Dr. Zhang Hongjiang’s Insights on AI and LLM Trends in Southeast Asia

Dr. Zhang Hongjiang, advisor to WIZ.AI and a renowned computer scientist, shares his perspectives on recent AI and Large Language Model (LLM) trends. This interview with WIZ.AI explores the transformative impact of foundation models like ChatGPT and GPT-4, examining their implications for AI and generative AI adoption primarily in Southeast Asia. We also delve into the challenges, risks, and opportunities that AI companies face, plus the crucial role of talent development and AI governance.

Key Takeaways

• ChatGPT’s adoption by millions of users marks a key milestone in AI history. In ten to fifty years, people will remember this year as when we saw and realized AI’s full potential.
• Foundation models will emerge from regions beyond the US, including Southeast Asia. Singapore will become a regional hub for new AI technology due to its dense AI talent pool and rapid technology adoption.
• Unlike previous AI waves, current foundation models require significantly larger resource pools for training.
• AI companies should invest resources in addressing AI parsing and ensuring safety from potential risks. Companies must tackle system failure and misuse to enhance security.
• While this AI wave will replace some jobs, it will also create new opportunities. Retraining people becomes critical. AI Singapore focuses on training non-technical staff in local traditional companies, equipping them with AI and prompt engineering skills.
• Currently, robots can only perform specific tasks. However, future robots equipped with LLMs and multimodality models will act autonomously based on observations.
• Technology-focused startups, not just big enterprises, could become foundation model providers in the future.

Large Language Model Trends and AI Advancements

WIZ.AI: What do you think are some large language model trends and AI advancements happening around us?

Dr. Zhang Hongjiang: The entire world has experienced rapid change in AI technology over the last twelve months. We witnessed the power of foundation models, particularly ChatGPT and GPT-4, changing how people view AI, its future, and how AI transforms our societies and business practices.

ChatGPT’s adoption by over 100 million users represents a genuine milestone in AI history. I believe that in ten, twenty, or fifty years, we will recognize this as the year we truly saw and realized AI’s full potential.

AI Trends Across Global Regions

WIZ.AI: AI advancements happen not just in the US but across Europe, Asia, Southeast Asia, and other parts of the world. What are your views on AI trend differences in these regions, especially in Southeast Asia?

Dr. Zhang Hongjiang: Foundation model development extends beyond the US to Europe, the UK, China, and Southeast Asia. Singapore has assembled excellent AI talent at the National University of Singapore and Nanyang Technological University.

The density of AI talent in Singapore impresses me, as does how quickly people learn and adopt new technology—from government to private enterprises, from big multinational corporations to small startups. The Singapore government’s speed and effectiveness in promoting AI technology surprises me.

Looking at potential societal impact, I see Singapore as a regional hub for new AI technology and innovative application scenarios. Singapore will continue playing a critical role in AI adoption throughout Southeast Asia. As a fast-growing economic region, Southeast Asia offers several opportunities to apply foundation models and fine-tune them for specific scenarios that fit various countries, industries, and commercial sectors.

Current AI Wave vs. Previous Advancements

WIZ.AI: How does the current AI wave differ from previous AI technology advancements, and what changes have foundation models brought during this time?

Dr. Zhang Hongjiang: The current AI wave differs from previous ones because foundation AI models require much larger resource pools than before. They need CPU resource cards and pre-trained models for performing alignment training. Future AI model training will continue requiring these resources.

However, some challenges remain consistent for Southeast Asia and other regions: finding resources, helping companies access them, and supporting startups. We need to invest significant effort in these areas.

Foundation models have brought unprecedented changes. Previously, developing a service or application scenario—such as a medical application to analyze X-ray pictures—required creating a vertical model for that specific scenario. Today, with large language models, especially multimodality models, you can fine-tune foundation models to align with particular scenarios.

Companies no longer need large AI teams. Using your data to fine-tune and align represents a paradigm shift in AI application development. Most applications no longer require developing new models from scratch. This accelerates AI application development and adoption, helping improve productivity and fostering new AI-powered apps for consumer services.

Addressing Southeast Asia’s Linguistic and Cultural Diversity

WIZ.AI: How do you think AI-based companies like WIZ.AI can strategically address the linguistic and cultural diversity in the SEA region by fine-tuning and aligning foundation models?

Dr. Zhang Hongjiang: Southeast Asia’s demographic diversity—with multiple languages, cultures, and traditions—makes aligning and fine-tuning models to fit local cultures critical. Companies like WIZ.AI have opportunities to adopt AI for customer services, corporate services, and more.

However, applications extend beyond customer service and can help improve company productivity for consumers and customers. WIZ.AI, as an AI service provider, can provide tremendous value to local companies. Challenges exist in preserving data privacy and security. New algorithms and systems can protect data privacy and integrity while improving security.

Improving AI Security

WIZ.AI: What suggestions do you have for generative AI solution providers like WIZ.AI to improve AI security?

Dr. Zhang Hongjiang: AI technology and application companies should spend significant portions of their resources and efforts addressing AI parsing. This resembles working with nuclear energy and nuclear power stations, where large R&D efforts focus on safety.

Previously, AI’s power was too limited to cause harm. Today, we have reached that critical point. You can categorize risks into different types. System failure represents one category—a persistent topic since IT technology began. System failures impact governments, students, and everyone. AI and LLMs face significant exposure to this risk.

Misuse forms the next category. When people ignore regulatory rules and principles, misuse occurs. Rogue AI use designed for evil purposes also falls under misuse. We need governance and regulations for AI and generative AI. Those considering technical aspects should understand the risks while knowing how to discover and handle them from a technological perspective.

Job Replacement and Opportunities

WIZ.AI: Will this AI wave replace jobs? What are your thoughts?

Dr. Zhang Hongjiang: Every new technology brings significant changes to our society and industry, meaning some jobs will face replacement. Hopefully, it will also create new job opportunities, allowing people to retrain for these new positions. However, this represents a much larger issue than technology alone.

I recently spoke with individuals from the AI Singapore program. One of their activities trains non-AI technical staff from local traditional companies. Training programs equip them with AI and prompt engineering skills and solid understanding of LLMs. These initiatives help those companies adopt AI.

Every Southeast Asian company, especially traditional companies, could benefit from such programs. The regional economy can benefit by training and retraining employees from these companies with AI skills. While AI represents advanced technology that can be difficult to grasp, foundation models have made this technology easier to use, lowering barriers to entry.

The Path to Artificial General Intelligence (AGI)

WIZ.AI: How far do you think we are from the era of AGI and what is your greatest hope for AGI?

Dr. Zhang Hongjiang: We have seen sparks of AGI from GPT-4 and newly released multimodality models. I believe they will change the industry and how we develop software and future models. Co-pilots that help with coding and design will accelerate development processes.

Today, LLMs handle most native language processing problems. I also see multimodality models, combined with vision and language, revolutionizing the software and robotic industries. Currently, many robots can only perform one task—they are hard-wired and precoded.

However, once we equip robots with LLMs and multimodality models, we give them real brains. Robots will act autonomously based on observations and your requirements and requests. This shift will transform robots into autonomous self-planning and self-action entities.

Of course, this poses potential risks. We have moved from a traditional information age to an intelligence age. We will transition into an age of autonomous agents, with many software agents helping us in various ways. They will work as co-pilots or autopilots, requiring developments with significant focus on safety and risk management.

Future of Foundation Models Providers

WIZ.AI: Do you think big players will continue as the primary foundation model providers in the future?

Dr. Zhang Hongjiang: Not necessarily. OpenAI was a small startup and nonprofit two to three years ago, yet now they represent the most important player in the AI and generative AI space. I don’t believe only big players will have opportunities to become foundation model providers.

Technology-focused startups also have potential. Several exist, and some will surely succeed. Hopefully, we can see one emerge from Southeast Asia.

About Dr. Zhang Hongjiang

Zhang Hongjiang is a Chinese computer scientist and executive who serves as advisor to WIZ.AI. He has served as Kingsoft’s CEO, Microsoft’s managing director of Advanced Technology Center (ATC), and Chief Technology Officer (CTO) of Microsoft China Research and Development Group (CRD). The National Academy of Engineering elected him in 2022 for his technical contributions and leadership in multimedia computing.

Ready to Transform Your Business with AI?

Discover how WIZ.AI can help your organization harness the power of foundation models and generative AI.