Singapore vs "AIers"

14 May 2024

The Singaporean government's request to use local literary works for training a large language model (LLM) sparked a backlash from prominent writers due to concerns over consultation, compensation, and copyright protection. This reaction mirrors global resistance against using copyrighted content in AI training. While Singapore's approach to seeking consent is considered considerate, historical tensions between the government and the arts community fuel skepticism. Balancing technological progress with the rights of creators is crucial for Singapore's ambition to lead in AI by 2030.

When the Singaporean government sought permission from local writers to use their works for training a large language model (LLM), it likely did not anticipate the strong backlash from the country’s small literary community. In March, an email from the National Multimodal LLM Programme (NMLP) highlighted the need to counter the Western bias in existing LLMs by creating one rooted in Singapore’s unique linguistic and cultural context, a context that is rich and diverse, encompassing local languages like Malay, Mandarin, and Tamil.

However, prominent literary figures like Gwee Li Sui, author of the first book written in Singlish, voiced significant concerns. Gwee criticized the need for early-stage consultation with writers and the absence of precise compensation and copyright protection terms. The government's initial communication provided only a ten-day window for response and scant details on these critical issues, leading Gwee and others to refuse participation.

This reaction is not isolated, but part of a global trend of resistance against the use of copyrighted works in AI training. High-profile lawsuits in the U.S. by authors like Sarah Silverman and John Grisham against companies such as OpenAI and Meta reflect similar concerns over copyright infringement, underscoring the gravity of the issue. 
While some countries are developing their own LLMs in Asia, legal protections and compensation for authors must be clarified.

Despite these challenges, the Singaporean government's attempt to seek consent is noteworthy. Nuurrianti Jalli, an assistant professor at Oklahoma State University, highlighted that most LLM developers do not seek permission from copyright holders, making Singapore's approach relatively considerate. Yet, writers demand specific guarantees and protections.

Historically, Singapore's relationship with its arts community has been fraught with censorship and stringent regulations, exacerbating writers' skepticism about the NMLP. However, as Singapore strives to become an AI leader by 2030, balancing technological advancement with the rights and concerns of its creative community will be crucial. Engaging writers constructively could not only address their concerns but also enhance the quality and cultural relevance of the LLM, fostering public trust and collaboration in this ambitious project.
Thanks for reading. You can support and reward me via: 
Pay Pal —
ADA — addr1q9vfs6nqz4xmtnpljwhv4tukyskd2g7enxd87rpugkwwvfun5pnla5d5tes2mvurrc77e7837yd0scrfk063qlha8wgs8d4ynz
Bitcoin 3HbxyDXE9MhNQ8RqsirqgYvFupQzh5Xby2
ETH — 0x8982cdb97bd23f092f78a16a4fc93c5c4607a285
Seeds — vladlausevic
Skycoin — ZxjhWMJRbTNCRQzy5MekZzH4fhdWFCqBP8
Tezos — tz1QrRzkTAKuPKF8dmGW6c1ScEHBUGvoiJBM

Write & Read to Earn with BULB

Learn More

Enjoy this blog? Subscribe to Vladlau89


No comments yet.
Most relevant comments are displayed, so some may have been filtered out.