These are slides used in CMU's panel on AI for Social Good in Washington DC, which I moderated. I talk about some of the positive uses of AI technologies, some of the benefits of AI in general, as well as some of the challenges.
June 20, 2018
https://cmuaijune2018.splashthat.com/
3. Previous Panels
AI 101
AI and the Future of the Workforce
AI for National Defense
Today, how can we use AI for social good?
4.
5. • Pilot test in Pittsburgh of
smart traffic light
• Reduced 25% by eliminating
stops and reducing wait
time, not by increasing
travel speeds
• Reduced emissions by 20%
by reducing stops and idling
• Over 40% less time waiting
at intersections
6. Crisis Text Line
• 33M text messages,
5300 active rescues
• Wednesdays are most
anxiety-provoking day
• “Ibuprofen”, “Advil”,
and more
predictive than “kill”,
“suicide”, “harm”
9. Some Benefits of Using AI
Potential for reducing human bias
Increased accuracy (better than humans)
Ability to process big data, especially sensor streams
Available 24/7
Can help prioritize limited resources
11. Pittsburgh’s Hill District
Was ground zero for Jazz
musicians in 20th century
8,000 residents and 400
businesses, construction
decimated the economic
center of African-American
Pittsburgh
Median Income (2009):
$17,939
12. Some Challenges for AI
Data Quality – AI is only as good as its data
Privacy – Same data can be used for good / bad
Ethical issues – Ex. jobs, dual-use, bias
Note: These challenges not unique for AI for social
good
13. Fei Fang
Assistant Professor, Institute
for Software Research at
Carnegie Mellon University
Research in artificial intelligence,
with applications to security,
sustainability, and mobility
14. Jay Qi
Data Science Lead at Uptake,
industrial AI software company
headquartered in Chicago
Uptake.org (philanthropic and
social innovation arm) works
with orgs around world to build
data-driven tools and to grow
their own data science talent
15. Anna Bethke
Data scientist at Intel’s Artificial
Intelligence Product Group where
she heads their AI4Good efforts
AI in crop growth, determining how
algorithms can aid greenhouse
operators achieve highest crop yields
while minimizing the use of resources