Increasing public engagement in AI research has led to the use of surveys to understand public values and perceptions. This paper analyzes a pilot survey and reviews 44 papers on AI-related surveys, revealing Western biases, lack of critical discourse, and inconsistent transparency. Suggestions are offered to improve survey design and interpretation for responsible public engagement.

By analyzing 128 papers (31% of proceedings) from 2018 to 2022 in an AI research conference, we discovered 84% solely feature Western participants, mainly U.S. Efforts to diversify stem from researchers' local data collection signaling that researchers need to collect data from under-represented populations to obtain an inclusive worldview.

Our study uses interviews with 19 developers and a survey of 309 users to reveal shared privacy concerns in permission systems: trust, app reputation, and data access. Developers sometimes seek multiple permissions due to confusion or third-party needs. Users perceive permission granting as their choice, aligning with developers.

Mobile apps and ad networks blur privacy lines, posing regulatory risks. We study ad network interfaces via empirical studies and think-aloud sessions involving 11 developers. Insights reveal scattered, confusing privacy info, complicating compliance. Ad networks shift the responsibility, while developers seek user-friendly tools. Our recommendations: centralize privacy details, simplify configurations, provide testing mechanisms, and enhance documentation through multimedia resources for app developers, bolstering privacy adherence.

Acquiring skilled programmers remains a challenge for tech-focused empirical studies. We delve into this using established surveys instruments, targeting CS students and diverse crowdsourcing platforms (Appen, Clickworker, MTurk, Prolific). 613 participants were recruited, gauging programming skills, security outlook, and development self-efficacy. Our exploration weighs costs, quality, skill levels, and privacy attitudes across recruitment channels.

Ad networks offer personalized ads to boost developers' revenue, but their framing influences their decisions, affecting user privacy. We performed an online survey experiment with 400 app developers, varying choice frames around ad personalization across six conditions. Key outcomes: Highlighting privacy impact raises non-personalized ad preference by 11 times compared to a condition where privacy is not salient; Developer choices are impacted by revenue, user privacy, and relevance.

We engaged 12 Privacy Champions in software teams to explore their motivations, tactics, and barriers in safeguarding user privacy. Barriers include weak privacy culture, internal conflicts, tech constraints, and elusive metrics. Privacy Champions combat these through dialogue, management backing, and strategic guidelines. Notably, hands-on code reviews and practical training trump generic awareness efforts.