Artificial Intelligence Operations

THINK | 2025

Slide
00:00 / 00:00

TxDOT's Blueprint for AI

Building on a storied history of innovation, Texas is harnessing artificial intelligence’s potential to improve safety, efficiency and system performance on the state’s roadways through a highly collaborative, human-centered approach

By Erika Kemp, Director, Strategic Initiatives and Innovation Division | Texas Department of Transportation

Texas has an enduring reputation for thinking big and being big, thanks to its geographic footprint and growing population. Texas also holds claim to the country’s most extensive road network, with more than 80,000 centerline miles on the state highway system. Accordingly, our transportation agency workforce has more than 13,000 professionals committed to enhancing mobility, safety and quality of life for Texans every day.

To pursue its mission, the Texas Department of Transportation (TxDOT) has been entrusted with funding – more than $100 billion over the next 10 years – to operate, preserve and optimize our system. Innovation and continuous improvement are key to delivering measurable value for every invested dollar, which is why our leaders support us in assessing emerging technologies and techniques that improve efficiency and quality.

One high-profile outcome of this mandate was TxDOT’s public release of its first Artificial Intelligence Strategic Plan in December 2024. This 70+ page document includes strategic priorities, hundreds of possible use cases, best practices and recommendations to drive our adoption of AI through mid-2027. This document guides our pursuit of high-impact AI applications that enhance decision-making, streamline processes and unlock deeper data insights – all while keeping humans in the center of the action.

The plan’s release was the culmination of 18 months of collaborative development including the efforts of specialists throughout the agency and the creative ideas of TxDOT employees across the state. Importantly, the AI plan marked a milestone in a journey of innovation that Texas has been embarking on for decades. It sets in motion a broad set of actions that will transform how we run our operations, provide services and accelerate improvements to safety, mobility and quality of life for the state and the nation.

Innovative mindset, collaborative process

TxDOT has long supported transportation innovation. Our legislature has supported the examination of emerging technologies to gauge how they may impact mobility in our state. Early on we analyzed the potential of AI with the help of Ben McCulloch, our team’s strategic data scientist, and fast-tracked our AI strategic plan. We also engaged two critical partners – Anh Selissen, chief information officer, and Darran Anderson, director of strategy and innovation – and together agreed that AI warranted swift, but highly collaborative strategic planning to leverage it for our success. McCulloch was named the project’s champion and led the effort.

Around this same time, the Information Technology Division launched an AI Program to support, enable and educate employees on AI and machine learning technologies. Led by Kristina Miller and Michelle Brockdorf, the team developed a comprehensive policy and governance framework to guide responsible AI innovation, and the policy principles were echoed in the strategic plan. The team also began evaluating and implementing foundational technologies to support a growing portfolio of AI use cases, ensuring that TxDOT is well-positioned to harness AI in ways that are secure, scalable and impactful.

We managed the development of our AI strategic plan internally, methodically drawing out the best ideas and inputs from our teams across the state. We conducted workshops with representatives from our 25 districts, which cover a vast array of geographies, population concentrations and transportation priorities. We also engaged with our 34 divisions, which specialize in activities that impact all areas of the state.

These workshops probed for answers to an essential question: “What should TxDOT be doing with AI?” Participants were challenged to identify specific applications – tools, capabilities, applications and time-savers alike – and submit their ideas. Our team reviewed inputs, consolidated similar ideas and themes and created a compendium of more than 200 proposed AI use cases, which comprise the lion’s share of the strategic plan.

Interestingly, most proposed use cases fell into one of two categories:

1) improve process efficiencies, or

2) gain transformative insights.

Focusing for maximum effect

Concurrent with our workshop collaborations across our districts, our core team developed the body of the strategic plan, which involved formalizing our guiding principles for AI use, identifying the assets and knowledge already at our disposal and envisioning the future state of TxDOT after AI implementation. Our risk management lead and IT Division were instrumental in framing our approach to AI governance, which our technology team developed into an Acceptable Use Policy. These workstreams resulted in one cogent AI development reference for all employees and stakeholders.

AI has the potential to revolutionize virtually every area of our agency, so we have articulated Five Focus Areas to guide TxDOT’s AI investments and adoption in the coming years:

  1. Optimizing infrastructure
    AI has the potential to improve the efficiency and resiliency of the state’s transportation network. This encompasses design and engineering, construction and maintenance of facilities, traffic management and operations and many other activities that enhance infrastructure performance. Such optimization will ultimately serve the traveling public, which will benefit from safer and more responsive intelligent transportation systems.
  2. Data-informed decision making
    AI-enabled analytics will have a big impact on how staff members make decisions involving roadway operations and maintenance. On the front end, too, data-informed decisions can strengthen our agency’s planning process to uncover and address critical factors. AI needs high-quality data to deliver high-quality results. We are fortunate in this regard because, well before our AI planning began, we were working to standardize our enterprise data platform to improve interoperability and provide cleaner data for digital analysis. We have a cloud-based solution for our enterprise data platform, Snowflake, where we are compiling many of our data sets for our machine learning algorithms to ingest.
  3. Enhancing stakeholder experience
    TxDOT will harness AI to create real-world solutions to help our employees and partners deliver better services to the traveling public. We know that our agency cannot deliver what the public needs without strong partners to help us engineer, build and maintain our transportation infrastructure. One of our use cases for AI involves checking contractor invoices for accuracy. Our goal is to increase payment efficiency. With AI’s help we will strengthen relationships with contractors and suppliers and solidify our standing as a top-tier business partner.
  4. Unlocking workforce potential
    We envision that AI will reduce demands that routine, manual tasks place on employees, freeing them to perform higher-value work and drive innovation. With AI tackling repetitive tasks, TxDOT employees can focus on setting priorities, coordinating work efficiently and communicating proactively with stakeholders. Our intent is to amplify people’s impact so they can work on what is most important.
  5. Ensuring security and privacy
    As AI expands our capacity for utilizing data, robust security and data privacy safeguards to protect the agency and our stakeholders become paramount. Our cybersecurity experts incorporate gold-standard privacy and cybersecurity measures while we consider an additional major factor: Humans. When we shared our draft plan with agency leaders last year, they offered many helpful suggestions. But, one resounding imperative, across the board, was to always have a human in the loop. This precept is visible throughout the strategic plan and is central to our approach as we move forward.
Pricing
Use Cases
Deploying AI
Twenty-two of the 230 use cases in the AI strategic plan are active at this time. Seven of these cases were already incorporated into existing pilots or were initiated upon publication of the plan. This is an important point, because our goal is to facilitate and guide – not impede – the adoption of AI to get results. The ideal way to do this is to connect our AI strategy with the natural evolution of solutions that people are interested in pursuing, because their everyday experience illuminates pathways to producing impact.
Here are three representative use cases that are currently being implemented:

Predicting Battery Failure

Our Fleet Division is using Geotab vehicle trackers in combination with the Snowflake cloud data platform to predict vehicle battery failures across the fleet. The process uses AI to predict – with 96% accuracy – when a battery is likely to fail, two months in advance. This is a tier of accuracy that wouldn’t have been possible before AI’s statistical sophistication. This new utility was prompted by a maintenance supervisor in one of our focus groups, who hopes to get early failure warnings on a range of vehicle systems, to minimize downtime and extend lifespans.

Reviewing Engineering Invoices

Our agency receives a constant flow of invoices for engineering work that can be hundreds of pages long, with complicated approval routing. Up until now people have laboriously gone through these documents line-by-line to confirm accuracy. Our Transportation Programs Division conducted a proof of concept that used AI to streamline line-item verification and found that this significantly reduced turn-around time. Our next step is to pursue full-scale development and implementation.

Detecting Roadway Incidents

We have a pilot program in Austin that uses AI to identify incidents more quickly to improve safety and optimize traffic flow. This pilot involves a traffic management platform that gets data from agency road sources (ITS cameras and sensor data), as well as external data sources from navigation solutions and telematics, to create a rapid, holistic view of roadway traffic patterns. AI machine learning technology normalizes disparate data sets to enhance incident detection—accelerating issue responses and resolutions.

Moving forward

All of us at TxDOT are looking forward to embracing AI – responsibly, safely and with a human touch. We understand this is only the beginning of our journey and unforeseeable opportunities and risks are likely to emerge as AI tools and infrastructure mature to influence more aspects of our operations and culture.

We intend to stay focused on our major priorities, collaborating across our agency and with our many stakeholders, to derive the greatest value from AI for our agency and the public. We invite our colleagues in other state DOTs to explore our Artificial Intelligence Strategic Plan and to apply relevant ideas and process details to their own explorations of AI.

ABOUT THE AUTHOR

Erika Kemp
Strategic Initiatives and Innovation Division Director
Texas Department of Transportation

Erika Kemp is director of the Strategic Initiatives and Innovation Division for the Texas Department of Transportation (TxDOT).

Erika leads the agency’s enterprise-wide programs focusing on data science, performance management, risk management, continuity of operations, policy development, and governance. She also sponsors collaborative consortia that bring together stakeholders from multiple sectors to facilitate the exchange of best practices and implement innovative initiatives.

Before joining TxDOT, Erika was a senior leader for a leading mobile technology services company, serving as general manager of a two-state region with 400+ employees, and overseeing sales for a nine-state region with 6,000+ employees.

She holds a Bachelor of Business Administration (BBA) from South University and is a certified Project Management Professional (PMP) and Prosci Change Management Practitioner.