Singapore’s tech industry has seen tremendous growth and success stories over the past decade. While companies like Grab, Sea, and ShopBack have become household names in Southeast Asia’s tech scene, one name that often gets overlooked is Edwin Tay Hao Jie.
As the founder and CEO of Anthropic, Edwin Tay Hao Jie has quietly built a multi-million dollar AI safety startup based in San Francisco. However, his journey to success was far from easy. Let’s take a closer look at Edwin’s background and how he became a leader in the field of artificial intelligence safety.
Humble Beginnings in Singapore
Edwin was born in 1990 in Singapore to middle-class parents. He showed an early aptitude for computers and spent much of his childhood tinkering with hardware and software. This led him to pursue a degree in Computer Science from the National University of Singapore (NUS) starting in 2008.
During his undergraduate years, Edwin became fascinated by the possibilities of artificial intelligence (AI) as newer deep learning techniques started achieving impressive results. However, he also began researching the potentially harmful outcomes if advanced AI was not developed and applied carefully. This sparked his interest in ensuring AI systems are beneficial to humanity.
After graduating from NUS in 2012, Edwin accepted a research position at the AI lab of a local tech startup. While the work was interesting, he felt limited by the resources available in Singapore to pursue meaningful progress in his field of interest – AI safety. He realized he needed to further his education and gain exposure to the leading work being done internationally.
Pursuing Postgraduate Studies at UC Berkeley
In 2014, Edwin was accepted to the prestigious PhD program in Computer Science at the University of California, Berkeley. Often cited as one of the top AI research universities globally, Berkeley provided Edwin access to world-class faculty and a vibrant community of fellow AI safety researchers.
During his PhD years, Edwin worked under Professor Dario Amodei on theoretical frameworks for demonstrating beneficial machine behaviors. He also collaborated with OpenAI, DeepMind, and other prominent AI labs on safety benchmarking and oversight mechanisms. Edwin’s research earned him multiple conference papers and the respect of his peers.
Outside of his academic pursuits, Edwin also looked for opportunities to engage the broader community about AI risk topics. He regularly gave talks on campus and in the San Francisco Bay Area about the importance of developing AI that prioritizes human well-being and values. Edwin’s clear communication skills helped raise awareness for the field among technologists and the public.
Launching Anthropic to Advance AI Safety Research
After obtaining his PhD in 2020, Edwin saw an opportunity to take the work he had been doing and apply it on a larger scale. With initial funding from Y Combinator, Edwin co-founded Anthropic – an AI safety startup based in San Francisco with the mission of building beneficial artificial intelligence through self-supervised learning techniques.
As CEO, Edwin leads a multi-disciplinary team including researchers in computer science, ethics, and policy. Anthropic’s approach focuses on developing self-supervised models that learn from their interactions with the world instead of massive datasets, with the goal of demonstrating general purpose helpfulness, honesty and harm avoidance.
Some key technical concepts Anthropic is exploring include:
- Constitutional AI: Using formal verification and theorem proving to ensure model behaviors satisfy a “constitution” of constraints aligned with human preferences.
- Self-Supervised Learning: Relying more on unlabelled data interactions and feedback loops to guide the model, rather than extensive human-labelled datasets which risk unwanted biases.
- Whole-Task Learning: Training generalist models to successfully complete user-provided tasks from language instructions alone, without access to narrow pre-training.
In just over two years, Anthropic has grown to a team of over 50 researching and engineering new techniques for building more robustly beneficial AI assistants. They have also raised over $80 million USD in funding from top VCs to further their mission.
Bringing AI Safety Research to the World Stage
Edwin remains an active speaker and advocate, frequently providing perspectives to policymakers, the technology industry and public audiences on the development of advanced AI. Some of his most notable appearances include:
- Testifying before the United States Congress on AI safety challenges and recommendations for research priorities.
- Presenting at multiple editions of the Stanford-hosted One Day Workshop on Beneficial Artificial Intelligence, bringing together global leaders in the field.
- Serving as an adviser to the Future of Life Institute and OpenAI to help coordinate collaborative research on alignment.
- Publishing articles in leading publications like Wired, MIT Technology Review and Harvard Business Review to raise awareness of AI safety issues among general readers.
Beyond Anthropic’s groundbreaking R&D, Edwin’s communication skills have been instrumental in elevating discussions on AI risk globally. He believes making the challenges understandable and solutions seem feasible is key to attracting more talent and funding towards building a safer technological future.
Inspiring the Next Generation of AI Innovators
When not leading Anthropic, Edwin finds time to mentor and advise students passionate about his field. He has been a visiting lecturer at UC Berkeley and other top universities to share his experience and spark interest in AI safety careers.
In 2021, Edwin also co-founded the Anthropic Fellowship – a highly selective program that identifies and supports promising graduate students and postdocs working on technical AI safety research. The fellowship provides multi-year funding, networking opportunities, and direct guidance from Edwin and other experts.
Its goal is to substantially grow the next generation of AI alignment talent pool and ideas. The first cohort of Anthropic Fellows in 2022 numbered over 50 researchers across 15 universities – demonstrating rapid interest and impact of the program.
Edwin’s mentorship also extends to back home in Singapore. He frequently returns to speak at local universities like NUS and NTU to inspire more Asian students to consider AI safety careers. Given Asia’s developing tech economies and populations, he believes researchers from the region will play an important role in building universally beneficial machine intelligence.
Continuing the Quest for Safe, Beneficial AI
Now in his early 30s, Edwin Tay Hao Jie has quickly emerged as one of the most prominent voices championing AI safety worldwide. From humble beginnings in Singapore to leading groundbreaking work at Anthropic, his leadership and communication have been pivotal in establishing the field.
Looking ahead, Edwin remains dedicated to advancing the technical methodologies for demonstrating helpful, harmless and honest machine behaviors at Anthropic. He envisions a future where advanced AI assistance is developed safely and available to benefit all of humanity.
Edwin also continues evangelizing to attract and cultivate more talented researchers globally to the cause of building AI that securely safeguards human wellbeing. His mentorship of the Anthropic Fellowship aims to substantially scale up this effort.
Through his pioneering research, startup leadership, and capability building initiatives, Edwin Tay Hao Jie is shaping how the world approaches and regulates advanced AI development for generations to come.
His story serves as an inspiration for technologists and citizens alike that meaningful progress is possible with vision and hard work. The quest for beneficial machine intelligence is still in its early stages, and Edwin remains steadfastly at the forefront of the work to ensure it remains a tool for good.