During my term at Dewan Kesenian Jakarta / Jakarta Arts Council, every Monday at 10 a.m. I posted the same question in a Notion workspace I had introduced: what happened last week, and what are we doing this week?
For the first two weeks, there was some movement. People opened the page, added notes, and tried to follow the structure. By the third week, the page was quiet. Work returned to WhatsApp, where messages kept moving but decisions became harder to locate. I had assumed that introducing a better system would improve coordination. The institution had absorbed plenty of new tools before; one more should have helped.
I was wrong. The problem was not the tool. Notion could store information, but it could not give anyone a mandate, settle who owned a decision, or make a new routine matter to people already carrying too much work through informal channels.
During that term, I encountered the same limit while working on an institutional position about AI in arts and culture. AI helped me discover policies, cases, and ethical questions from different countries. Humans checked the sources, argued over the language, and wrote the document. The draft was discussed, then deferred to a later pleno. It was never approved during my term.
That experience changed how I understand AI adoption. The technology may help produce a document, analyse information, or suggest a course of action. It cannot decide whether a document is authoritative, whether a discussion has become a decision, or who will remain responsible after the meeting ends.
Andrej Karpathy, an AI researcher and former head of AI at Tesla, offers a useful technical frame for this change. In what he calls Software 1.0, programmers write explicit instructions. In Software 2.0, they shape behaviour through data and training objectives. Large language models move the work again: people increasingly express intent through natural language, provide context, then judge the result.
This makes more creation possible, but it also moves responsibility into places that are less visible. The person using the system may see a prompt and an output, while the training data, commercial terms, cultural assumptions, and ownership decisions remain elsewhere.
In music, I see two responses developing at the same time. Some artists experience these tools as an expansion of what they can make. Others see their voices, images, recordings, or livelihoods being pulled into systems whose terms they did not choose. Both responses are real. Treating one as progress and the other as resistance would miss the conditions underneath them.
Nada offers one glimpse of a better direction. The app turns humming, singing, or whistling into editable MIDI. A musician who cannot play a keyboard can still capture a melody, change the notes, assign another instrument, and continue working. The voice does not disappear after the machine processes it; it remains the source of the musical decision. Nada reduces the distance between an idea and the ability to develop it.
Nadascore addresses another distance: the gap between independent artists and the data surrounding their work. Streaming platforms produce numbers constantly, but access to numbers does not automatically produce understanding. Nadascore presents a score, explains its reasoning, and suggests a next action. This cannot give an independent artist the bargaining power of a major label, but it can make platform signals less mysterious and help an artist decide what deserves attention.
Algorapture approaches technology from the other side. Instead of giving artists a finished interface, it treats code as material for sound, visuals, and live performance. Its Syntax and the City workshop introduces tools such as Strudel and Hydra.js without requiring previous coding experience. The value here is not the claim that every musician should learn programming. It is the possibility that code can enter a local artistic scene as something people perform, share, modify, and occasionally break.
These examples do different work. Nada helps a musical idea survive long enough to become editable. Nadascore helps an artist read some of the forces acting on a release. Algorapture lets artists enter the technical language itself. None removes the conflicts around ownership, access, labour, or platform power, but each gives the person making music more room to act.
That is the direction I mean by a new paradigm in music. I do not think we have arrived there, and I do not think AI companies can define it for us. Artists need enough access and technical confidence to shape the tools rather than wait for finished products. Institutions need to support that experimentation without turning every workshop into a campaign, every prototype into a success story, or every ethical concern into another document that no one owns.
My experience during that completed term inside an arts institution made this uncomfortable for me because I had believed, for a while, that clarity could be designed into a new platform. The three-year WhatsApp archive I later examined contained 16,717 entries and 72 sampled coordination episodes. Discussion appeared in 71 of them, decisions in 63, and follow-up in 64. Clear closure was visible in only 28. The institution was active, sometimes intensely so, but activity and completion were not the same thing.
The harder lesson was that much of the institution lived inside people. Someone remembered which version was current. Someone translated an artistic programme into a government budget category. Someone knew whom to call when a process stopped moving. Software could support those people, but it could not replace the trust, authority, and memory they carried.
This is also shaping the direction of ALUN. I want to work with cultural and public institutions while a technology decision is still open, when disagreement can change the outcome and experimentation can still teach us something. Readiness, to me, is the ability to make that decision with artists and affected communities in the room, then remain accountable for what follows.
As part of that work, I am building an independent public map of Indonesia’s AI and music-governance developments. It will separate official statements, stakeholder positions, research evidence, and my own analysis, including questions that remain unresolved. The dashboard will not speak for Konferensi Musik Indonesia / Indonesian Music Conference, Kementerian Kebudayaan Republik Indonesia / Ministry of Culture of the Republic of Indonesia, AXEAN Festival, or any participating institution unless a statement is explicitly attributed to them. It is a way to make the conversation easier to inspect before we start treating any answer as settled.
Without artistic agency, institutions become defensive and technology is defined elsewhere. Without institutional capacity, experimentation is easily absorbed by platforms with greater resources and bargaining power. The future of culture depends on building both together.
