What experience do you have using or working with artificial intelligence, either personally or professionally, and where do you get your information about AI? How have those experiences and sources shaped your views of the technology?
My legal career is focused on litigating easements and title disputes, and in my research I occasionally uncover fraud (see Asheville Watchdog’s Equity Erased series for an example). That required me to carefully review documents, speak with my client about what I found, share my concerns, answer their questions, and make recommendations for a course of action based on my decades of experience helping people in complex title situations. I don’t think those human steps of interpretation, communication, clarification, and context-based problem-solving can or should be replaced by AI.
I learn about AI from sources as close and trustworthy as my son who studied AI as part of his recent computer science masters. He’s deeply skeptical about AI and he’s also pursuing a PhD that will be based on research that seeks to harness AI for good to accurately predict rainfall to protect people from climate change disasters. He is pursuing that study because of Helene’s devastating impact on Asheville, and what a poor job the rainfall modeling did to warn people of the risk. Like many people, I also hear about AI from sources that are distanced and untrustworthy, including social media and news coverage that reads like advertorials for AI companies and the AI supply chain.
My takeaway is that AI is not neutral. AI has the biases and assumptions of the people who created the code for it and whose intellectual property has been used to create the “I” in “AI”.
Where, if anywhere, do you think Asheville should use AI to improve city services or operations? Are there uses you believe should be off-limits?
For context, the City of Asheville has an AI policy that applies to staff. The first two sentences of the AI policy instituted last year by our City Manager are: “Generative AI is a tool. We are responsible for the outcomes of the tools we use.”
Also, there’s a list of prohibited uses, including: “Fully automated decisions that do not require any meaningful human oversight but substantially impact individuals. Social scoring, or the use of AI systems to track and classify individuals based on their behaviors, socioeconomic status, or personal characteristics”
My opinions follow.
Used: Efficiency tasks that are unlikely to infringe on human rights and don’t invade privacy, such as infrastructure monitoring and maintenance and service routing.
Off-limits: Everything listed as off-limits in our current staff policy, including facial recognition (stated here for clarify even though it is a form of biometric identification and already prohibited), and also predictive policing and benefit/risk-assessment tools because AI is not neutral and those uses are known to perpetuate racial and class biases and existing power structures, and designed to prevent accountability.
More could be added to this list, but this gives you the idea.
When AI is used in a city decision that affects an individual resident, what human oversight and ability to challenge that decision should be required?
The wording of the question is so broad that it would be difficult to argue that AI hasn’t played some role in every decision since WordPerfect and Microsoft Word first offered spellcheck in the 1980s. And harms from inaccurate name changes and erroneous punctuation are real.
I will answer what I think is your substantive question.
We can start with the aspiration for ensuring human oversight and the ability to challenge decisions, and land on actionable policy that protects, prevents harm, and serves us. And then we need to continue evolving, because technology will keep evolving.
As a first step, in order to know what AI is used, so there is a basis for lodging a challenge, we would need to determine effective ways of publishing a directory of all City AI software/systems in use. Without this, how will residents know if AI was used in a process that led to a decision?
What policies or safeguards should Asheville put in place for its use of AI, particularly when it comes to transparency, privacy, bias and accountability?
I support the transparency, privacy, bias and accountability policies and safeguards in the current AI policy for staff, those proposed in my response to question 3, and am open to hearing other ideas.
