Same for every track

How judging works

Every team is scored on its main track, out of 100.

The five criteria

25 points

Community impact

Does it solve a real problem for real people?

20 points

AI component

Does what you built use AI or machine learning, and use it well? For example, it answers questions, sorts reports or predicts demand. Using AI only to help write code or slides isn't enough here.

20 points

Responsible use

Privacy, fairness, and being honest about limits.

20 points

Technical execution

Does it work?

A team without code is scored on how well what it built works, including a no-code tool or a mockup. A plan alone scores low here.

15 points

Feasibility and pitch

Could it really be used, and did you explain it well?

Presenting

  • Each team gets 5 minutes: 4 to present and 1 for questions.
  • Your presentation is your demo. Show it live. No videos.
  • If your live demo breaks, show screenshots of it working.
  • Judges score the live presentation. Your team folder is the record they can open to check links, code and slides.
  • Include one slide called "AI we used and how we checked it".
  • Presentations start after submissions close. See haihackathon.com (opens in a new tab) for the schedule.

Scoring

  • Judges score each team right after it presents.
  • A judge who knows someone on a team, or mentored it, doesn't score that team.
  • We expect judges to score fairly and professionally. HAI staff and the judging lead review every score before results are final.
  • The program director rules on anything the rules don't cover.

For judges

  • HAI staff will give you a judge card before presentations start.
  • Score each team right after it presents, on the main track named in its "Our project" Doc.
  • Score the live presentation. Open the team's folder (the link is on its table card) to check links, code and the "AI we used and how we checked it" slide.
  • Technical execution ("Does it work?"): a working no-code tool or clickable mockup can score well. A plan alone scores low.
  • "AI component": score AI or machine learning inside what the team built, not AI that only helped write code or slides.
  • Each challenge's "Build at minimum" is a target, not a rule. Use the five criteria.
  • If a live demo breaks, the team may show screenshots. No videos.
  • Ask questions only in the 1-minute question time.
  • If you know someone on a team, or mentored it, don't score that team. Tell the judging lead.
  • Send one score form per team, under your own name. HAI staff and the judging lead review every score before results are final.