<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Dr. Sandra Whitehead on Medium]]></title>
        <description><![CDATA[Stories by Dr. Sandra Whitehead on Medium]]></description>
        <link>https://medium.com/@sandrawhitehead?source=rss-4988eb5fe7e8------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*I1CYS2QmjvXEyQeQ4k0IPA.jpeg</url>
            <title>Stories by Dr. Sandra Whitehead on Medium</title>
            <link>https://medium.com/@sandrawhitehead?source=rss-4988eb5fe7e8------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 01:01:09 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@sandrawhitehead/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[From Noise to Insight: Using AI to Rethink Public Comment Analysis]]></title>
            <link>https://medium.com/@sandrawhitehead/from-noise-to-insight-using-ai-to-rethink-public-comment-analysis-58fa27c13900?source=rss-4988eb5fe7e8------2</link>
            <guid isPermaLink="false">https://medium.com/p/58fa27c13900</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[decision-making]]></category>
            <category><![CDATA[lived-experience]]></category>
            <category><![CDATA[community-engagement]]></category>
            <category><![CDATA[cities]]></category>
            <dc:creator><![CDATA[Dr. Sandra Whitehead]]></dc:creator>
            <pubDate>Mon, 29 Sep 2025 20:51:38 GMT</pubDate>
            <atom:updated>2025-09-29T20:51:38.182Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="Diverse stakeholders with thought bubbles" src="https://cdn-images-1.medium.com/max/1024/1*kDoUoYRWpe91iZn4T46VQA.png" /></figure><p>If you’ve ever submitted a comment on a government project or regulation, you may have wondered: <em>does anyone actually read this?</em></p><p>The short answer is yes — but the way comments are handled often falls short of meaningful engagement. Agencies are required to collect and respond to public comments, and they do. But when thousands of people submit feedback, staff are under immense pressure to summarize, categorize, and respond efficiently. That means a lot of nuance is lost in the shuffle.</p><p>Public comments are a cornerstone of democratic decision-making (<a href="https://onlinelibrary.wiley.com/doi/10.1111/puar.12361">Fung, 2015</a><a href="https://onlinelibrary.wiley.com/doi/10.1111/puar.12361).">)</a>. They give ordinary people a chance to shape projects that affect their lives: energy facilities, zoning rules, environmental standards, transportation plans. But in practice, the system has become overwhelmed by volume. For major rules and infrastructure projects, it’s not uncommon for agencies to receive <em>tens of thousands</em> of comments. Some rule making processes — like those from the Environmental Protection Agency or the Department of Transportation — have drawn hundreds of thousands (<a href="https://www.gao.gov/products/gao-21-103181">GAO, 2021</a>)</p><p>So here’s the question: what if artificial intelligence could help?</p><h3><strong>The Overload Problem</strong></h3><p>Public comment systems were designed for a slower, paper-based era. In today’s digital world, submitting a comment is as easy as sending an email or clicking “submit” on a form. The result: more participation, but also more data than staff can realistically process.</p><p>The traditional approach is labor-intensive:</p><ul><li>Staff read a sample of comments to identify major themes.</li><li>Comments are sorted into categories (support, oppose, suggest modification).</li><li>A response is drafted, often at a very general level.</li></ul><p>This makes it difficult to capture nuance or to demonstrate to the public that their input truly mattered. As the GAO (<a href="https://www.gao.gov/products/gao-21-103181">2021</a>) noted, agencies often provide “boilerplate” responses that leave stakeholders dissatisfied.</p><p>The irony is clear: the more people participate, the harder it is to demonstrate that participation was meaningful.</p><p><strong>Where AI Could Help</strong></p><p>Artificial intelligence is designed to find patterns in large volumes of information. That makes it well-suited to public comment analysis, where the goal is not to read every word manually, but to extract meaning and identify recurring concerns.</p><p>Here are some practical applications — with tools already available:</p><ol><li><strong>Clustering and Thematic Analysis</strong></li></ol><ul><li><em>How it works</em>: NLP models can group thousands of comments into clusters of related themes.</li><li><em>Tools</em>: Topic modeling with <a href="https://maartengr.github.io/BERTopic/index.html">BERTopic</a> , <a href="https://cloud.google.com/natural-language">Google Cloud Natural Language</a> or <a href="https://www.ibm.com/watson-natural-language-processing">IBM Watson NLP</a>.</li></ul><ol><li><strong>Sentiment Analysis</strong></li></ol><ul><li><em>How it works</em>: AI can identify whether comments are broadly positive, negative, or neutral.</li><li><em>Tools</em>: <a href="https://azure.microsoft.com/en-us/services/cognitive-services/text-analytics/">Microsoft Azure Text Analytics</a>, <a href="https://aws.amazon.com/comprehend/">Amazon Comprehend</a> ), and open-source libraries like <a href="https://ojs.aaai.org/index.php/ICWSM/article/view/14550">VADER</a>.</li></ul><ol><li><strong>Summarization</strong></li></ol><ul><li><em>How it works</em>: Large language models can generate concise summaries of lengthy submissions, enabling staff to process complex arguments more efficiently.</li><li><em>Tools</em>: OpenAI <a href="https://arxiv.org/abs/2005.14165">GPT models</a>, Anthropic <a href="https://www.anthropic.com/claude">Claude</a>, or <a href="https://docs.cohere.com/docs/summarize">Cohere Summarize</a>.</li></ul><ol><li><strong>Duplicate Detection</strong></li></ol><ul><li><em>How it works</em>: AI can flag identical or near-identical comments generated by campaigns, grouping them together so staff don’t waste time on repetition.</li><li><em>Tools</em>: <a href="https://scikit-learn.org/stable/">scikit-learn</a> or <a href="https://spacy.io/">spaCy</a> similarity pipelines.</li></ul><ol><li><strong>Trend Tracking Over Time</strong></li></ol><ul><li><em>How it works</em>: Models can compare one round of public comments to another, identifying shifts in public sentiment or emerging concerns.</li><li><em>Tools</em>: <a href="https://www.tableau.com/">Tableau</a> with NLP plug-ins, or custom Python pipelines using <a href="https://huggingface.co/transformers/">Hugging Face transformers</a>.</li></ul><p><strong>A Real-Life Example: The EPA and Power Plant Rules</strong></p><p>We don’t have to imagine this in theory. In 2018, the EPA proposed revisions to its Affordable Clean Energy (ACE) rule. The agency received <a href="https://www.epa.gov/stationary-sources-air-pollution/affordable-clean-energy-rule"><strong>over 500,000 public comments</strong></a>. Traditionally, staff would spend months manually categorizing comments. Instead, EPA piloted a text-mining tool to help identify duplicate comments, cluster themes, and prioritize unique submissions.</p><p>The result: staff were able to process the comments more efficiently and produce a clearer summary of public concerns. While the tool wasn’t perfect, it demonstrated the potential of AI-assisted analysis to handle scale while still highlighting substance.</p><p>Similarly, the <a href="https://www.cpuc.ca.gov/">California Public Utilities Commission</a> has begun exploring NLP tools to analyze public input on utility rate cases, where technical and community concerns collide.</p><h3><strong>Benefits of an AI-Assisted Approach</strong></h3><p>AI won’t replace the need for human judgment, but it can make the process more efficient, more transparent, and more responsive:</p><ul><li><strong>Efficiency</strong>: Staff time can be redirected from manual sorting to meaningful engagement and response.</li><li><strong>Transparency</strong>: Agencies could share AI-generated thematic maps or dashboards to show the public how their input fits into the larger conversation.</li><li><strong>Responsiveness</strong>: By identifying themes early, agencies can address concerns before they escalate into opposition or litigation.</li></ul><p>Imagine if, instead of reading a generic “thank you for your comment,” participants saw an interactive dashboard showing that their concern — say, traffic impacts near a school — was echoed by 437 others and is being addressed in a revised plan. That’s not just efficiency. That’s trust-building.</p><h3><strong>Risks and Limitations</strong></h3><p>Of course, AI is not a magic wand. Introducing it into public comment analysis carries real risks:</p><ul><li><strong>Bias in Training Data</strong>: Models may misinterpret or underrepresent voices if the training data is <a href="https://dl.acm.org/doi/10.1145/3457607">skewed</a>.</li><li><strong>False Confidence</strong>: AI can produce summaries that look authoritative but miss nuance. Oversight is essential.</li><li><strong>Transparency and Defensibility</strong>: Agencies must be able to explain — in plain language — how AI contributed to the analysis. Otherwise, outputs <a href="https://ojs.aaai.org/index.php/aimagazine/article/view/2741">may not hold up under legal scrutiny</a>.</li><li><strong>Human Nuance</strong>: Some comments are deeply personal, telling stories that don’t fit neatly into categories. Reducing those to “themes” risks erasing what makes them powerful.</li></ul><h3><strong>A Model of Co-Creation</strong></h3><p>The best path forward is not automation, but co-creation:</p><ul><li><strong>AI organizes and summarizes</strong> the comments.</li><li><strong>Human analysts review and interpret</strong> the results, ensuring nuance is respected.</li><li><strong>Agencies communicate findings back to the public</strong> in accessible, transparent formats.</li><li><strong>Stakeholders validate outputs</strong>, providing feedback on whether the summaries and themes reflect their input.</li></ul><p>This approach preserves human accountability while making the process faster, clearer, and more participatory.</p><h3><strong>Why This Matters</strong></h3><p>Public comments are one of the few direct ways citizens shape government decisions. But when the process feels like a black box, it erodes trust.</p><p>AI provides a chance to make engagement more meaningful:</p><ul><li>Participants see how their input connects with others.</li><li>Agencies can respond with clarity and specificity.</li><li>Decision-making becomes more transparent.</li></ul><p>Done right, AI could transform public comment from a frustrating exercise into a genuine tool of democracy.</p><h3><strong>Join Conversation</strong></h3><p>I’ve laid out one vision of how AI could reshape public comment analysis. But I want to hear from you:</p><ul><li>For those who’ve worked in <strong>government agencies</strong>: what’s the hardest part of managing public comments today?</li><li>For those in <strong>technology</strong>: what safeguards could make AI outputs reliable enough for public trust?</li><li>For <strong>community members and advocates</strong>: what would make you feel your comment really mattered?</li></ul><p>Public participation is too important to leave stuck in the past. Let’s use the tools of the future — carefully, transparently, and collaboratively — to make it stronger.</p><p>Connect with me on <a href="https://www.linkedin.com/in/sandra-whitehead-4227068/">LinkedIn</a> and let’s chat.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=58fa27c13900" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Using AI to Accelerate Permitting for Energy Generation Projects: Promise, Pitfalls, and a Path…]]></title>
            <link>https://medium.com/@sandrawhitehead/using-ai-to-accelerate-permitting-for-energy-generation-projects-promise-pitfalls-and-a-path-fa7f2238f49f?source=rss-4988eb5fe7e8------2</link>
            <guid isPermaLink="false">https://medium.com/p/fa7f2238f49f</guid>
            <category><![CDATA[renewable-energy]]></category>
            <category><![CDATA[big-data]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[green-energy]]></category>
            <category><![CDATA[solar-energy]]></category>
            <dc:creator><![CDATA[Dr. Sandra Whitehead]]></dc:creator>
            <pubDate>Tue, 23 Sep 2025 13:11:11 GMT</pubDate>
            <atom:updated>2025-09-23T13:11:11.179Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*iGH87hiHynuWOLJmxwiD6w.png" /></figure><h3>Using AI to Accelerate Permitting for Energy Generation Projects: Promise, Pitfalls, and a Path Forward</h3><p>When Energy Secretary David Wright recently described permitting reform as “the biggest remaining thing” in the U.S. energy agenda, he was pointing to a bottleneck that many stakeholders — utilities, developers, regulators, and the communities where projects are proposed — recognize as a persistent barrier to progress.</p><p>Energy generation projects — whether wind farms, solar arrays, geothermal plants, or grid-modernization infrastructure — are slowed not by a lack of technical feasibility, but by the complexity of permitting. We know how to generate reliable, cleaner power. What takes years is securing the approvals to build.</p><p>The permitting challenge is not new. Scholars like Rossi and Sovacool (2021) have traced inefficiencies back decades. But as energy infrastructure grows in scale and urgency, permitting has shifted from being a bureaucratic hurdle to a major constraint on progress.</p><p>And this is where artificial intelligence — if applied thoughtfully — can help.</p><h3>The Problem: Permitting as the Slow Lane</h3><p>Large-scale energy generation projects can take 5–10 years just to move through approvals. Federal reviews, state siting boards, local zoning hearings, and interconnection studies each play a role — but together they create a maze where projects stall.</p><p>An interconnection study, for example, is a technical review conducted by utilities or grid operators to determine whether a new power plant — say, a solar farm — can safely connect to the electric grid. These studies evaluate how the added generation will affect grid stability, transmission capacity, and reliability. While essential, they are resource-intensive and can add years to a project’s timeline.</p><p>So the problem is not ambition. The problem is throughput.</p><h3>Why AI Could Help</h3><p>Permitting is, at its core, an information challenge:</p><ul><li><strong>Collecting</strong> data (environmental baselines, cultural resource surveys, community feedback).</li><li><strong>Synthesizing</strong> that data into coherent impact assessments.</li><li><strong>Analyzing</strong> tradeoffs across health, cost, and resilience.</li><li><strong>Communicating</strong> decisions transparently to stakeholders.</li></ul><p>AI tools can reduce bottlenecks in each of these domains.</p><p><strong>Environmental Data Processing</strong><br> Satellite imagery, sensor networks, and other big data sources generate terabytes of information. Machine learning can detect habitat fragmentation, estimate pollution impacts, and forecast changes faster than manual review.</p><p><strong>Document Review and Drafting</strong><br> Environmental Impact Statements (EISs) often run over 1,000 pages. AI can summarize public comments, flag regulatory inconsistencies, or generate draft sections for staff review — accelerating the paperwork without cutting corners.</p><p><strong>Scenario Modeling</strong><br> AI-enhanced models can forecast grid stability under various project build-outs or simulate long-term impacts of multiple facilities, helping decision-makers compare alternatives.</p><p><strong>Stakeholder Engagement</strong><br> Permitting involves multiple perspectives, from technical experts to community members. AI can assist by identifying recurring themes in meeting transcripts, sorting public comments, and providing multilingual support for outreach materials. These tools don’t replace direct dialogue, but they can help agencies manage the volume of input more efficiently and ensure key issues are recognized early.</p><h3>The Academic Lens: Risks and Realities</h3><p>The literature is clear that AI is no panacea.</p><ul><li><strong>Bias</strong>: Models trained on incomplete or skewed data can underestimate important risks (Benjamin, 2019).</li><li><strong>Legitimacy</strong>: Communities already skeptical of permitting agencies may resist decisions that appear to come from algorithms. Procedural fairness — being able to understand and question the process — is as important as outcomes (Schlosberg, 2007).</li><li><strong>Legal Defensibility</strong>: Courts require transparent administrative records. AI-assisted analysis must be explainable, not a black box.</li></ul><p>So the challenge is not whether AI can speed things up — it can. The challenge is whether it can do so while maintaining trust and legal durability.</p><h3>A Hybrid Model: Human + AI</h3><p>The most promising approach is not automation, but co-creation.</p><ul><li><strong>AI drafts; humans refine.</strong></li><li><strong>AI identifies potential impacts; practitioners ground-truth the findings.</strong></li><li><strong>AI models alternatives; policymakers and stakeholders weigh tradeoffs.</strong></li></ul><p>This model reduces bottlenecks while preserving accountability and shared ownership of the process.</p><h3>Political Realities</h3><p>Permitting reform is politically contentious. Some stakeholders emphasize speed to enable energy security and competitiveness; others emphasize robust safeguards to protect communities and ecosystems. The result is often gridlock.</p><p>AI offers a potential reframing: faster permitting does not have to mean weaker review. If AI can make reviews both faster and stronger, it opens the possibility of moving past the binary “speed versus safeguards” debate.</p><h3>Practical Steps to Get There</h3><p><strong>Pilot Projects</strong>: Agencies should fund small-scale pilots in defined permitting tasks, with rigorous evaluation of accuracy and public acceptance.</p><p><strong>Standards and Transparency</strong>: Like any NEPA methodology, AI use should be disclosed — what data, what model, what limitations.</p><p><strong>Capacity Building</strong>: Local governments and tribal nations need resources and training to interrogate AI outputs.</p><p><strong>Ethical Guardrails</strong>: Use frameworks like NIST’s AI Risk Management Framework to ensure responsible deployment.</p><p><strong>Collaborative Governance</strong>: Engage planners, technologists, community leaders, and developers in co-designing AI-assisted permitting processes.</p><h3>Why This Matters</h3><p>Every year of permitting delay locks in dependence on older infrastructure, increases costs, and slows innovation. But every shortcut risks weakening safeguards that protect people and natural systems.</p><p>AI provides a possible way to move at the speed of energy transition while upholding transparency and trust. It’s not about letting machines make decisions. It’s about enabling people to make better decisions, faster.</p><h3>Invitation to the Conversation</h3><p>I’ve outlined one vision of how AI might accelerate permitting for energy generation projects while maintaining safeguards. But this conversation needs more voices:</p><ul><li><strong>Agency staff</strong>: What specific bottlenecks in permitting could AI realistically help with?</li><li><strong>Community leaders</strong>: What assurances would build your trust in AI-assisted decisions?</li><li><strong>Technologists</strong>: How do we ensure models are explainable enough for courts and stakeholders?</li><li><strong>Policymakers</strong>: What kinds of incentives or standards would help AI serve both efficiency and accountability?</li></ul><p>Permitting reform isn’t just about process — it’s about shaping the energy future we want, and how quickly we can get there. AI is a tool that can help, but only if we shape it together.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=fa7f2238f49f" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>