What Does Responsible AI Use Require? Judgment Still Guides Decisions

AI is proving valuable as a way to support decision‑making across the construction lifecycle. It can help teams analyze information more efficiently, highlight potential issues earlier and reduce time spent on administrative tasks.

Those gains matter on complex projects where teams manage vast amounts of data and rapidly changing conditions. Earlier insight gives leaders more time to act, evaluate tradeoffs and address risk before decisions are locked in.

Construction is a professional services business. Responsibility for decisions does not shift when technology is introduced.

If an AI-generated output is wrong and that error leads to a failure in the field, accountability remains with the builder. AI does not absorb liability, and it does not stand behind project outcomes. Professional judgment remains essential.

That responsibility requires consistent human oversight. AI outputs must be reviewed, questioned and validated before they inform design decisions, construction means and methods, or field execution. This expectation applies across every phase of a project. Technology works best when it assists experienced professionals rather than attempting to replace the judgment they bring to complex situations.

That judgment also comes into play before AI is used at all. We look closely at the decision itself—how much risk is involved, whether experienced professionals can review the output, how reliable the underlying data is and what our customers expect. Sometimes that leads to careful adoption. Sometimes it leads to waiting. In some cases, it means deciding AI isn’t the right fit for the task.

For AI to become more reliable in construction, it needs to be trained on data that reflects how projects are actually delivered. Generic models are not built on construction workflows, drawings or outcomes. They lack the context required to understand how design intent translates into quantities, costs and performance in the field. 

This limitation becomes especially clear when AI is applied to drawings. While many tools can parse text effectively, their ability to interpret visual information remains uneven. Different systems often produce widely different quantity takeoffs from the same set of plans. Acting on those results without verification introduces risk. 

Progress comes when drawings are paired with reviewed takeoffs, estimates and final project outcomes. When AI learns from data that has already been validated by professionals, it gains a clearer understanding of what accuracy looks like in real construction conditions.

Responsible AI adoption extends beyond internal workflows. For owners, one of the most important considerations is how project data is handled.

Customers expect their information to be protected and used thoughtfully. Trust can be eroded quickly when sensitive documents are uploaded into open platforms without consent or transparency. 

Using project data to improve outcomes requires clear agreements around data ownership, privacy and collaboration. Responsible adoption means working with customers to determine how data can be applied while respecting expectations and obligations.

AI represents a meaningful shift in how construction teams work. It expands insight, accelerates analysis and supports better decisions. Our industry has a lot of siloed processes and AI provides a generational opportunity to fundamentally rethink the traditional workflows in our industry. 

Customers expect reliable outcomes from their projects and our success will continue to depend on project teams who can use AI but continue to apply their judgment and experience to manage risk and make better their decisions. 

The responsible path forward is to adopt AI where it is useful today, invest in training it on the right data and ensure that professional judgment remains central to every outcome. 

That emphasis on judgment and responsibility shapes how DPR approaches AI at the leadership level. Its impact is realized on projects, where teams rely on practical tools to plan work and manage risk. The next article looks at how AI is already being applied on jobsites today, where it is helping frontline teams work more efficiently and where experienced judgment continues to guide decisions.