Cooperative AI

Get Started. It's Free
or sign up with your email address
Cooperative AI by Mind Map: Cooperative AI

1. AI–AI cooperation

1.1. Mastering team games requires many skills essential to cooperation

1.2. Research avenues

1.2.1. AIs nderstand what teammates are thinking and planning

1.2.2. communicate plans

1.2.3. ad hoc teamwork

1.3. AI agents need to learn to manage harder cooperation problems as humans do

1.4. Devising variants of Diplomacy that modify the difficulty of challenges

1.4.1. introducing an agreed simple vocabulary

1.4.2. permitting binding commitments

2. From autonomy to cooperation

2.1. People become most productive when working cooperatively

2.2. Certain kinds of autonomy in AI systems are useful

2.3. clusters of work

2.3.1. AI-AI cooperation

2.3.2. Human-AI cooperation

2.3.3. AI for human collaboration

2.4. Four elements of cooperative intelligence

2.4.1. Understanding

2.4.2. Communication

2.4.3. Commitment

2.4.4. Norms and institutions

3. Human–AI cooperation

3.1. AI practitioners have dreamt of building autonomous and human-like intelligence

3.1.1. Autonomous systems might pose greater safety risks - displacing labour

3.2. AIs Help humans

3.2.1. Care robots help human health workers

3.2.2. Collaborative industrial robots labourers

3.2.3. Personal AI assistants (Amazon’s Alexa, Apple’s Siri and Google Assistant) help with schedule

3.3. Challenge facing is that human is involved

3.4. Researchers should assess the performance of human–machine teams

4. AI for human collaboration

4.1. Developing AI assistants that complement human intelligence and depend on us for tasks

4.2. Digital systems and AI can expand technology mediating our ability to find and process information communicate and self-organize

4.3. AI tools is strongly pushed to promote collaborations

4.4. Digital platforms provide tools to combine user-provided content

4.5. AI advances improved community infrastructure, user rating and reputation systems

4.6. Platforms for political deliberation can be designed to promote empathy

5. Next steps

5.1. Cooperative AI must connect with the broader science of cooperation

5.2. Cooperative AI research will gain momentum if investigators can adopt challenging benchmarks

5.3. Promoting research into cooperative AI will require social interventions

5.3.1. NeurIPS drew speakers from diverse array of disciplines => Open Problems in Cooperative AI review.

5.3.2. Cooperative AI Foundation backed by a large philanthropic commitment to is being established

5.4. Feasibly much-needed global cooperation in the future thanks to forward upward potential of AI