Knowledge Vault 4 /37 - AI For Good 2019
Distinguished Engineer, IBM Research
Noam Slonim
< Resume Image >
Link to IA4Good VideoView Youtube Video

Concept Graph & Resume using Claude 3 Opus | Chat GPT4o | Llama 3:

graph LR classDef project fill:#d4f9d4, font-weight:bold, font-size:14px classDef goal fill:#f9d4d4, font-weight:bold, font-size:14px classDef demo fill:#d4d4f9, font-weight:bold, font-size:14px classDef narrative fill:#f9f9d4, font-weight:bold, font-size:14px classDef team fill:#f9d4f9, font-weight:bold, font-size:14px A[Distinguished Engineer, IBM
Research] --> B[Project started in 2012,
post-Jeopardy! success. 1] A --> C[Goal: develop machine
to debate humans. 2] A --> D[Demonstrated system debating
top human debater. 3] A --> E[Collected arguments on
controversial topics. 4] E --> F[Generate compelling narratives
to aid decisions. 5] E --> G[People contribute arguments
on important topics. 6] A --> H[Enhances communication between
decision-makers, audiences. 7] H --> I[Helps decision-makers with
public opinion. 8] H --> J[Better than simple
yes/no surveys. 9] A --> K[Presented results on social
medias harm. 10] K --> L[Generated narratives supporting
and opposing social media. 11] K --> M[Narratives available on
project website. 12] A --> N[Slonim worked on project
over 8 years. 13] A --> O[Interdisciplinary team beyond
data scientists. 14] O --> P[Included experts in
various fields. 15] O --> Q[Diverse team made
project enjoyable. 16] A --> R[Project debater addresses pros/cons
for decisions. 17] A --> S[Few AI technologies
aid human decisions. 18] A --> T[Significant progress, more
expected in future. 19] A --> U[Collects sophisticated opinions
beyond surveys. 20] class A,C,D,K,M,N,R,S,T project class B goal class E,F,G,H,I,J narrative class O,P,Q team

Resume:

1.- Project started in 2012 as a grand challenge after chess, Go, and Watson's Jeopardy! success.

2.- Goal was to develop a machine that could debate humans, which seemed impossible at first.

3.- After 6+ years of work, they demonstrated the system debating a top human debater in San Francisco.

4.- Developed project debater speech by crowd to collect arguments on controversial topics from many people.

5.- Aim to generate compelling narratives from the arguments to enhance decision-making.

6.- People tend to contribute arguments about topics they care about.

7.- Can enable better communication between decision-makers and target audiences, e.g. governments and citizens.

8.- Won't replace decision-makers but will help them make better decisions by providing public opinion.

9.- Superior to simple surveys that just ask yes/no questions and take the average.

10.- Presented results from a May question on if social media does more harm than good.

11.- System generated two narratives - one supporting (54% of arguments) and one against social media (46%).

12.- Narratives are available on the project debater speech by crowd website.

13.- Project has been worked on for over 8 years by Noam Slonim.

14.- Required an interdisciplinary team beyond just data scientists and programmers.

15.- Included experts in system architecture, natural language understanding, AI, text analysis, and even philosophy.

16.- The heterogeneous nature of the team made the project more fun.

17.- Project debater is just the beginning, touching on the fundamental problem of understanding pros/cons for better decisions.

18.- There aren't many AI technologies explicitly designed to help humans make better decisions currently.

19.- Significant progress has been made and they believe further progress will come in the future.

20.- Allows collecting sophisticated opinions from people impacted by decisions, beyond just simple survey questions.

Knowledge Vault built byDavid Vivancos 2024