What I Learned Judging Global Awards In The AI Era

Judging used to feel like a quiet craft. You read a case study, check the brief, weigh the results, and argue for the entries that moved the field forward. In the last two years, the job changed. Generative AI did not replace creativity, but it did change how we recognize it. The work is faster, more synthetic, more collaborative, and sometimes more confusing. Good juries had to evolve.

This is a personal field note from serving on juries that evaluate creative, product, and technology work.

Clarity on process
The juries I serve on do not use AI systems to render scores or decisions. All evaluation is performed by human judges through reading, discussion, and voting. Any references to AI in this essay describe how entrants used AI in their work, not how jurors evaluate. Organizers may use basic administrative tools for logistics, but the judgment is human. The goal is simple. Share practical lenses that help panels reward real innovation, filter synthetic noise, and keep judging fair in a world where models can draft, design, and even decide.

1) Originality looks different now
Many entries show a polished surface. That is not proof of originality. In the GenAI era, we look for creative decisions that a model would not take on its own. Where is the human constraint, the insight that shaped the prompt, the data curation, the custom tool, the unexpected choice that solved a hard problem for a real user? If an entry can explain that chain of choices in plain language, it earns credibility.

2) Disclosure builds trust
Panels reward clarity. If AI tools or automation helped produce the work, say so. Explain which tools, where in the pipeline they were used, and what was human. Hidden use erodes trust. Honest disclosure lets jurors judge intent and craft instead of guessing.

3) Provenance is a feature, not a footnote
We ask for version history, content credentials, or other proof of authorship when entries hinge on media integrity. A simple provenance trail protects teams and users. It also speeds up deliberation because judges can verify rather than speculate.

4) Outcomes still matter most
A beautiful demo without adoption is a prototype. Juries look for outcomes: real users served, measurable lifts, fewer harms, repeatable results. Creativity is the spark. Impact is the flame. The best entries show both.

5) Safety and dignity are part of quality
Great work does not harm people to hit metrics. We evaluate how teams handled privacy, bias, misuse, and user dignity. Did they test edge cases. Did they retire risky features. Did they give people a way to reach a human when stakes were high. Responsible craft is now table stakes.

6) The five questions that keep debates honest
When a panel is split, these questions help:

1. What problem did this solve that actually mattered to someone.
2. What did the team do that was hard to copy next week.
3. What evidence shows the result was not a fluke.
4. What could go wrong at scale and how did they reduce that risk.
5. What did this teach the field that others can build on.

7) How juries evaluate AI assisted work fairly
Fair judging does not punish tool use. It rewards craft. We separate three layers:
• Ideation and direction. The human insight that frames a solution.
• System design. The pipeline, data care, and constraints that guide the tool.
• Execution. The final outputs and how they perform in the world.
An entry can use AI at any layer and still show excellence if the team demonstrates judgment across all three.

8) Red flags juries notice
• Vague claims about training data or user consent.
• Stock prompts or style transfers with no clear purpose.
• Synthetic testimonials, inflated metrics, or screenshots without context.
• Over indexing on novelty without a plan for safety or maintenance.
These are fixable, but they slow trust.

9) What entrants can do to stand out
• Show your brief and constraints.
• Document one key decision where human judgment beat a generic prompt.
• Include a short risk log and what you changed because of it.
• Provide before and after metrics and one independent validation.
• If you built internal tooling, include a diagram or short video of the workflow.
Entries that teach the jury something new tend to win.

10) What award programs should update
• Add a disclosure field for AI use and automation.
• Ask for provenance or version control in categories that hinge on authenticity.
• Provide a short rubric for safety and dignity.
• Publish category specific definitions of originality with examples.
• Give juries time to request clarifications, not just scores.
These changes reduce noise and improve fairness without killing speed.

11) Cross border judging makes the work better
Panels that mix creative directors, product leaders, engineers, ethicists, and security minds make sharper calls. The best debates come when someone who has shipped at scale sits next to someone who defends users, and both sit next to someone who lives in the culture the work serves. Diverse rooms see blind spots faster.

12) The big shift
Juries used to ask, is this beautiful and effective. We still ask that. We now also ask, is this responsible and repeatable. The teams that thrive in this era do not treat safety as PR or AI as a gimmick. They treat both as craft. That is the work worth honoring.

The GenAI era will not end judging. It will make judging a little more like the work we celebrate. Collaborative, transparent, and focused on outcomes that keep people at the center.