The Founder of CollectiWise and Venture Partner at Post Urban, Johannes Castner, attended Worldwide AI Webinar and delivered an insightful keynote on the ethical dilemmas of human-centric AI. Catch the highlights of his speech here.
Systems should empower human
Claiming that systems should empower humans, Johannes listed out how sound algorithm systems could help with world innovation and solve global issues:
-
increase revenues
-
better understand customers
-
optimize marketing expenses
-
enable self-driving cars
-
enable smart parking
-
eliminate patients’ suffering
-
rewild the wastelands
-
eliminate pollution
-
make society more inclusive
The list goes on.
The bottom line is AI in its power can be a great assistant to humans. Good algorithmic systems can be beneficial.
However, since algorithms represent humans, usually the original creators and AI does what humans teach it to do, biases and stereotypes are inevitable.
3 main issues with algorithms
Johannes gave examples of Amazon’s recruiting tool that turned out to show biases against women as they trained the system on previous CVs and racial bias being found in a major healthcare risk algorithm, resulting from healthcare costs which were used as the differentiator.
Since human biases are reproduced in algorithms by flawed data selection processes, the three issues with algorithms are beyond the social choice problem and biases:
-
Business models are not based on human freedom
-
Algorithms lack transparency
-
Users are not empowered to improve the system
The recipe for ethical pluralism
Johannes believed that we should not ask if artificial intelligence is good or fair but rather ask how it shifts power and leads us to participatory AI.
To enable and build participatory AI systems, he advised companies to:
-
rule out the unethical
-
use a participatory toolkit
-
create vibrant participation
-
gamification