In a groundbreaking development, the Southeast Asian Large Language Models (SeaLLMs) have emerged as a new frontier in artificial intelligence, designed specifically to cater to the diverse linguistic landscape of Southeast Asia. Developed by Alibaba DAMO Academy in partnership with researchers from Nanyang Technological University in Singapore, SeaLLMs supports nine regional languages including Thai, Vietnamese, and Indonesian, alongside English and Chinese.
Despite the advancements made by current large language models, many still harbor significant language biases. SeaLLMs addresses these issues, striving to ensure a more equitable AI landscape that enhances access to advanced technologies across Southeast Asia. This initiative is particularly important given the region's rich cultural diversity, allowing local businesses and communities to leverage advanced AI technologies without language barriers that have posed challenges in the past.
One of the standout features of SeaLLMs is its multilingual support, particularly evident in its SeaLLM-base version, which has been trained on a diverse and high-quality dataset. Linguists fluent in the respective languages verified its capabilities, ensuring that the model comprehensively understands local contexts and communication nuances. The model's effectiveness is showcased in its chatbots, designed to resonate with cultural contexts, thereby enhancing user interaction while observing local societal norms and styles of communication.
Offering both 7 billion and 13 billion parameter versions, SeaLLMs has made its models available as open-source on Hugging Face, opening avenues for research and commercial applications. The latest SeaLLM-7B-v2.5 model excels in tasks such as general knowledge, logical reasoning, and natural language processing, often outperforming previous models like GPT-3.5 in tests relevant to Southeast Asian languages. This achievement marks a significant step in standard-setting for AI capabilities in the region.
Moreover, SeaLLMs also embraces multimodal capabilities with SeaLLMs-Audio, which processes audio data, recognizing speech from multiple languages in the region, as well as English and Chinese. This feature enhances its utility in various applications, expanding the potential reach and versatility of AI in Southeast Asia.
In other exciting news from the art world, Christie’s recently held a historic auction showcasing artworks created by artificial intelligence (AI), which generated a total of $1.1 million USD, marking a significant moment in contemporary art history. The auction featured 34 pieces, including physical artworks and digital art sold as NFTs (non-fungible tokens). This auction, branded as “Augmented Intelligence,” highlighted the growing intersection of technology and art, fostering discussions on the evolving role of AI as a creator.
Dr. Jasmin Pfefferkorn, a researcher from Melbourne University, expressed enthusiasm for the potential integration of AI within the art sphere. “If we want in the area of studies, I think everything will get involved in the art,” she noted, suggesting that infused creativity may span across disciplines due to AI’s expanding capabilities. The auction successfully sold six artworks, though it faced backlash; over 6,500 artists signed a petition opposing the integration of AI in auctions.
Despite the protests, many artists recognize the nuanced relationship between AI and human creativity. Multi-disciplinary artist Sue Beyer commented on her mixed feelings, stating, "I like using AI because it’s not perfect, and that’s what I love," indicating a collaborative view between human intuition and AI ability. Beyer's approach highlights how artists can utilize AI to explore self-identity and to initiate deep personal questions about their humanity.
The rise of AI in the art world also comes with specific challenges, particularly reflected in the Brisbane Portrait Prize's recent policy shift. After contentious debates, the award adjusted eligibility criteria to allow AI to contribute but only as part of the creative process instead of the primary medium. Artist Dennis McCarthy, who received the Digital Award, illustrated how much traditional and AI methods can coexist in creating meaningful art.
The sentiment within the indigenous art community remains cautiously critical, particularly concerning how AI applications could misappropriate Indigenous cultures. Birrunga Wiradjuri, an award-winning artist and founder of Birrunga Gallery, voiced concerns over the commodification of First Nations art, likening it to modern-day cultural colonization. “I believe art should open space for debate and challenge traditional ideas,” he asserted, pressing that artists’ identities must not be compromised amid technological advancements.
With the advent of AI, issues surrounding copyright and cultural representation have surfaced prominently. Dr. Huw McKinnon from the Arts Law Centre Australia provided legal insight, informing artists about how their creations can be used and recognizing the potential for misrepresentation. As AI-generated artworks can sometimes eclipse original cultural expressions, protecting artists’ rights and intentions becomes paramount in this evolving landscape.
Moreover, the collaboration within the arts community to combat misrepresentation and inappropriate use of indigenous styles is evident. Tools like Nightshade and Glaze, developed by institutions like the University of Chicago, aim to prevent AI from training on unauthorized artwork, ensuring that artists maintain control over their intellectual property.
Dr. Pfefferkorn emphasized that while AI presents a new frontier, the dialogue surrounding its implications in art is essential. She mentioned, “I hope that if we engage and question these issues more, we can push back and reclaim our power,” showcasing the ongoing struggle artists face in an AI-enthused society.
As the intersection of art and technology continues to expand, the challenges and opportunities ahead signal a transformative chapter for both creators and consumers. It is imperative for artists, technologists, and advocates to navigate this new terrain collaboratively, ensuring the integrity of cultural expressions are respected as AI continues to evolve.
