AI May Be Coming to Take Your Job…to the Next Level

Welcome to NASPO Pulse, your gateway to the latest insights and trends in government procurement. In this blog post, we embark on an enlightening exploration of Artificial Intelligence (AI) and its transformative potential in the procurement landscape. 

AI, a cutting-edge technology, has captured the attention of public agencies seeking to harness its capabilities to optimize procurement processes and drive innovation. Join us as we demystify the essence of AI, and delve into its game-changing applications, shaping the future of procurement in the public sector. (This introduction was written by ChatGPT)[1]

What is AI?

Recent news has focused on new generative AI products or artificial intelligence programs that can create and compose. Large language models like ChatGPT use data that is scraped from millions of websites and continue to recognize and learn patterns from the inputs and prompts users provide. This is not unlike the predictive text feature on a mobile device, just on an exponentially larger scale. Image creators like Midjourney, Dall-E, and Canva generate images based on textual descriptions provided by users by connecting terms from the prompt with features of existing images from a large dataset that has been used to program the content creator. The Washington Post recently published a look behind the curtain at the sources of information used to train generative AI.

Despite its current popularity, AI programs are not a new concept. Machine learning, automated decision systems, and big data analytics are all AI-related concepts. In essence, AI and the like are terms for data analytics processes that create algorithms and models based on multiple data points. Behind the scenes, neural networks and knowledge graphs allow for relationships to be identified between different data sets or types of information and provide AI systems with the ability to assess the prompt from the user and respond with related information coherently. The same principle allows a search engine to produce useful results, an online marketplace to display products related to the items you are shopping for, and your social media apps to show you content you may be interested in.[2]

While Open AI’s ChatGPT may have kicked off the recent interest in the field, major tech companies like Google, Amazon Web Services, and Microsoft have been working in this area for many years. Remember Microsoft’s Clippy? Once again, AI models are being designed to assist with and enhance work tasks.

AI Enhancements in the Private Sector

In the private sector, AI is used to assist with routine tasks, freeing up resources and increasing efficiency. Accounts payable software provides automated invoice data extraction to speed up payment processing and detect fraud. AI is also used for compliance monitoring and can compare contract terms and payment rates to identify discrepancies and catch duplicate invoices.[3]

Supply chain management teams can map supply chains end-to-end, combined with production data, supplier performance data, geographical data, and other contextual data; this information allows AI to identify conditions that threaten supply chains and predict disruptions before they happen.

AI allows procurement managers to test out new strategies. Current and historical market and purchasing data are used to create simulated environments to test strategic sourcing activities. Procurement teams can gauge the results to determine the risks before committing to action.

Public Use Considerations

Governments are exploring uses for AI in various departments and agencies. The successful incorporation of AI tools in other entities could be a preview of future use by procurement offices. The automation of simple or routine procedures may increase officials’ capacity to focus on more complex or consequential tasks and could be especially valuable to agencies and offices facing staffing shortages. Tennessee has already put this into action under the guidance of an Executive Director of Intelligent Automation.[4] An emerging example of the use of generative AI for these purposes includes the adoption of chatbots that can serve as the first points of contact for communicating with constituents and answering common questions.

The U.S. Census Bureau created Census GPT to assist users with searching the census database using simplified queries. In essence, you can ask it a research question, and it can provide the relevant census data to answer with neighborhood-level precision.

For procurement, some experts make a case for using AI to:

• vet suppliers

• track supplier performance

• identify problematic suppliers

• detect potential fraud

AI algorithms and models can be used to predict future spending patterns and purchasing needs for government entities. They can be embedded into eProcurement platforms to search punchout catalogs and suggest the most cost-effective solutions. This particular use, as with most algorithms, comes with some risks like incomplete data and bias in algorithms which could unintentionally disadvantage some suppliers. Such patterns could take significant time to identify.

The U.S. General Services Administration (GSA) is using AI in a variety of ways, from determining the responsiveness of proposals against RFP requirements to coding qualitative feedback from survey comments and call center transcripts to identify patterns and potential issues.[5] The National Artificial Intelligence Initiative Office website lists all the ways the Federal government is implementing AI technology, along with several other informational resources.

States Taking Actions

Although current and future uses of AI in government can carry immense potential, they also present significant risks. As with any government IT solution, protecting sensitive and personal information takes priority and presents a significant challenge. Many states are taking a cautious approach to how they implement and monitor AI-powered processes.

• Maine – In late June 2023, Maine’s CISO enacted a 6-month ban on the use of generative AI for any state business and on any devices connected to state networks. This measure is intended to provide executive leadership time to assess the risks involved with the new tools and develop plans for mitigation and safe usage.

  “Although these systems have many benefits, the expansive nature of this technology introduces a wide array of security, privacy, algorithmic bias, and trustworthiness risks into an already complex IT landscape.”[7]

  Agencies may continue to use previously approved chatbots, and waivers of the ban can be awarded on a case-by-case basis.  You can read more about this and other measures in this article from Government Technology.

• Connecticut – In an effort to increase transparency and mitigate potential risks involved in the use of algorithms and AI in government decision making, Connecticut passed a law establishing a legislative working group tasked with recommending policies for the regulation of AI use in government functions. The bill was informed by a study from Yale Law School’s Media Freedom and Information Access Clinic and the White House Office of Science and Technology Policy’s proposed AI Bill of Rights.[6] This action also requires the Dept. of Administrative Services to monitor and measure the impacts of current and future AI use.

To Learn More

Take a closer look at what states are doing to govern AI, visit the National Conference of State Legislatures’ AI Legislation Tracker.

For more information on the potential roles and risks of AI in government procurement, read this primer from New York University.

For more about procuring AI technology for government use, check out this toolkit from the World Economic Forum.

[1] ChatGPT by OpenAI, July 3, 2023, https://openai.com/blog/chatgpt.

[2] “Knowledge Graphs,” The Alan Turing Institute, accessed July 7, 2023, https://www.turing.ac.uk/research/interest-groups/knowledge-graphs.

[3] “The AI Opportunity in Sourcing and Procurement: Opportunities in the Market Today,” Deloitte, accessed July 7, 2023, https://www2.deloitte.com/content/dam/Deloitte/ca/Documents/deloitte-analytics/ca-en-omniaai-supplychain-pov-aoda.pdf.

[4] Route 50, “The New Government Workplace: The innovative strategies and technologies helping public agencies work more efficiently,” Q2, 2023.

[5] “Ai Inventory,” AI Inventory – Tech at GSA, accessed July 7, 2023, https://tech.gsa.gov/ai-inventory/.

[6] Chris Teale, “Governor Signs Bill Paving the Way for State-Level AI Bill of Rights,” GCN, accessed June 13, 2023, https://gcn.com/emerging-tech/2023/06/governor-signs-bill-paving-way-state-level-ai-bill-rights/387472/.

[7] “State of Maine Department of Administrative and Financial Services Maine Information Technology (MaineIT) Cybersecurity Directive 2023-03,” State of Maine, accessed July 9, 2023, https://www.maine.gov/oit/sites/maine.gov.oit/files/inline-files/FINAL-Cybersecurity-Directive23-03Artificial-Intelligence.pdf.