Not everything that holds a project up is visible.

This section maps the systems, tools, contradictions, and values that shaped this project, from its ethical commitments to its infrastructural choices. It’s about the bones beneath the body, the scaffolding behind the page, the methods that made the method possible.

On Content:

This report is not authored in isolation. It is a composite of multiple voices, conversations, and enquiries, gathered through semi-structured interviews, field notes, institutional case studies, desktop research, and interactions with AI language models.
The research process includes:

1. Interviews and field observations with mentors, mentees, and arts educators
   
2. Desktop research and policy review on mentorship models
   
3. AI-assisted exploration using OpenAI’s ChatGPT to gather perspectives, structure questions, and test hypotheses
   
4. Reflections on lived experience from the author and collaborators
   

As such, the final document is an assemblage of enquiry, not a singular assertion. The AI tools used were trained on human-generated data, reflecting multiple cultural, social, and geographic contexts. The aim here is not to produce fixed knowledge but to map a terrain of questions, tensions, and emerging models around mentorship in the arts.

On Infrastructure:

This website is hosted on a solar-powered Raspberry Pi — a tiny, energy-efficient computer positioned near a window in New Delhi, India. It runs on a minimalist web server (Lighttpd) and serves plain HTML pages with dithered black, white, and red images. The intention was simple: to build something light. Light on energy, light on bandwidth, light on ego.

But to make something light, I had to use something heavy.

Nearly every part of this project — from writing structure and research organisation to server configuration, HTML scaffolding, and image compression — was assisted by AI, especially ChatGPT. I asked questions I didn’t know how to frame, rewrote lines, corrected grammar, and translated concepts into code — all while learning, iteratively and imperfectly, as I went. The text you’re reading now was drafted, redrafted, and refined in a feedback loop between me (a human artist) and a machine trained on billions of words.

The irony isn’t lost on me.

To build a slow website, I leaned on real-time machine intelligence.

To reduce file sizes, I used an online dithering tool —ditherit.com — powered by distant servers.

To escape the cloud, I still had to pass through it.

But that’s what makes this infrastructure “quiet.”

It doesn’t pretend to be pure.

It holds the mess — and makes it transparent.

This project was not AI-generated.

It was AI-assisted — human-led, artist-driven, and built with intentional friction.

Every line was questioned, reworked, or challenged. Every tool was chosen not for trendiness, but for what it allowed me to build — ethically, lightly, and on my own terms.

In keeping with responsible practice:

1. AI is not listed as an author, but is fully acknowledged as a collaborator and mentor in code.
   
2. All text, code, and visuals were curated, edited, and finalised by the author
   
3. No proprietary data or models were used unethically
   
4. This work does not claim objectivity — it claims process, care, and curiosity

This attribution aligns with current best practices for AI use in artistic and research-based projects, where the human remains the primary author and curator, and AI is transparently acknowledged for its contributions to structure, language, and technical support.

And so, this architecture holds not just a website — but a set of questions:
What does it mean to publish slowly in a world obsessed with speed?
What happens when we treat infrastructure as a form of authorship?
How do we honour both sunlight and silicon, without pretending one is better than the other?
This project is one small answer.
“To make something light, I used something heavy.
To publish slowly, I moved through a tool trained to answer instantly.
To build something sustainable, I turned to something that is not.”

Credits & Hosting

This project is the result of many layers of collaboration, reflection, and experimentation — human and non-human.

1. Author, Concept & Design: Pooja Bahri, Multidisciplinary Artist, Co-Founder, Art For Art Foundation
   
2. Research Assistant: Suparna Aggarwal
   
3. Technical Assistance: Jagadeesh Reddy
   
4. Technical Guidance: Kunal Kalra
   
5. Coordination Support: Tarini Wadhawan
   
6. AI Mentorship: OpenAI ChatGPT
   
7. Server Inspiration:Low-Tech Magazine
   
8. Published by: Art For Art Foundation
   
9. Hosted on: Raspberry Pi 4, powered by solar energy
   
10. Website Design: Minimal HTML/CSS(w3schools) — no trackers, no cookies — built by Jagadeesh Reddy
11. Generous Sponsor: Sonalika CSR

With gratitude for her continued vision and support: Archana Sapra, Co-Founder, Art For Art Foundation

Tools & Infrastructure Notes

1. Web Server: Lighttpd on Raspberry Pi 4
   
2. Hosting: Powered by a 150W monocrystalline silicon solar panel connected to a 12V 40Ah battery, charge controller, and 200W power inverter (off-grid system from Amazon India)
3. Static Site: Custom HTML, CSS, and image optimisation
   
4. AI Assistance: OpenAI’s ChatGPT (writing refinement, coding help, conceptual structuring)
   
5. Image Compression:ditherit.com for dithered image conversion
   
6. Tunnel & DNS: Cloudflare Tunnel with dynamic DNS
   
7. Content: Authored and curated by Pooja Bahri with research support from Suparna Aggarwal
   

This project is not automated. It is assisted. And it is accountable.
 — Pooja Bahri & ChatGPT

This attribution aligns with current guidelines and ethical considerations published by major institutions and journals on the responsible use of AI in creative and research-based work. For example:

1. The Modern Language Association (MLA) states that AI tools may be cited as sources but should not be credited as authors (MLA Style Center, 2023).
   
2. Nature and Science journals recommend that large language models (LLMs) like ChatGPT should not be listed as co-authors, but their use must be disclosed transparently.
   
3. OpenAI’s usage policy encourages transparency and discourages misrepresentation of AI-generated content as solely human.
   
4. Royal College of Art, MIT Media Lab, and the Stanford Institute for Human-Centered AI have each released internal guidance encouraging reflective, human-led use of AI in authorship and publishing.