” and more “why didn’t the system catch it?”
And usually, all of this comes with a sense that adoption needs to be quick. If you don’t move, someone else will. If you hesitate, you risk falling behind. It’s hard to sit with a longer-term question when the shorter-term pressure is right in front of you.
Which is interesting, because in this industry, we spend a lot of time training ourselves to recognize one of the most consistent warning signs of fraud, urgency. We know how it can cause us to stop thinking critically. When the situation as presented makes it feel like pausing would cost you something, urgency has a way of narrowing focus; of shortening the space between question and action.
In combatting fraud, we’re trained to slow down when something feels outside of our control and urgent. I’m not sure we’ve fully applied similar critical thinking to how we think about the systems we rely on today. This isn’t about any one provider, or any one decision. The modern tools on offer are, in many cases, necessary, exciting and impressive. We welcome them to the arena.
And, we also need to notice we are also in the middle of a larger and longer shift. We may be trading away something we once protected zealously.
The systems keep getting better. The question isn’t whether they work. It’s how the balance between control, responsibility, and ownership continues to shift…and what that means over time. For example, while it’s undeniably true that Chat GPT is able to give some very good health ideas based on the complex and unique lab test results, brain scan or mammogram image someone uploads into it. It’s also true that Chat GPT isn’t subject to HIPPA.
Until Next Time,
Mary Schuster
Chief Knowledge Officer
October Research, LLC