[Season 6] AI THINKING LAB — Midterm Presentation Report (2026/3/19)
This lab is a hands-on program where participants combine AI with their own skills, hobbies, and professions to create something uniquely their own. At session 6 of 12 — the halfway point — each participant presented what they've accomplished so far.
The event was held in a hybrid format, both in-person and online. Guests joined remotely, and the session was filled with lively questions and comments.
Here's a summary of each participant's presentation.
Participant 1: M — Using AI to Support Career Counseling for Young People
What's the purpose?
M is active as a mentor for young people. Currently, he is supporting a young woman who hopes to eventually settle abroad permanently. Having moved from a rural area to Tokyo, she needs help building an independent life without becoming isolated or losing her mental well-being.
What are you building?
M is accumulating information gathered through online consultations and face-to-face conversations in NotebookLM, then linking it with Gemini to create a system that can address the core concerns of the person he's advising. The goal is to build a database that leads to concrete advice — organizing conditions for living abroad, local network information, and more.
What's the happy outcome?
The person being advised can envision their future without feeling rushed
M himself gains new industry knowledge and AI skills
Eventually, by swapping in different consultation content, this could become a versatile "counseling agent" usable for other young people as well
Instructor's comment: The connection between NotebookLM and Gemini is still in an exploratory stage. The key question is: "How do you feed M's own experiences and perspectives into the AI to generate original, personalized responses?" Turning his own data into a database is likely the crux of this project.
Participant 2: O — Using AI to Create Time and Focus on People
What's the purpose?
O works full-time at a construction company while also running a side business as a planner/journal instructor and fortune teller. Working from 8:30 AM to 5:30 PM with children at home, she has no shortage of things she wants to do — but a serious shortage of time. SNS posts, customer management, creating reading reports... the administrative burden was overwhelming.
What are you building?
Three areas of AI application are planned:
SNS draft creation — Streamlining posts for Facebook, Instagram, and note
Fortune-telling report & material design — Reducing the time spent creating from scratch
Customer & student database — Also useful for LP creation and form design
She has also started accumulating her own information in NotebookLM.
What's the happy outcome?
Operating under the theme "Mari's Waku-Waku (Excitement)," O envisions AI handling the organizational work so she can focus on connecting with people. The goal is higher-quality service, more thoughtful one-on-one dialogue, and becoming someone clients truly appreciate.
Instructor's comment: Beyond her own data, incorporating feedback from students and clients into the AI could also help improve her own approach. A system that captures both her voice and the voices of those she serves would make this even stronger.
Participant 3: Y — Using AI to Streamline Podcast Audio Editing
What's the purpose?
Y runs a podcast called "Mitaka no Mikata-ban" (Mitaka's Ally Edition), where he personally covers and reports on local community news. His biggest struggle was audio editing — removing filler words like "uh" and "um" from his speech was an enormous task that drained his motivation to move on to the next episode.
What are you building?
The process: upload audio data → convert to text → detect frequently occurring filler words → let the editor decide whether to cut each one → auto-edit to generate a clean audio file. The design intentionally keeps the final decision in human hands rather than being fully automated.
He has already had ChatGPT write Python code and has run it in Google Colab.
What's the happy outcome?
Easier editing keeps the motivation to publish going
Being able to tell interviewees "filler words will be edited out automatically" lowers the barrier for people who aren't confident speakers
More frequent local coverage means more information and people flow into Mitaka, energizing the community
Challenges shared honestly:
Adobe Premiere Pro already has a built-in filler word removal feature (since 2024)
AI voice synthesis to replicate his own voice was attempted, but the result sounded flat and monotone
Japanese filler detection still has accuracy issues — sometimes cutting parts that shouldn't be removed
In conversations with good rhythm, filler words aren't necessarily bothersome
Instructor's comment: Y has made the most hands-on progress among all participants. Japanese filler detection is genuinely difficult, and if built out properly, this could even have monetization potential. Publishing it as an API with a paid model is conceivable — this is a remarkably advanced project.
Participant 4: I — Using AI to Suggest Healthy Menus from Home Ingredients
What's the purpose?
I joined the AI Lab after retiring, with time on his hands and pure curiosity about what AI actually is. After some deliberation about what to create, he landed on the idea of inputting household ingredients and his own medical history to have AI suggest suitable healthy menus.
What are you building?
He organized a list of ingredients and medical conditions in Excel, then input it directly into ChatGPT. Following Steps 1 → 2 → 3, the AI properly output menus along with cooking instructions.
What's the happy outcome?
Personal health improves
Happiness spreads to family and those around him
Contributes to reducing medical costs (addressing the issue of rising healthcare expenses for the elderly)
Helps reduce household food waste
Instructor's comment: I says "it's already done!" — and simply experiencing AI firsthand is itself a huge achievement. Going forward, there's room to explore further: using custom instructions for personalized settings, generating different menus each time, or even recognizing ingredients from photos. No rush — let's take it easy over the remaining six sessions.
Participant 5: S — Using AI to Lighten Promotional & Reporting Work and Focus on Care
What's the purpose?
S works as a nurse at a primary care clinic while also pursuing community activities as an aromatherapist. Every event brings scheduling coordination, promotional copy writing, and activity report organization — eating into the time she most wants to protect: time spent face-to-face with people.
What are you building?
A system with a clear division of labor between AI and humans:
AI handles: Information organization, drafting Instagram announcements, Facebook posts, and activity reports
Humans handle: Final review of expression, checking that the intended message and feelings are properly reflected, and making corrections
Input uses Google Forms and Notion; generation uses ChatGPT and Canva. Zapier and Make automate the workflow, with Buffer and Meta Business Suite handling SNS posting. Activity history is stored in Notion and Airtable, creating a cycle that feeds into future announcements.
What's the happy outcome?
A change in time — Less time on announcements and reports means more focus on preparation
A change in connection — Better announcements deepen relationships with participants
A change in sustainability — Accumulated history makes future announcements easier, helping activities continue
Instructor's comment: The concept is remarkably specific — she's already at the tool-selection stage, so the next step is simply to start using them. There's significant overlap with O's project, so they can likely learn from each other. Accumulating her own evolving thoughts and feelings as data will help the AI understand her direction over time.
Looking at the Big Picture
What stood out most was that every participant created their presentation slides using AI. The quality of slides that would have been unthinkable three years ago can now be produced in a short time with AI.
A common thread among many participants was the desire to "feed my own data into AI and get output that's uniquely mine." Accumulating information in NotebookLM, setting up custom instructions for personalized use, turning one's own thoughts and experiences into a database — this fusion of "AI × Self" is truly the heart of this lab.
In the second half of the program, participants will move into the phase of actually building their projects. The final presentation is scheduled for Thursday, April 30. We plan to hold it as an open event once again, so stay tuned!
AI THINKING LAB operates out of "BASES," an atelier and café in Gakuen-Nishimachi, Kodaira City, Tokyo. Each session is documented on note. Writing these articles with the help of AI is itself one of the lab's ongoing experiments.
