YC
A.I. Policy and Public Perception - Miles Brundage and Tim Hwang
@ Y Combinator
04/25/18
Takeaways
YC
Y Combinator Cast
04/25/18
@ Y Combinator
The rise of cryptocurrency has shifted attention away from artificial intelligence, which is a positive development for the AI field.
MB
Miles Brundage
04/25/18
@ Y Combinator
My paper is titled 'The Malicious Use of Artificial Intelligence,' focusing on how AI can be deliberately misused, including generating fake news and using AI with drones for terrorist attacks.
MB
Miles Brundage
04/25/18
@ Y Combinator
AI is a dual-use technology, similar to biotechnology, and we need to establish norms for its responsible use, including responsible disclosure of vulnerabilities.
TH
Tim Hwang
04/25/18
@ Y Combinator
While my research focuses on the intersection of AI and disinformation, I emphasize that AI can be intentionally used for harmful purposes, such as creating believable fake content.
TH
Tim Hwang
04/25/18
@ Y Combinator
My methodology involves analyzing current research to predict potential uses of AI, especially as tools become more accessible through cloud services.
TH
Tim Hwang
04/25/18
@ Y Combinator
The public's perception of AI is often shaped by popular culture and media, leading to misunderstandings about what AI actually is and how it functions.
TH
Tim Hwang
04/25/18
@ Y Combinator
Concerns about AI include international competition, particularly with China's investments in AI, and the need for interpretability in AI systems to explain their decision-making processes.
MB
Miles Brundage
04/25/18
@ Y Combinator
The AI community is actively addressing issues of fairness, accountability, and transparency, which are crucial for the responsible deployment of AI technologies.
MB
Miles Brundage
04/25/18
@ Y Combinator
There is significant progress in AI applications in healthcare, particularly in image recognition tasks, which are becoming increasingly effective and reliable.
MB
Miles Brundage
04/25/18
@ Y Combinator
The workshop 'AI in the Wild' highlights the shift from lab-based AI research to addressing practical deployment challenges in real-world scenarios, including the deployment of AI in healthcare, which is still largely in pilot phases, with specific applications like skin lesion analysis being among the few widely used examples.
MB
Miles Brundage
04/25/18
@ Y Combinator
There's a significant amount of ongoing research and pilot projects in AI for healthcare, but many technologies have not yet been widely rolled out due to issues of interpretability and fairness that need to be addressed first.
MB
Miles Brundage
04/25/18
@ Y Combinator
Interpretability in AI is a crucial debate, questioning whether it truly matters to end-users, such as doctors, and if the field is overemphasizing its importance.
TH
Tim Hwang
04/25/18
@ Y Combinator
The regulatory landscape for AI varies significantly across regions, with the U.S. taking a case-by-case approach while Europe tends to implement broader regulations for automated decision-making.
TH
Tim Hwang
04/25/18
@ Y Combinator
In the U.S., the regulatory focus is often on high-risk areas like medical applications, whereas Europe has a more comprehensive framework for AI governance.
MB
Miles Brundage
04/25/18
@ Y Combinator
AI policy discussions often lack the methodological rigor seen in other fields, such as climate change, where expert consensus is synthesized into authoritative reports.
TH
Tim Hwang
04/25/18
@ Y Combinator
There's a need for better collaboration between AI researchers and policy experts to ensure that technical developments align with societal needs and ethical considerations.
MB
Miles Brundage
04/25/18
@ Y Combinator
The AI field is still developing its understanding of policy implications, and there's a growing need for researchers to engage in policy discussions and consider the societal impact of their work. People in both AI research and civil society are often concerned about fairness, but they operate under different constraints.
MB
Miles Brundage
04/25/18
@ Y Combinator
Researchers in industry focus on practical applications of fairness, while those in academia may take a broader perspective but rely on industry connections.
MB
Miles Brundage
04/25/18
@ Y Combinator
The Future of Humanity Institute has interdisciplinary teams, including philosophers, ethicists, and political scientists, working on AI and biotechnology issues.
MB
Miles Brundage
04/25/18
@ Y Combinator
Existential risk is a major motivation for the Future of Humanity Institute's work, but they also focus on long-term issues and the positive applications of AI.
MB
Miles Brundage
04/25/18
@ Y Combinator
There is a common cause between those addressing immediate AI issues and those focused on long-term risks, as both aim to maximize benefits and minimize risks.
MB
Miles Brundage
04/25/18
@ Y Combinator
Value alignment is a key concept in AI safety, addressing how to ensure AI systems reflect human values and preferences, especially in terms of fairness and bias.
MB
Miles Brundage
04/25/18
@ Y Combinator
Transparency in AI development could help avoid international conflicts, similar to historical arms control agreements that emphasized verification and cooperation.
MB
Miles Brundage
04/25/18
@ Y Combinator
There may be specific domains where norms around openness in AI research should be reconsidered to prevent potential harms, such as in the case of driverless cars.
MB
Miles Brundage
04/25/18
@ Y Combinator
The AI research community generally favors openness, as it aligns with both broad interests and individual goals of researchers and companies.
MB
Miles Brundage
04/25/18
@ Y Combinator
Open publishing in AI research is crucial for realistic public policy and understanding the current state of the field, which benefits everyone involved.