AI 2040 Asia

Building a Plan A for AI in Asia

Inspired by Europe 2031 and AI 2040, AI 2040 Asia seeks to inform governance and industry in Asia, as well as global AI forecasting efforts.

Pilot: Philippines and Vietnam

Hover over a country to preview · click to open its cardTap a country to open its card

Why Asia

About sixty percent of the world’s population reside in Asia. Thus, any plan for AI going well cannot succeed without Asia. Despite this, ideas on how Asia might comply, resist, or even bargain with their positions are still limited and left to assumptions.

Why AI 2040

Global forecasting projects have a good track record of being cited in policy because they convert abstract risks into questions and scenarios governments can scrutinize and deliberate on. We chose AI 2040’s model because it offers a unique perspective – it is currently the most comprehensive and “optimistic” public plan for humanity being able to live alongside Artificial General Intelligence and Artificial Superintelligence.

An Asian insider perspective

Our project’s contribution is to provide an Asian insider perspective on these scenarios. For instance, how plausible is buy-in for the Philippines and Vietnam? Or maybe, rather than Mongolia, it is more likely that the Philippines will host compute for Mutually Assured Compute Destruction (MACD)?

The gap

AI 2040 was primarily informed by the developments within the frontier, particularly US frontier labs and policy communities. As a result, parts of Plan A, such as worldwide buy-in, verification, compute allocation, and MACD, assume Asia’s ready participation, which might not be the case.

Asia’s leverage

ASEAN alone is a major export market for both the US and China and a key security partner for the US. Compute chokepoints and chip smuggling routes run across Asia.

2027–2040 · Discussion framework

AI 2040 covers 14 years, from 2027 to 2040. However, in acknowledgment that positions of ROW countries would not qualitatively change year over year, we have decided to structure this exploration in the three stages we find most meaningful for consumer states in Asia.

2027–2030 · Stage 1 · Choosing verified slowdown

The negotiation period between US and China where ROW also plays a major part. For non-G2 states, this is the period of bargaining – leverage here is either lost permanently or exercised.

In AI 2040 · scroll or pick a year

2031–2035 · Stage 2 · Enforcing the deal and the fallout alternative

Unsurprisingly, the existence of a US-China AI treaty does not wave away the reality of uneven distribution of risks and benefits. Compute-robot allocation and Mutually Assured Compute Destruction (MACD) are also within this stage, both of which involve ROW countries, specifically key Asian countries like Mongolia in the latter.

In AI 2040 · scroll or pick a year

2036–2040 · Stage 3 · Post-AGI world

In this period, labor ends as a basis for income and we hand control over to AI. The question here for Global South ROW countries would be: in the past decade, have they built enough resilience for their citizens to survive in a world without jobs?

In AI 2040 · scroll or pick a year

The pilot

The first batch is two pilot expert-led working sessions held in Manila, Philippines and Ho Chi Minh City, Vietnam to discuss AI 2040’s Plan A and its implications for the two countries and Asia as a whole.

Next batches

We have already scouted interested country leads and jumped on a call already with some who would like to continue this pilot in their own countries.

Project leads

Advisors

Organisations

Safe AI Philippines AI Safety Vietnam (antoan.ai)

Methods and framework

Why local experts

Mid- to long-term scenario work for Asia is necessary to envision the region’s survival but the authors of AI 2040 aren’t well-placed to engage in this type of forecasting nor do they have enough local legitimacy to get actors in Asia moving.

There are two main reasons why we think the best way to bridge this gap is to involve locally respected experts: first, they are well-placed to do regional forecasting work and can verify the claims of AI 2040 in their own fields of expertise; and second, their participation lends the findings the type of local credibility and legitimacy that local governments and regional formations actually take seriously.

Why roundtables

We chose structured expert sessions and roundtables because these methods are proven to build credible forecasting projects, as is the case with AI 2027, which involved hundreds of participants across more than a dozen tabletop exercises. The shared expertise of everyone in the room allows for stress-testing of assumptions, co-creation of ideas, and a sense of ownership toward the project that makes participants more likely to cite it in conversations with their network.

The method, country by country

Each country lead will go through the following steps to standardize the method across multiple contexts and proceedings.

  1. Discussion guide draft and local expert review. The country lead will draft a discussion guide, complete with necessary local context, a hypothesis for each of the three stages in the discussion framework, and conclusions. Afterwards, they ask respected local experts in economics, geopolitics, labor, and law to read the AI 2040 document and review the draft. Comments to the draft will be resolved before the roundtable proceedings.
  2. Roundtable. The country leads will organize a roundtable of around 30–50 participants and invite leaders and representatives from a variety of sectors, from academia, civil society, and government. The experts invited to comment on the draft will also be invited to speak for about 15–20 minutes on their specific subject matter and how it relates to the draft. Then, there will be an open forum and FGDs following the discussion guide so that we can get a multitude of opinions on the topic. Volunteers will be present to transcribe the proceedings.
  3. Country-level analysis and report. The country leads will then compile the data gathered from the roundtable and draft the final report. While the local experts can be invited to be co-authors, this will not be the default case. We will, however, ask them to review the draft of the report and note their feedback. AI governance experts will also be consulted to provide feedback on the draft. After consolidation of all comments, the final report is published on this website.
  4. Asia summary report (future). After interested countries have held their roundtables and published their findings on this website, a summary report will be co-authored by the country leads to synthesize the findings and recommendations for Asia. The maximum goal is to become a replicable starting point for other Asian countries so that a summary report for Asia can be ideally produced by the end of 2027 to better inform global AI forecasting.

What the discussions ask

From the Philippine discussion guide:

  • Stage 1. What, if any, can the Philippines demand as a condition to join the Consortium? Can ASEAN negotiate as a regional bloc? Or will states have to negotiate individually? Would regional negotiation be preferential to individual negotiation? If so, what would be the benefits and tradeoffs?
  • Stage 2. How should the Philippines exercise and develop its leverage on the semiconductor supply chain to maximize robots and compute in the 2032 allocation? If the Philippines were asked to host US data centers for MACD, what would be the likely response?
  • Stage 3. How would the Philippines survive the end of labor given that labor is the country’s primary export? What sustainably and substantially replaces OFW remittances and BPO industries?

Measuring success

We will consider this pilot successful if six local experts review the draft and speak at the sessions, each session draws the estimated 30–50 participants, both country reports are published, and each report is submitted to at least one policy channel.

In the Philippines, we will cite the report in upcoming technical working group (TWG) hearings on the AI Development and Regulation Act and share it with the newly established National AI Council (NAIC). In Vietnam, we will submit the report to the Ministry of Science and Technology’s future public consultations on decrees for the AI Law.

Roundtables

  • Ho Chi Minh City Date to be announced

    Plan A for 2040: what should Vietnam aim for?

    An expert-led working session of about 30–50 people, with talks by local experts and an open forum and focus group discussions following the discussion guide.

  • Manila Date to be announced

    Plan A for 2040: what should the Philippines aim for?

    An expert-led working session of about 30–50 people, with talks by local experts and an open forum and focus group discussions following the discussion guide.

Country reports will be published here after the roundtables. Announcements of roundtables in other countries will also be posted here.

Interested in contributing?

For the pilot in Manila and Ho Chi Minh City, we are looking for:

  • Roundtable panelists and participants: leaders and representatives from academia, civil society, government, and industry.
  • Local experts in the Philippines and Vietnam in law, geopolitics, labor, and economy. For discussion draft review and feedback, a 15–20 minute presentation at the roundtable, and final report review and feedback.
  • AI governance experts, for final report review and feedback.

Beyond the pilot

Plan A for Asia is meant to grow country by country. If you are in another Asian country, you can lead or support a roundtable there, share a discussion guide your group has already drafted, or offer your expertise to one of the country leads on the map. Tell us your country in the form and we will connect you.

Funding

We are currently fundraising. If you are interested in funding Plan A for Asia, get in touch at team@ai2040.asia.

Agreeing to be a partner expert in this project does not mean you necessarily agree with the events and recommendations we describe in both the discussion guide and final report. Any mistakes are our own. Experts will be acknowledged in the final report with gratitude as contributors.

Or email team@ai2040.asia

I’m interested in