OpenAI Special Projects 2016

OpenAI has identified four key problem areas that it considers essential for the advancement of AI and its long-term societal impact. These priorities focus on a combination of strengthening AI capabilities—such as automated programming and complex simulations—and developing defensive measures against the potential misuse of AI systems.

AI Breakthrough Detection

OpenAI aims to develop methods to detect if a covert breakthrough AI system is being used in the world. This is driven by the probability that as more organizations allocate resources to AI research, an organization may make an undisclosed breakthrough and deploy a system for malicious purposes. Proposed detection methods include monitoring news, financial markets, and online games.

Automated Programming

OpenAI is seeking to build an agent capable of winning online programming competitions. The goal is to create a program that can write other programs, which OpenAI describes as a "powerful" capability.

AI-Driven Cybersecurity Defense

OpenAI is prioritizing the development of AI techniques to defend against sophisticated hackers. The company anticipates that an early use of AI will be to break into computer systems, and seeks to create defensive AI methods to counter hackers who are themselves using AI methods.

Complex Multi-Agent Simulations

OpenAI is interested in building a large-scale simulation containing numerous long-lived agents. The objective is to create an environment where agents can interact, learn over extended periods, and discover language, and achieve a variety of complex goals.