30-second overview: Born in 1976, Ethan Tu (杜奕瑾) set up PTT, Taiwan's largest BBS forum, with a single 486 computer in a university dormitory in 1995. That text-based forum has survived the dot-com bubble, the rise of Facebook, and coordinated disinformation attacks—thirty years later it remains a central node of Taiwan's public discourse. He traveled to the United States to pursue genetic research at the NIH, then joined Microsoft, where he led the development of Cortana across multiple platforms as a Principal Development Manager. In March 2017, he abandoned his Seattle salary to return to Taiwan and establish Taiwan AI Labs, Asia's first nonprofit AI research organization. This is the story of a person who chose not to become Taiwan's Mark Zuckerberg, but instead to place the most advanced technology in places where it does not generate profit.
On the night of September 14, 1995, in a dormitory at National Taiwan University, a second-year student connected an aging 486 computer from Guanghua shopping mall to the network, ran a Linux command, and named the site after his handle—"Ptt." His classmates called him "Panda" for the dark circles under his eyes (from all-nighters), and his surname was "Tu (杜)." "PT" felt too short, so he added another "T" to make it flow better when spoken aloud.
No journalists were present. No press conference. No investors. PTT simply came online.
Free Expression by Design, Not Retrofit
Before building PTT, Tu was already the administrator of the campus BBS at National Taiwan University, Coconut Grove. The problem was the university exercised tight control. When a student posted using a handle to seek a one-night stand, the administration simply disabled anonymity. Tu disagreed with this logic. He and a few classmates decided to start over, building PTT independent of official oversight.
"PTT has free expression and equal voice distribution—it doesn't use algorithms to push advertisements. The fact that it has lasted twenty-five years is already a kind of success; very few comparable cases exist worldwide," he said in a 2020 interview.
PTT's system of upvoting and downvoting, its democratic mechanism for electing moderators by user vote, its non-commercial financial structure—all of these were written into the system's DNA from the beginning. This decision sent it down a completely different path from every commercial social platform.
A Forum That Appears Again and Again at Critical Moments
At peak, PTT saw concurrent users exceeding 150,000. That number alone doesn't explain why it matters, but a few dates might:
- 2013 Hung Chung-chiu incident: When an army sergeant died from abuse, the news fermented on PTT. Netizens spontaneously formed a civic group, "Citizens 1985 Action Alliance," which led to 250,000 people marching across Taiwan—one of the largest grassroots demonstrations since the 1990s.
- 2014 Sunflower Movement: Students occupying the Legislative Yuan to protest the hastily-passed trade agreement used PTT as a command center, and the platform became a thermometer for measuring public sentiment.
- Early 2020 COVID-19 outbreak: A warning from Dr. Li Wenliang on Weibo was reposted to PTT, enabling the Taiwan CDC to grasp the severity by late January—earlier than most governments worldwide.
Taken together, these three moments show that PTT is not just a place for online chatter, but a crucial piece of infrastructure with a unique role in Taiwan's civil society.
📝 PTT's Anti-Commercial Structure
PTT does not sell advertisements, go public, or accept corporate acquisition. Its servers depend on the bandwidth of National Taiwan University's academic network and are managed by a student-run society, "Department of Electronic Bulletin Board System Research." Its legal counsel is a volunteer attorney. This structure leaves it free from commercial pressure to monetize or from advertising algorithms—but equally, it is always starved for resources.
To America and Back Again
In 2003, Tu decided to "vote with his feet" and leave Taiwan.
The reason was straightforward: Taiwan at that time did not value software. He had long wanted to build instant messaging as a cross-device service, but was told that only telecommunications carriers could do so. He created an early digital currency called "P-coin," but was shut down for violating financial regulations. "People in Taiwan had already explored digital economics, but it was strangled. The industry looked down on it too. Back then, everyone thought online chatting was a dead-end career," he said, a note of long-suppressed disappointment in his voice.
Later, a small Chinese company partnered with Yam and built an instant-messaging app that resembled his original vision. That company was Tencent; the app was QQ.
Arriving in the United States, Tu first entered the National Institutes of Health's Human Genome Research Institute, where he worked on automated detection of genetic sequences and cancer. In 2006 he moved to Microsoft, starting in storage technology, then joining the Bing search engine team led by Dr. Harry Shum (沈向洋).
In 2012, he entered Microsoft's AI division with the title Principal Development Manager, responsible for the Cortana product. In 2015, his Cortana team released the world's first cross-platform AI voice assistant—not just on Windows, but on Android and iOS. In 2016, Microsoft established a 6,000-person AI initiatives unit (AI.R.), and he took charge of Cortana's cross-platform development and Microsoft's AI strategy for the Asia-Pacific region.
📝 Cortana's Significance: The AI Assistant Battle of 2016
2016 was the most intense year on the voice-AI battlefield: Microsoft launched cross-platform Cortana, Google unveiled Allo at I/O, Facebook rolled out Messenger Bot. All three shipped in the same year—in essence, three companies betting whether AI assistants would be the next operating system. Tu stood on Microsoft's side, leading dozens of engineers against the world's largest tech companies in pursuit of the same future.
March 2017: Abandon Seattle, Return to Taiwan to Build Nonprofit AI
His mother died of sepsis in 2013. He wrote on Facebook: "I held her hand as it went from warm to swollen, then ice-cold, through agony. Before her heartbeat stopped, I touched her head and said: 'Yi-chin is here to see you.'" Four years later, he formally returned to Taiwan.
In March 2017, Tu founded Taiwan AI Labs, established as a non-governmental, nonprofit organization and Asia's first open AI research institute. His reasoning: Taiwan has the world's most comprehensive national health insurance database and world-class software engineers, yet because of its market size, it sits at the margins of AI discourse and decision-making.
"Taiwan nurtures international-caliber professors and elite software talent. I plan to assemble Taiwan AI Labs to work concretely with leading Taiwan enterprises on AI experiments," he said in his founding declaration, his tone more energetic than his usual reserved public demeanor.
Taiwan AI Labs' work divides into several pillars:
Healthcare imaging: Partnering with major medical centers, the lab trains AI models on de-identified health insurance data and has developed TaimedImg, a medical imaging assistance platform for tasks like detecting lung X-ray abnormalities and auto-segmenting brain tumors, deployed in clinics and hospitals.
Human-computer interaction: Developing Yating (雅婷逐字稿), a speech-to-text tool tuned for Taiwan accents (including Taiwanese, Hakka, and Taiwan Mandarin), widely used by media reporters and for hearing-impairment assistance. In 2021, the lab launched FedGPT, a large-language-model-powered system letting enterprises build custom knowledge bases while protecting data privacy.
Countering information warfare: Established the Infodemic project, which uses AI to analyze abnormal coordinated behavior on social media, exposing bot networks and narrative strategies of disinformation campaigns.
COVID-19: An App and a Privacy Debate
In 2021, Taiwan AI Labs assisted the government in developing the "Taiwan Social Distancing App." This application uses Bluetooth-based contact detection, collects no personal identity data, and uploads no location information—one of the few systems globally that achieves contact tracing without building a centralized database.
This design embodied Tu's long-standing position: technology should serve the public and must not, in the name of public health, expand surveillance capacity. Critics noted, however, that this system's effectiveness depends on adoption rates. If too few people download it, tracing is limited; if government mandates it, voluntary adoption comes into question.
📝 Open-Source AI as Political Choice
In the race for large language models, Taiwan AI Labs chose to advance open, Traditional Chinese-language models rather than pursue commercial licensing. The logic is clear: if Taiwan's AI systems depend entirely on OpenAI or Google's models, Taiwan has no voice in how that AI sees and interprets the world. Tu's argument is that an open-source model trained on Taiwan-specific data and grounded in democratic values ensures Taiwan is not sidelined. But others question whether open-source models can compete commercially well enough to sustain ongoing research funding.
Critical Voices
Tu's public image is not without controversy, and these controversies deserve to be taken seriously.
Political alignment concerns: In 2021, when writer Lin Wei-feng was caught using reversal tactics on PTT, some commentators suggested this revealed connections between PTT's management and the ruling party. Hsieh Han-bing criticized Tu on CTi News, saying "this entire industry chain was set up by them." Tu sued; in 2022 the Taipei District Court ruled against him, finding that Hsieh's criticism was "not substantially divergent from the main facts."
Government contract dependency: Wikipedia notes that between 2018 and 2021, eight companies newly founded by Tu won 22 government contracts totaling approximately NT$198.12 million. For someone who claims his organization is non-governmental and nonprofit, this figure prompted observers to question its independence.
Internet army remarks: In July 2021, in an interview with United Daily News, Tu stated: "Internet armies have never been inherently evil. When a group of people online share a common vision, they might be called an 'internet army,' and the question is whether they use reasonable means to achieve their goals." This sparked widespread backlash; critics argued that the founder of an organization researching cognitive warfare and disinformation should not legitimize internet armies this way.
Digital Intermediary Services Act: In August 2022, Tu publicly supported the "Digital Intermediary Services Act," originally proposed by the NCC but widely criticized for overly broad language. This provoked strong pushback from PTT users. A founder of PTT—criticized by the very community that created it for standing on the side of "speech regulation"—this scene itself may be the most complex footnote to his thirty years.
He Says He Doesn't Care About Success
In a 2020 Mirror Media interview, a reporter asked him: "You became an AI team lead for Asia-Pacific at Microsoft, with a spouse and two daughters. Is returning to Taiwan to work in AI moving upward or downward?"
The moment he heard the word "success," he objected: "What does success even mean? Success is only a definition."
Then he said something that seemed to explain all thirty years of his choices: "PTT never wanted to become Facebook or sell ads. People generally think success means looking at market value, going public—I don't think that way."
If there is a thread through Tu's story, it may be this: each time he reaches the top of one position, he decides to pursue something far more uncertain. PTT was like that. Leaving Taiwan was like that. Coming back was like that.
Controversy makes his image more complicated, but also more authentic. A person who never turned PTT into a commodity, and a person who accepted nearly NT$200 million in government contracts—they are the same person. The guardian of public discourse and the backer of digital regulations—also the same person.
Contradiction is not his weakness. Contradiction may be his whole truth.
References
- Mirror Media — The Whole Picture: Loving Freedom Over Success, Ethan Tu (2020)（secondary source, celebrity interview）
- Wikipedia — Ethan Tu（secondary compilation with original references）
- Wikipedia — PTT（PTT founding date, historical timeline）
- Taiwan AI Labs Official Website（primary source, organization overview and research directions）
- Taiwan AI Labs — Healthcare（primary source, TaimedImg medical imaging platform）
- Taiwan AI Labs — Human Interaction（primary source, Yating, FedGPT, Infodemic projects）
Further Reading
Related topics: Audrey Tang, Sunflower Movement, Miin Music App, Taiwan's internet history, Taiwan Semiconductor Industry