Overview:
Ngan Vu, a research engineer at DeepMind, hosted a discussion on AI in Vietnam, focusing on inclusivity and the underrepresentation of Southeast Asian communities in the field. Jeff Dean, a Google Senior Fellow and leader of Google AI, joined the conversation to support these communities and engage in important discussions.

Ngan Vu’s Introduction:
Vu expressed her desire to make the field of AI more inclusive of Vietnam and Southeast Asia, recognizing the underrepresentation of these regions. She appreciated Jeff Dean’s willingness to support these communities and participate in the discussion.

Jeff Dean’s Background and Interest in Vietnam:
Dean has never visited Vietnam but has always wanted to. He enjoys Vietnamese food and has Vietnamese colleagues with whom he has collaborated on research papers. Dean appreciates the opportunity to work with people from diverse backgrounds and learn about different cultures through shared stories and anecdotes.

Jeff Dean’s Experience with AI and Technology in Vietnam:
Dean believes that Vietnamese people have a strong background in math and science, with a primary and secondary education system that prepares them well for careers in technology. He has seen Vietnamese AI researchers making significant contributions to the field, both in academia and industry.

Challenges and Opportunities in Vietnam’s AI Landscape:
Vietnam possesses a wealth of talented young people with great potential in AI and machine learning. The creation of the VinAI Research Institute, the first AI lab in Vietnam dedicated to fundamental research, aims to bridge the gap between potential and career development for young talents. The AI residency program at VinAI attracts top talents from local universities, providing them with exposure to high-quality research and hands-on experience. Vietnamese AI researchers exhibit remarkable dedication and hard work, sometimes surpassing that of their Silicon Valley counterparts. However, Vietnam faces challenges such as environmental issues, traffic congestion, and limited infrastructure, which hinder the development and deployment of AI technologies.

Addressing the Gap between Developed and Developing Worlds in AI:
Democratizing AI through open-source software libraries like TensorFlow enables broader participation in AI research and application worldwide. Collaboration between multinational technology companies and local research institutions can facilitate knowledge transfer and capacity building in developing countries. Supporting the development of local AI talent through education, training, and mentorship programs can help bridge the skills gap and foster a vibrant AI ecosystem. Encouraging international collaboration and partnerships among researchers, companies, and governments can promote the sharing of knowledge, resources, and best practices. Promoting responsible and ethical AI development and deployment can ensure that AI benefits all of society, especially in developing countries.

Recommendations for Closing the AI Development Gap:
Jeff Dean emphasizes the importance of democratizing AI through open-source software libraries and promoting collaboration between multinational companies and local research institutions. Hung Le suggests supporting the development of local AI talent through education, training, and mentorship programs, as well as encouraging international collaboration and partnerships. Ngan Vu highlights the need for promoting responsible and ethical AI development and deployment to ensure that AI benefits all of society, particularly in developing countries.

Education and Language Barriers:
Jeff Dean emphasizes the importance of education, particularly in math and science, as a foundation for AI and machine learning careers. Dean acknowledges the language barrier as a challenge, as most research papers are written in English. Hung Hai Bui suggests localizing teaching materials, such as subtitling video lectures and having Vietnamese-language instruction, to address the language barrier.

Fostering a Thriving AI Ecosystem:
Dean stresses the value of strong education at all levels, from secondary to graduate studies, to cultivate a thriving AI ecosystem. He highlights the importance of having local talent work on advanced problems and the potential for attracting those who have studied abroad to return to Vietnam. Bui emphasizes the need for fundamental research to build a strong research ecosystem and attract top talent.

AI’s Impact Beyond Computer Science:
Dean points out that AI and machine learning are not limited to computer science but are transforming numerous fields, including education, healthcare, robotics, climate science, and basic science. He encourages individuals interested in these fields to learn about AI and apply it to their areas of expertise.

Scaling Educational Initiatives:
Bui discusses the challenges of scaling educational initiatives, such as localized online courses, and explores the possibility of adopting platforms like Coursera or Eddex to reach a wider audience.

Balance between Fundamental and Applied Research:
Phong, a researcher from Vietnam, asks for advice on the ideal ratio between fundamental and applied research in Vietnam, considering that many companies cannot afford extensive fundamental research. Dean responds that any research endeavor should include some fundamental research to drive innovation and attract top talent. Bui adds that fundamental research is essential for solving long-term problems and building a strong foundation for applied research.

Portfolio Management of Research Projects:
Research investments should be managed as a portfolio, considering projects with varying time frames and probabilities of success. A mix of fundamental and applied research is desirable, with long-term investments targeting transformative advancements.

Balancing Fundamental and Applied Research:
Fundamental research, although challenging and uncertain, can lead to significant breakthroughs and advancements in the field. Applied research can uncover fundamental research problems that require further investigation. These two approaches complement each other, driving progress and innovation.

Recognizing Opportunities for Applied Research Impact:
Research groups should actively seek opportunities for their work to impact applied domains. Collaboration between research and applied teams can accelerate progress and solve real-world problems.

Feedback Loop between Research and Applied Domains:
Fundamental research insights can improve applied technologies and practices. Applied challenges can inspire fundamental research directions and agendas.

Harnessing AI for Emerging Markets:
AI has immense potential to address challenges and create opportunities in emerging markets like Vietnam. AI-driven solutions can improve healthcare, education, agriculture, and other vital sectors.

Collaboration and Partnerships:
International collaboration and partnerships are crucial for AI development in emerging markets. Sharing knowledge, resources, and expertise can accelerate progress and drive inclusive growth.

Ethical Considerations:
AI deployment in emerging markets should prioritize ethical considerations, ensuring responsible and beneficial outcomes. Local communities should be actively involved in shaping AI policies and applications to reflect their values and needs.

AI in Healthcare:
AI can enhance healthcare decision-making, especially in developing countries with limited healthcare infrastructure. Machine learning models can replicate expert medical opinions, providing effective decision-making tools for basic care. Example: Google’s work on diabetic retinopathy screening using AI models trained to match the expertise of retinal specialists.

Benefits of AI in Developing Markets:
AI can address global issues like climate prediction, weather forecasting, and educational tutoring. These applications have the potential to disproportionately benefit developing markets.

Google’s Potential Presence in Vietnam:
Google is constantly evaluating opportunities for expanding its global presence. Vietnam’s pool of talented developers is a potential factor in considering an engineering office in the country.

Expanding Google Research Offices:
Jeff Dean considers expanding Google Research offices to locations with strong backgrounds in math and science, including Vietnam. He mentions Vietnam’s relatively young engineering student population as a positive factor for potential office expansion. While there are no immediate plans, Vietnam is on the list of potential locations for future expansion.

Research During COVID-19:
Dean notes that the COVID-19 pandemic has forced companies and organizations to adapt to remote work. He highlights the effectiveness of virtual events like this one, despite not being physically present. Dean emphasizes that research is particularly suited for distributed work due to its flexible nature and the ability to communicate asynchronously.

Gender Diversity in AI:
Ngan Vu brings up the importance of gender diversity, equity, and inclusion in AI. She mentions Jeff Dean’s advocacy for gender diversity and his participation in the Grace Hopper Conference. Vu cites an article where a CTO in Vietnam made controversial statements about women’s technical abilities and persistence under pressure. She emphasizes the danger of reducing individuals to stereotypes and the need to address gender biases in the tech industry.

Jeff Dean’s Perspective on Diversity in Computing:
Jeff Dean, an expert in computer science and machine learning, emphasizes the importance of welcoming everyone, including women, into the field of computing. He advocates for diversity in gender, nationality, language ability, and other factors, believing that a diverse workforce leads to more innovative and impactful solutions. Dean’s personal experience of moving around a lot and living in various places shaped his belief that no one place or group of people is superior.

AI’s Impact on Employment:
AI and machine learning are advanced technologies that can solve problems more efficiently and with better outcomes in many cases. AI and machine learning are often tools that empower people to tackle more challenging problems, with computers handling repetitive tasks. Dean views AI and humans as partners, working together to enhance effectiveness and capabilities. AI will cause shifts in what people do in their jobs, automating certain tasks but also creating opportunities for new tasks and expanding job roles. Societies and governments need to consider how jobs will change due to AI and take steps to support workers affected by these shifts. For example, autonomous vehicles will impact jobs in transportation, and governments must plan for this transition.

Ethical Considerations in AI Development:
AI and machine learning technologies themselves are neutral, and their ethical implications depend on their application. Google employs a set of seven AI principles to evaluate potential AI uses within its products. These principles address issues such as privacy, human rights, bias, and the positive or negative impact on society. Google’s AI principles review process ensures that every machine learning-related launch or development undergoes ethical scrutiny. The company’s ethical framework has helped its engineers and researchers approach AI development with a consistent and responsible mindset.

Societal Shifts Due to Automation:
Automation, including self-driving vehicles, will lead to job shifts and the need for upskilling. Governments and society must ensure opportunities for affected workers to acquire new, in-demand skills. Historical examples, such as the transition from train workers to airplane pilots, demonstrate society’s ability to adapt to technological advancements.

Personal Development and Leadership Habits:
Jeff Dean emphasizes the importance of curiosity and learning as key habits for computer science leaders. Continuous learning and exploration of new concepts and technologies are essential for staying relevant and influential. Leaders should prioritize understanding the impact of their work on society and strive to make a positive contribution. Dean’s experiences at Google, including designing and building software systems at scale, have taught him the importance of collaboration, iteration, and focusing on the user’s needs.

Building a Diverse Toolkit:
Embrace new problems, even if they require skills you don’t have. Collaborate with people from different backgrounds to complement your skills. Engage in cross-disciplinary learning to expand your expertise.

Effective Learning Strategies:
Skim multiple papers to gain broad knowledge across different areas. Prioritize understanding concepts over mastering details. Read abstracts to gather a wide range of ideas and approaches.

Enjoying the Work:
Surround yourself with colleagues you enjoy working with. Find collaborators who are ambitious and fun to be around. Create a work environment that promotes enjoyment and productivity.

Collaboration and Partnerships:
Seek long-term collaborations with talented colleagues. Work closely with others to make numerous important decisions together. Pair programming and co-designing systems can foster innovation.

Early Career in AI:
Take machine learning classes to gain a solid foundation in the field. Focus on research that has practical applications and real-world impact. Pursue problems that align with your interests and passions.

Favorite Machine Learning Applications:
Jeff Dean expresses excitement about the transformative potential of machine learning in healthcare and climate science.

Other Technological Trends of Interest:
Dean mentions the advancements in genetics and reusable space vehicles as compelling areas of development. He also emphasizes his recent interest in urban planning, seeking ways to design livable and effective cities.

Motivations for Daily Work:
Dean’s drive stems from the ability to solve important problems through computer science, impacting millions of lives. He finds fulfillment in finding new and interesting problems that can improve people’s lives, whether through software development or product-oriented advancements.

Overcoming Learning Struggles:
Dean stresses the importance of selecting appropriate learning material suited to one’s level of understanding. Collaboration with experts in different fields is valuable for personalized tutoring and targeted knowledge acquisition. He highlights the effectiveness of personalized tutoring in both early education and specialized research topics.

Parting Thoughts:
Dean expresses gratitude for the opportunity to interact with people in Vietnam and apologizes for not being able to visit in person due to COVID-19. He acknowledges the contributions of the event’s participants and organizers and looks forward to visiting Vietnam in the future.