Week 12 (Mar. 30, 2026 – Apr. 3, 2026)

Just like how the weather in Pittsburgh keeps changing every day, Week 12 was full of different things for us. Because of that, this might be one of our longest blogs so far.

We started the week strong with our playtesting session with a naive audience on Saturday, March 28, 2026, from 9:30 AM to 3:00 PM. For this, we created a short performance and gathered feedback. Our idea was a late-night talk show hosted by our AI goose, Alice. The setup included a human actor as the guest, the AI goose as the host, and a hidden team member under the table controlling the puppet and UI.

We ran about 10 sessions, each around 20 minutes long. Our audience ranged from teenagers to adults over 50. Overall, the playtesting went well, but we got mixed feedback. Some key takeaways stood out:

• The AI works best when it is short, clear, and responsive. Long or repetitive dialogue breaks immersion.
• The show needs a stronger structure, especially a better opening and more variety throughout.
• Adding unexpected moments (like glitches or surprises) can make the experience more engaging.
• Audience interaction could make a big difference—like reacting to claps or laughter.
• An animatronic beak would make the goose feel more real.

After some much-needed rest on Sunday, Monday was our second process grade meeting with instructors. These one-on-one sessions were a chance to reflect on how things have been going—our work, the team, and the project as we move toward finals.

Later, instead of our usual presentation-style meeting, Mo and Brenda came into our project room and walked around to see what each of us was working on. It felt more engaging and real. They also mentioned they prefer live demos over slides, which was good to know going forward.

In the afternoon, Anthony Palyszeski approached us with an exciting opportunity—to perform our project for kids from Assemble.

So on Tuesday from 2:30 PM to 3:30 PM, we ran a version of our talk show for kids. We had three groups of 6–7 students each, along with a teacher. We adjusted our approach a bit for them. Before the show, we explained the idea and asked them to help us come up with a funny invention.

We asked them to suggest:

• something you find in a house or kitchen
• and something that thing should not do

The answers were hilarious—like a coffee machine that turns your skin green or an oven that cools mac and cheese.

We didn’t reveal the goose until the performance started, and we only told them about the AI after it ended. The show lasted about 4 minutes. We heard lots of giggles and laughs, which felt great.

Afterward, we did a short Q&A. We asked what they understood, whether they thought the voice was AI, and if knowing that changed how they felt. Their responses were really interesting—honestly, some of the 3rd graders gave very thoughtful answers. We ended up enjoying this session a lot.

On Wednesday, we had our second retrospective as a team. Like last time, we talked about what we should start, stop, and continue doing. We also discussed risks, concerns, and gave shoutouts. It was a helpful session to reflect and improve how we work together.

Thursday was our second meeting with Mark Piszczor. We shared our progress, and he mentioned it was interesting how much we’ve discovered through this project. That felt reassuring, especially since the project has been quite open-ended. Mo had earlier said that a good sign is when you have enough material to start cutting things out—and we’re starting to reach that point.

On Friday, we met with Brenda to plan for softs and the final presentation. The discussion was really helpful, and we now have a clearer direction for the next steps.

Throughout the week, the team made some solid progress:

• Cleaned up our Git and combined multiple project directions (talk show, dating show, detective game, improv exercises)
• Continued building our design document (project context, design principles, system flow, tech)
• Ran a Twitch session where two AIs talked to each other—it got repetitive quickly, which led to an interesting insight
• Built a simple Arduino setup to move the goose’s beak based on audio
• Recorded a demo performance using improv exercises
• Tried Gemini AI for real-time voice and prompts—it looks promising so far

Next week, we plan to build on all of this, refine our ideas, and start preparing seriously for softs and finals.

And with that, we reached the long (and slightly exhausting) end of Week 12.