BG Blog: Austin’s AI Policy Seeks To Balance Trust, Transparency

With artificial intelligence (AI) technology showing tremendous potential to impact all areas of public and private life, Austin’s city government is poised to assess how to enhance city services and decision-making processes while addressing the myriad concerns with any technological advancements. Looking toward the guidance of technology experts in the community, City Council voted at its Feb. 15 meeting to have city staff implement a comprehensive resolution aimed at integrating AI responsibly into various facets of municipal operations.

Chris Stewart, Austin’s former Chief Information Officer, reflected on the city's current landscape on a recent episode of the BG Group Podcast, emphasizing the necessity of a collaborative effort involving multiple city departments. He underscored the importance of seeking external feedback and expertise to ensure that the city's AI initiatives align with community needs and values.

Stewart highlighted that while AI has become a prevalent topic of discussion in recent years, it gained further prominence due to instances like ChatGPT's rising popularity and concerns about its impact on personal lives, with the question of issues surrounding children and cheating making it clear how pervasive the technology has become.

City Council's resolution outlines key principles to guide the responsible use of AI within city operations. These principles are:

Innovation and Collaboration: the resolution emphasizes the use of AI systems to promote teamwork between city employees and AI systems. By leveraging the strengths of both human expertise and AI capabilities, the city aims to enhance its decision-making processes and improve the delivery of services to residents. This approach ensures the final decision-making remains with trained city personnel, preserving accountability and transparency in governance.

Data Privacy and Security: safeguarding the confidentiality, integrity, and availability of data is of critical importance. Through rigorous inventorying and evaluation of AI systems, the city aims to minimize security risks and ensure data privacy standards are upheld at every stage of AI system development, training, testing, deployment, and use. Policies will be implemented to decrease the risk of data breaches and privacy loss, preserving trust and confidence among residents and stakeholders.

Transparency: the AI strategy must ensure that the development, use, and deployment of AI systems are evaluated for compliance with all relevant laws and regulations. By making documentation related to the purpose and impact of AI systems publicly available, the city can promote accountability and facilitate informed decision-making.

Explainability and Interpretability: the city sees that AI systems and their outputs must be explainable and interpretable. The rationale behind AI-generated decisions should be clear and understandable, with outputs communicated in clear language that reflects the context of their use and deployment. This is another example of the acknowledgment of the trust and confidence needed in AI systems among residents and stakeholders.

Validity and Reliability: ensuring the validity and reliability of AI systems is paramount to their successful implementation and use, which means the systems must be evaluated thoroughly for their performance under expected conditions of use. Ongoing evaluation and auditing processes must be in place to monitor system accuracy throughout their development and deployment lifecycle, so as to mitigate the risk of unintended consequences or errors.

Bias and Harm Reduction: AI systems must operate through an equity lens, aligning with the city’s anti-racist and anti-discriminatory commitments. This involves assessing AI systems for potential biases and unintended harms arising from data, human, or algorithmic biases, and addressing them proactively prior to purchase and deployment. By prioritizing bias and harm reduction, the city can ensure AI technologies serve all residents equitably.

Central to the resolution is the establishment of accountability and oversight mechanisms for AI usage. This includes implementing ethical procurement standards, conducting regular audits and evaluations of AI systems, engaging with expert stakeholders and the community, and establishing an oversight board dedicated to evaluating responsible AI practices.

Importantly, the resolution emphasizes that employees should not be held accountable for errors beyond their control, emphasizing a culture of accountability and support within city operations.

The resolution sets a clear timeline for progress, with an update scheduled for May 28, right before the Council's summer recess. This update will provide insight into the development of AI guidelines, accountability strategies, and considerations for the city's workforce.

It represents a commitment to transparency and ongoing evaluation as Austin navigates the evolving landscape of AI technology. Council also wants the city manager - in fairly short order - to develop a comprehensive plan for managing the impact of AI on the workforce.

Stewart said the primary departments involved in these efforts will be Communications and Technology Management, the Innovation Office, and the Information Security Office, with all departments having to take part in the inventory of what AI technology is already in use, such as chatbots or other tightly constrained functionalities that have been deployed to make the city operate more efficiently.

With the plan prioritizing job protection, training, and support for employees, while also addressing concerns related to job displacement, labor standards, workplace safety, and equity, there will be training programs implemented to equip employees with the necessary skills and knowledge to engage with AI systems responsibly, emphasizing ethical considerations, privacy protection, and responsible AI practices.

With the federal government and major cities such as Seattle, New York City, and San Jose already having AI policies in place, Stewart said that many decision-makers at city hall will be able to reach out to their counterparts in other municipal governments for guidance and best practices.

A natural question of “Who owns this?” is sure to arise as the city manager and department heads begin to parse out the different components of the resolution. It’s possible the late May deadline may come with more of a progress report rather than a finely tuned comprehensive plan to address one of the most disruptive and promising technologies to come along in decades.

//A.J. Bingham, Founder & CEO, Bingham Group

🟪 Contact A.J. at: aj@binghamgp.com