Convergence Analysis

Designing a safe and flourishing future for humanity in a world with transformative AI

[Convergence Analysis @CvgAnalysis]

Hey! We're Convergence Analysis - a new AI scenario and governance research institute.

Our mission: designing a safe and flourishing future for humanity in a world with transformative AI.

Read more:
https://t.co/g3QMLlshwi

[Convergence Analysis @CvgAnalysis]

Do you believe that red lines for AI are needed? Retweet this thread if you agree.
Learn about the Global Call: https://t.co/yhYR2hNHPF

[Convergence Analysis @CvgAnalysis]

Just Launched: The Global Call for AI Red Lines, with 70+ leading orgs & 200+ signatories, including former heads of state & Nobel laureates.

Together, we are calling for international agreement to prevent the most severe risks to humanity and global stability #AIRedLines

[Convergence Analysis @CvgAnalysis]

Signatories include Mary Robinson, Juan Manuel Santos, Joseph Stiglitz, Maria Ressa, Daron Acemoglu, Enrico Letta, Yoshua Bengio, Geoffrey Hinton, Andrew Chi-Chih Yao, Csaba Kőrösi, Ian Goodfellow, Stephen Fry, Yuval Noah Harari, and other respected thinkers and leaders.

[Convergence Analysis @CvgAnalysis]

35+ Organizations & Individuals in Whistleblowing and AI Focus Support This Call.

Only OpenAI has published their whistleblowing policies after public pressure in 2024. AI leaders shouldn't wait for pressure or worse; scandals, to drive transparency.

https://t.co/daen54FuOl

[Convergence Analysis @CvgAnalysis]

We’ve published a new post today on research we’re calling soft nationalization!

In short, we’re describing how the US government will eventually control AI labs due to national security concerns.

Want to see the full post? Click here:
https://t.co/FaqTsgC9uB

We believe existing descriptions of “nationalization” along the lines of a new Manhattan Project are unrealistic and reductive. The state of the AI industry — with trillions in private funding, millions of participants, and pervasive economic impacts — looks unlike nuclear research, or any previously nationalized industry. Government consolidation of frontier AI development is legally, politically, and practically infeasible.

We expect that AI nationalization won't look like a consolidated government-led “Project”, but rather like an evolving integration between the US government and frontier AI labs. The US government can select from many different policy levers to gain influence over these labs, and will progressively pull these levers as circumstances (particularly around national security) seem to demand it.

We believe the US government will achieve its goals in nearly all circumstances by combining sets of these policy levers, and will only turn to complete “nationalization” as a last resort.

Instead, government control of AI labs will incrementally escalate as concerns over national security grow.  The boundary between "regulation" and "nationalization" will become hazy.

We’re calling this process of increasing government control over frontier AI labs via iterative policy levers soft nationalization.

[Convergence Analysis @CvgAnalysis]

We just launched the 2024 State of the AI Regulatory Landscape!

https://t.co/bfqoFixQVf

It's a primer for individuals seeking to develop a high-level overview of the current state of AI regulation across the US, EU, and China.

In it, we examine a sequence of key topics on AI governance, summarizing legislative text and providing a short analysis for each topic.

Topics include:
- Evals
- Model Registries
- Incident Reporting
- Open-Source AI Models
- Cybersecurity
- Discrimination Requirements
- Disclosures
- CBRN Hazards and AI

We'll be discussing some key takeaways in the next few weeks from this report - stay tuned!

[Convergence Analysis @CvgAnalysis]

We're about to drop a new report...stay tuned! 📚🚨

[Ben Garfinkel @bmgarfinkel]

This is a major milestone for international AI governance - the first time companies from several different countries (including both the US and China) have signed onto joint safety commitments. Great to see #AISeoulSummit keeping up momentum from last year's AI Safety Summit

[Convergence Analysis @CvgAnalysis]

Our AI Governance team just published our first major report by @0xDeric evaluating an AI chip registration policy for the US! You can find the full report below: https://t.co/gDivQDts5z

Quick summary:

This report evaluates the feasibility and potential impacts of a US policy requiring the registration and tracking of high-end AI chips. It provides an overview of the current AI chip landscape, proposes a policy design, and assesses the policy's likely effectiveness, implementation challenges, and potential externalities for AI development and US-China relations.

Key takeaways:

1. The US executive branch is likely to be interested in implementing an AI chip registration policy in the near future due to its geopolitical AI competition with China.

2. A US AI chip registration policy passed in the near future would likely consist of a manufacturing registration requirement and transfer reporting requirements.

3. Implementing a chip manufacturing registration requirement at this time has a streamlined enforcement strategy, requiring compliance from only 2-3 American chip vendors.

4. Implementing a chip transfer reporting requirement globally would require a drastic expansion of current FDPR laws.

5. The US executive branch has the authority to create and enforce such an AI chip registration policy, but it lacks the ability to authorize additional funding.

6.A globally enforced AI chip registration policy would immediately improve enforcement of existing export controls, and lay the groundwork for future compute regulations based on an AI chip registry.

7. An AI chip registration policy would have minimal negative externalities on overall AI innovation and Western competition, but may spur greater concerns about US governmental overreach.

[Convergence Analysis @CvgAnalysis]

4/

[Convergence Analysis @CvgAnalysis]

1/ We've published our first governance research report: Evaluating an AI Chip Registration Policy

You can find the full paper here:

https://t.co/UnVB3GLnsr

[Convergence Analysis @CvgAnalysis]

10/ Interested in leaving feedback? Check out our EA Forum post below:

https://t.co/iMiD0ZM05b