The End Of Knowledge - Vault 4/13 - AI For Good - 2018

graph LR classDef trust fill:#d9e6f2, font-weight:bold, font-size:14px classDef ibm fill:#e2f2d9, font-weight:bold, font-size:14px classDef partnership fill:#f2e1d9, font-weight:bold, font-size:14px classDef summits fill:#e1d9f2, font-weight:bold, font-size:14px classDef general fill:#d9f2ea, font-weight:bold, font-size:14px A[Interview] --> B[AI Ethics Leader,
IBM Research. 1] A --> C[Trust in AI track
had three themes. 2] A --> D[Trust needed between users,
stakeholders, cultures. 3] A --> E[Nine projects demonstrated
trustworthy AI building. 4] A --> F[Trust Factory
global incubator launched. 5] A --> G[AI needs
appropriate trust levels. 6] C --> H[Recent initiatives aim for
beneficial, trustworthy AI. 7] C --> I[IBM Research works on bias,
explainability, value. 8] C --> J[IBM doesnt reuse client data,
attracts clients. 9] C --> K[Principles for Cognitive Era -
AI augments humans. 10] B --> L[IBM co-founded Partnership
on AI. 11] L --> M[Partnership grew from
6 to 53 partners. 12] M --> N[Pillars: fair/transparent/accountable AI,
safety, jobs, collaboration. 13] M --> O[Corporate presence in Partnership
for real-world deployment. 14] D --> P[AI for Good summits
evolved with ethics field. 15] P --> Q[2017 summit introduced
UN, AI communities. 16] Q --> R[2018 summit action-oriented
with projects, timelines. 17] R --> S[Four tracks, 9-15 projects
each in 2018. 18] class B,C,D,E,F,G general class H,I,J,K ibm class L,M,N,O partnership class P,Q,R,S summits

Resume:

1.- Francesca Rossi is the AI Ethics Global Leader at IBM Research.

2.- The "Trust in AI" track at the AI for Good Global Summit 2018 was structured around three themes of trust.

3.- Trust is needed between users and AI systems, among different AI stakeholders/communities, and among different cultures.

4.- Nine example projects were presented in the trust track to demonstrate how to build trustworthy AI.

5.- The "Trust Factory" concept was launched as a global incubator to expand the work beyond the one-day track.

6.- AI needs the right level of trust - undertrust prevents realizing benefits while overtrust attributes unwarranted capabilities.

7.- Many new initiatives in the past 2-2.5 years aim to make AI beneficial and trustworthy for individuals and society.

8.- IBM Research publishes work on bias detection/mitigation, explainability, and value alignment to develop responsible AI.

9.- IBM has a data responsibility policy to not reuse client data for other purposes. This attracts clients but limits data.

10.- IBM's "Principles for the Cognitive Era" state AI should augment human intelligence, not replace it, to help professionals work better.

11.- IBM co-founded the Partnership on AI with 5 other major tech companies to discuss AI's societal impact.

12.- The Partnership has grown from 6 to 53 multi-stakeholder partners to identify and resolve AI deployment issues.

13.- Thematic pillars include fair/transparent/accountable AI, safety-critical AI, AI's impact on jobs, and human-AI collaboration.

14.- The Partnership's corporate presence provides a voice on real-world AI deployment that is unique compared to academic initiatives.

15.- The AI for Good summits have evolved as the AI ethics field has progressed.

16.- The 2017 summit introduced the UN development and AI communities to each other's terminologies and issues.

17.- The 2018 summit was more concrete and action-oriented with specific projects and 12-month timelines.

18.- The four 2018 tracks had 9-15 projects each, demonstrating the readiness for tangible outputs compared to earlier discussion phases.

Knowledge Vault built byDavid Vivancos 2024