Machine learning for tax compliance: Machine learning analyzes real-time input by a taxpayer when they file their declaration and compares their input with data available to the tax agency. The system prompts the taxpayer based on applicable law and policy and provides those recommendations to the taxpayer. This customer-facing system gives people an opportunity to correct their tax return without additional enforcement and increases transparency and trust between the tax agency and its citizens.
AI technology has quickly moved beyond being a tool for generating written content. It is transforming industries with practical, real-world applications that solve everyday problems and streamline operations. From healthcare to logistics, customer service to personal productivity, the possibilities seem endless. As a part of this series, I had the pleasure of interviewing Jennifer Robinson, Global Public Sector Strategic Advisor for data and AI company, SAS, and member of the Town of Cary, NC town council.
Jennifer Robinson is SAS’ Global Public Sector Strategic Advisor, working to help governments maximize their use of data through data integration, data management, analytics and AI. Her career in software development is complemented by the opportunity to serve as a local elected leader for the last 25 years. Jennifer co-wrote the book A Practical Guide to Analytics for Government and is featured in the book Smart Cities, Smart Future. In addition to writing about data-driven governing, she speaks with government leaders about emerging technologies and how to strategically adopt them.
Thank you so much for doing this with us! Before we dig in, our readers would like to get to know you a bit more. Can you tell us a bit about your “backstory”? What led you to this particular career path?
While attending the University of Virginia, I landed a summer job as a ”Clerk Typist” with the Administration for Children and Families (ACF) within the US Department of Health and Human Services. On the first day of the job, my boss changed the course of my life when he assigned me to assist a programmer who was creating a financial system for the administration. Stan taught me how to program. This may not sound like exciting work, but I loved it. I returned every winter and summer break for three years, working on various aspects of the system. This led to being hired after graduation by a software company to work on a financial system for the Environmental Protection Agency. I later moved to North Carolina and opened my own little business doing data modeling for companies that were modernizing their computer systems.
Around that time, I became civically obsessed with my local government in Cary, NC. I attended council meetings, joined an advisory board, routed petitions, wrote op-eds, etc. I was appointed to fill the remaining two years of a vacated seat on the Cary town council in 1999. I thought I would serve for two years with little expectation of winning the election that followed. I planned to go back to my business after graciously accepting defeat. But I won.
I enjoyed serving and being a mom for many years until one day in 2015 I got a call from SAS, the global data and AI company headquartered in Cary. They wanted someone to come on as an industry consultant who could talk about data but also understood local government needs. Since then, I’ve also supported market strategy for the public sector where I follow trends, learn what governments need and position SAS to meet those needs. I’m still doing my part to make Cary an even better place to live, while helping communities around the world use data and AI to do the same. It’s the best of both worlds.
Can you share the most interesting story that happened to you since you began your career?
I don’t know if this is interesting, but it was poignant to me. As a municipal leader, I was asked to meet with a group of three citizens who called themselves “The Watchmen.” Their self-appointed role was to ferret out wasteful spending. The first man pointed to the tennis center as a drain on our budget, but the second man corrected him, noting that the tennis center made money for the town. The second man then asked, “Why are we running an amphitheater?” to which the third man said, “That’s the summer home of the symphony. But why are we in the recycling business?” The first man looked at him and said, “That’s our environmental responsibility!” This was a clear example of citizens’ values differing. One person’s waste is another’s smart investment. Leadership involves balancing the many values and opinions of your citizens and stakeholders. I told the Watchmen to talk among themselves and when they were in agreement, we would meet again. They never asked for a follow-up meeting.
Can you share a story about the funniest mistake you made when you were first starting? Can you tell us what lesson you learned from that?
I was interviewing for a publication internship and was asked what magazine I liked to read and who my favorite contributor was. I picked a writer off the top of my head, then referred to the writer as a female. You can imagine my embarrassment when the interviewer told me that the writer was a man, clearly revealing that I knew little about my “favorite” magazine writer. I learned the importance of being authentic and comfortable with not having all the answers. No one does.
None of us are able to achieve success without some help along the way. Is there a particular person who you are grateful towards who helped get you to where you are? Can you share a story about that?
The first person who hired me at ACF changed the course of my life by giving me the opportunity to learn how to program. As an English major, it was not a career I’d ever considered. The second person was the retired Town Manager for Cary who had gone on to a second career at SAS. He was familiar with my work history and knew the challenge I faced returning to the tech industry after taking a break to raise young children and serve in public office. Tragically, he passed away too young. Before doing so, he recommended me to my future SAS boss without my knowledge. It opened another door for me I never expected, and I’m forever grateful.
Ok, thank you for that. Let’s now jump to the primary focus of our interview.
Which industries do you believe will undergo the most significant transformation due to AI over the next 1 to 3 years?
The public sector will see many AI-driven changes in the near future, and recent research backs that up. According to a public sector productivity survey earlier this year of more than 1,550 government and public health care employees in 26 countries, most respondents believe that AI will have a significant or critical impact in improving productivity within their organization over the next three years. Within the public sector, we should expect to see government fraud fighters combat sophisticated AI-powered fraudsters with their own AI systems. A new research report, Trust and transparency: Combating fraud to maximize public program efficiency, indicates that government organizations are preparing to drastically increase investments in AI and GenAI, including a five-fold increase in the use of large language models, specifically to reduce fraud, waste and abuse.
The financial services industry has been at the forefront of AI and analytics for decades, and we should expect that to continue, from fraud fighting and anti-money laundering, to retail banking and marketing efforts. As AI accuracy improves, I think we’ll see many of the anticipated health care advances become a reality. We’ll see it more in a clinical setting, such as doctors using AI to summarize patient interactions or automating mundane tasks that will free up more time for meaningful patient interactions. We’re already seeing promising results in radiology where AI has shown remarkable accuracy in identifying cancer in images. Of course, health care is an industry in which keeping a human-in-the-loop will always be crucial.
All industries need to keep an eye on quantum AI. Compute times will plummet. I had an enlightening conversation with an executive at one of our large consumer goods customers. He believes it won’t be cost that keeps us from maximizing quantum; it will be a lack of imagination. The questions we COULD ask are currently unfathomable to us. Pretty wild stuff.
In your view, which aspects of the consumer experience — whether in retail, travel, healthcare, or other sectors — are poised to look dramatically different because of AI? And what do you think will remain largely unchanged?
I think anything involving important human interactions will be largely unchanged. Doctors and nurses will still need to be there for patients. I know there are AI “counselors” out there, but the relationships between patients and their mental health providers are too important to turn over to technology. AI can help with record keeping to free up time, but the core interactions will stay the same. Trades in which things can only be done with human hands, like auto mechanics, electricians, plumbers, etc. won’t change, though the back-office operations for all these professions will benefit from AI.
Tremendous changes are coming to anything involving the collection or dissemination of information. AI will optimize data collection itself, recommending which data is most important for decision making. In the public sector, this improved collection and integration will enable citizens to get real-time answers through digital portals, with the system anticipating many of their needs. A permit app, for instance, could guide a person through the information gathering process, prompting them when what they have provided is insufficient and giving them the best chance for an approved permit.
We’re already seeing AI applied to the analysis of public commentary to understand citizen and stakeholder sentiment. For example, Southern States Energy Board, an interstate compact composed of governors and state legislators from sixteen southern states, needed to make sense of 40,000 public responses to the potential transfer of Environmental Protection Agency regulatory oversight of CO2 injection wells to the state of Louisiana.
To enhance the efficiency of processing public comments, a multimodal strategy was implemented that combines traditional AI and GenAI. SSEB used visual text analytics alongside an LLM to streamline the process by automatically matching unique statements to specific regulatory aspects.
Traditional AI serves as a prefilter, organizing the extensive feedback into manageable categories. This foundational step allows generative AI to take over, providing concise summary statements for each category. Human reviewers then fine-tune these summaries, ensuring accuracy and clarity. This integrated approach not only accelerates the analysis of large volumes of comments but also enhances the overall quality of insights, empowering teams to navigate complex regulations more effectively. Then, using data visualization, the team created an easy-to-use dashboard in which stakeholders could examine a word cloud of more than 10,000 unique terms the software identified from concrete recommendations from organizations. For the dashboard, generative AI created and included summary statements, which users of the system could verify from the data provided by the text analytics, for each aspect of the regulation.
Government agencies are required by law to respond in some fashion to all citizen feedback to proposals. AI will be a game-changer for that process.
For companies aiming to develop consumer-facing AI tools, what is the most important piece of advice you’d offer?
Different industries will have different risk tolerances for what they operationalize in the world. If a retailer’s new algorithm makes bad style recommendations, nobody’s getting hurt. (Some fashion-forward people may disagree.) But if a faulty banking algorithm declines loan applications, someone’s financial security is jeopardized. These concerns are particularly high for government agencies, where benefits, taxes, health care and other serious issues are in the balance. Higher risk industries must carefully vet a system before it is exposed to citizens.
If a company is developing AI tools, they need to consider how those tools could show up in high impact industries. Many of the new AI regulations take this approach of impact or risk stratification. By preparing for the more stringent of those requirements, developers can not only help mitigate their own compliance risks, but help their customers do the same.
They can also empower their customers, like SAS has done, with built-in trustworthy AI capabilities like bias detection, explainability, decision auditability and model monitoring, governance and accountability. Model cards, which are like nutritional labels for models, highlight indicators like accuracy, fairness and model drift, which is the decay of model performance as conditions change. Additionally, helping customers understand their AI governance maturity, and a path forward, will be a major organizational advantage for them, while mitigating risks.
Are there any AI use cases you’re currently exploring — or excited about — that haven’t yet attracted much attention but have the potential to reshape how businesses function or how we live our everyday lives?
It may sound strange to be excited about dusty filing cabinets and old government reports, but I am! Government agencies have vast amounts of information gathered and stored in paper and images, with records dating back more than a century. The key to unlocking the value of that data is intelligent document processing, which uses composite AI systems, combining things like Computer Vision, Optical Recognition Models and Large Language Models to automate the “reading” and processing of scanned documents. It can extract structured contextual information from scanned document images to be used in analytics and decision making. That trove of newly digitalized content can also be used to train LLMs.
This is happening today. As an example, SAS engaged with a large US health provider to use document processing to enhance, optimize, and modernize the customer’s existing medical record review process. The existing process required nurses and medical coding specialists to manually review thousands of pages of patient documentation to decide on coverage eligibility. The technology helped automate the processing of these lengthy service claims and surface the relevant information for the reviewer to then assess and confirm. It now processes upwards of 10 million pages of documents per day.
The technology is sophisticated enough to decipher poor quality documents, copies of copies of copies and even handwritten doctor’s notes from WW2 medical records. It’s an innovation that’s emerging now, serving as an additional tool to be plugged into existing business processes. It could truly be the next great AI advancement by, ironically, looking backwards, unlocking value that’s been trapped since before the dawn of computing.
I also think AI agents and agentic AI (two different things!) will be transformative. AI agents are specific, task-oriented AI models designed to perform repetitive tasks on behalf of a user, whereas agentic AI is a broader framework that uses multiple AI agents to achieve complex goals autonomously. Both will redefine how organizations operate, make decisions and interact with technology. Governments will benefit from the productivity gains realized when rote, menial, time-consuming and complex tasks are automated. And as I mentioned, quantum computing will change our ability to analyze vast amounts of data quickly and accelerate the time to insights in ways we can’t even imagine.
How do you think about balancing AI-driven automation with the human element? In which areas is it essential to keep people meaningfully involved?
Tasks that require creativity, empathy and judgment should be performed by humans. As I mentioned, high impact and high-risk applications of AI will require more rigorous oversight, always with a human-in-the-loop. Tasks that involve huge datasets, repetitive tasks and complex calculations are good for machines. Low impact, highly repetitive, objective applications can be carried out with lighter oversight.
We must remember that it’s called artificial intelligence because it is the simulation of human intelligence in machines that are programmed to learn and think. These technologies are meant to augment the work we do, making it easier and faster to produce better outcomes for the people we serve. As governments deploy AI to perform more tasks, it will be imperative that the risks and concerns with autonomy are acknowledged and addressed. Responsible data practices and model management will be essential.
Can you share a surprising or particularly innovative everyday use of AI you’ve encountered that most people wouldn’t expect?
It’s not the most uplifting example, but the use of AI to combat child exploitation is showing great promise. From the minute a predator contacts a child online, every moment counts. To prevent these predators from meeting and harming children in real life, social media companies flag concerning profiles or messages — and then it’s in law enforcement’s hands.
Unfortunately, the individual review of profiles and chat logs is an inexact and arduous process that can create backlogs of flagged cases and delay intervention. Traditionally, it has taken a human to determine whether content is just disturbing talk, or if real plans to harm a child are in motion. That’s where specially trained, AI-powered language models can make a real difference.
These models can detect plans for real-world harm before they can transpire. This speeds up response times with real-time detection of grooming and exploitation attempts, enabling immediate intervention. The models can account for attempts to obfuscate meaning or hide intentions through advanced pattern analysis. This also reduces the mental strain on investigators as reviewing these intense materials day-to-day can become emotionally exhausted and burn out earlier in their careers, further reducing the number of people on hand to screen this kind of evidence.
Child exploitation is a horrible, global problem that is too large and complex for humans to handle alone. I am hopeful about AI’s ability to help combat it.
What is one common misconception people have about integrating AI into daily life, and how would you clarify or correct it?
Many people fear AI will force them to learn a difficult system and drastically change how they do their jobs. Fortunately, AI tools providers are not just creating more sophisticated applications, they’re also figuring out how to help users integrate it into their workflows. For example, Microsoft Copilot’s prompt library anticipates most of the common applications users need and provides great jumping off points for many tasks. If done right, AI tools should remove complexities, not add them, and make your job easier and faster.
Ok super. Here is the main question of our interview. Can you please share “Five Practical Everyday Use Cases for AI”?
I’ve already shared some throughout but here are a few more interesting public sector ones, and a bonus one from the Great Toilet Paper Shortage of 2020.
1. Machine learning for tax compliance: Machine learning analyzes real-time input by a taxpayer when they file their declaration and compares their input with data available to the tax agency. The system prompts the taxpayer based on applicable law and policy and provides those recommendations to the taxpayer. This customer-facing system gives people an opportunity to correct their tax return without additional enforcement and increases transparency and trust between the tax agency and its citizens.
2. Using Computer Vision to prepare for disasters and address wildfire risk in real-time: The use of aerial imagery to create shape files in determining fire boundaries, when done so in succession, determines a fire’s movement thereby informing first responders on faster, more accurate responses to prevent loss of life and destruction of property. Using Computer Vision to process imagery data automatically saves time and the manual intervention of a physical person drawing a fire boundary by hand once the image is received on the ground. This allows fire fighters to evaluate wildfires more efficiently and at a safe distance, in real-time, without traversing hazardous terrain, and putting first responders and equipment at risk. First responders can protect property better and reduce potential loss of life / injury. Side note, a similar approach has been used to uncover evidence of deforestation in the Amazon rain forest.
3. AI for property value assessments: Property values in the Triangle area of NC, which includes Wake County, have skyrocketed in the last 10 years. We worked with the county to create an AI-powered property valuation tool that provides objective backup to valuations of human assessors. It identifies changing market trends every day for approximately 400,000 properties. The models consider hundreds of factors and daily property sales to offer timely, objective, highly accurate market forecasts. It’s a great example of AI helping people do their jobs faster, better and easier and has led to fewer disputes from citizens.
4. Synthetic data to maximize college financial aid: Synthetic data is used by universities to fuel a machine learning model that optimizes the pool of financial aid funds, total enrollment, and net tuition revenue. Synthetic data is created using algorithms that generate student data that mimics the data of real students so that what-if scenarios can be run by machine learning models without compromising the privacy of students. This is leading to greater student retention, higher enrollment, money saved and higher net tuition revenue.
5. Digital twins support flood prediction and preparation: Governments are creating digital twins of their communities and water ways to predict floods, enabling them to anticipate needs and allocate resources. Machine learning models support real-time situational awareness, automated alerting, historical and forensic analysis, forecasted flood inundation modeling, and simulations for emergency planners, improving modeling for various disasters. While the model has been developed using historical data (I.e., past events and where data were prevalent), this model can be deployed where hyper localized data is scarce. The model, powered by AI, fills these data gaps.
BONUS EXAMPLE: Vibrations sensors and thousands of models keep Georgia-Pacific up and running: On everything from plywood to corrugated boxes and paper napkins, Georgia-Pacific runs more than 15,000 models to calculate the optimal production settings based on up-to-the-minute business needs. Using real-time data from 85,000 vibration sensors, the team can intervene early when their calculations predict an increased likelihood of part failure or electrical problems. By combining this information with historical data on asset performance, they can deliver insights to machine operators on what to change to optimize the life of the asset. During the COVID toilet paper shortage, Georgia-Pacific saw a 120 percent increase in demand for toilet paper, tissues and other products. At the same time, there was a breakdown in the global supply chain. Georgia-Pacific reduced unplanned downtime by 30 percent and ultimately improved equipment efficiency by 10 percent to get more products into stores — faster. That led to lower labor costs for maintenance, less scrap and waste, and increased production.
Is there an AI tool or app you personally use and love?
I’m not just plugging this because of the partnership between SAS and Microsoft but I do really love Copilot. As a content creator, I find it invaluable for getting me started on presentations and articles, as well as sparking creative ideas. I have access to the enterprise version, which can comb SAS internal content in addition to all the external Web content it can access. So, its recommendations combine everything I could access in my job role. It has boosted my personal productivity in major ways.
You are a person of great influence. If you could inspire a movement that would bring the most amount of good for the greatest number of people, what would that be? You never know what your idea can trigger.
As a local leader, I often reflect on the growing crisis of plastic waste. This challenge is too large for individuals alone — it demands bold action from private companies, governments, and service providers alike. I would like companies to move to producing products and packaging that are biodegradable or more easily recyclable. And, as an early step, I would like more governments to join the ranks of others that incent recycling with deposit-return programs. But perhaps the most immediate and transformative impact could come from local governments and private waste companies. By deploying small-scale recycling technologies — like shredders, granulators, and extruders — near the point of collection, we can dramatically reduce the volume of plastic entering our landfills and oceans.
We are very blessed that some very prominent names in Business, VC funding, Sports, and Entertainment read this column. Is there a person in the world, or in the US with whom you would love to have a private breakfast or lunch with, and why? He or she might just see this if we tag them.
Having spent most of my adult life serving in local government and working with data, I’d love to have lunch with Michael Bloomberg. He has been a pioneer of data-driven government, championing the use of data to improve municipal services. Through Bloomberg Philanthropies, he has funded and supported hundreds of local governments across the world to adopt innovative, data-informed practices. Bloomberg’s leadership blends business acumen with public service, showing how data and analytics can be used not just for efficiency, but for equity and long-term impact. For someone in local government, this is both aspirational and practical.
How can our readers further follow your work online?
Please engage with me on LinkedIn at https://www.linkedin.com/in/jenbrobinson/
Thank you for these fantastic insights. We greatly appreciate the time you spent on this.
About The Interviewer: Cooper Harris is a California-based entrepreneur and the founder and CEO of Klickly, a data-driven, AI platform that powers distributed commerce. Emerging as a pivotal figure in the west coast tech scene, Cooper has won Los Angeles Business Journal “Innovator of the Year,” InformationAge’s Women in I.T. “Entrepreneur of the Year” was nominated for Google’s “Young Innovator” award, L’Oreal’s “Digital Woman of the Year,”and was named by Adobe as a “Top Data Thought-Leader” at Cannes, alongside execs at JP Morgan Chase and Burger King. Harris invented a data technology that leverages AI to facilitate distributed commerce on 25M online destinations. This significant innovation creates a huge efficiency in the market — the platform’s machine-learning identifies best-fit matches for commerce opportunities and uses sophisticated algorithms to understand the consumer to both promote and power transactions.
Cooper has been invited to speak on eCommerce, tech, data, and AI at international summits including the United Nations, CES, Cannes Lions, Money 20/20, SXSW, ShopTalk, Sundance, Los Angeles TechWeek, London Technology Conference, and Jason Calacanis’ LAUNCH Scale. Ms. Harris speaks on topics including data / FinTech / Retail Tech, fostering women in STEM, and disrupting the status quo using technology/innovation. Cooper also writes about eCommerce, innovation, data and AI, and using tech to change the world. She is a contributor to Forbes and HuffPost and has been featured in Inc., Fast-Company, Entrepreneur, Mashable, Women2.0 and more.
