Return to the work overview

Do We Really Need a Human Here?

A short write up on my learnings as an AI focused intern this last summer

During the summer of 2026, I worked an internship as an AI enablement engineer where I looked at current business processes and assessed how they could be automated and updated using AI.

The most important thing I learned was that we should not just be asking how AI can improve existing processes. We should also be asking whether those processes need to exist in the same form at all.

When looking at these workflows and business processes, I started off from a traditional perspective, seeing what was already there and thinking about how we could apply technology to make it “better.” I realized that with the tools we now have, and how powerful AI has become, there is an opportunity not just to improve these processes, but to rethink them entirely.

We now have tools that are able to reason, act, and operate across increasingly complex tasks. With this in mind, we should be asking not just, “How can we improve this process?” but, “Should humans even be doing this anymore?”

Of course, human oversight is still a factor.

But a lot of digital processes, when broken down into their different actions, are actually highly automatable. The question I kept asking myself was, “Do we really need a human here?”

Right now, human oversight and judgment are still considered essential for most tasks because of the lack of trust and reliability in inherently probabilistic AI systems. But just because these systems do not produce the exact same output every time does not mean they cannot be trusted.

I think a better way to measure this is not just to ask how many times AI gets something wrong, but how often humans actually get it right. Humans do not have 100% accuracy either. They make mistakes, learn from their work over time, and improve through experience.

AI systems can increasingly do something similar through feedback, evaluation, and iteration.

Long gone are the days of simple chatbots. We are entering an era where more complex decisions and actions are being passed off to AI.

As these systems continue to prove themselves, I think we should become more willing to trust them with work that we previously assumed required a human.