Not only can companies benefit from the opportunities AI presents, but institutions – governmental, municipal, or public in general – can also experience significant advantages. Properly implemented AI tools can enhance the quality of institutional operations and better serve the community. At the same time, institutions face unique challenges that must be considered when thinking about AI implementation.
I’ve detailed the process of implementing AI tools in companies and institutions in a separate article (you’ll find the link at the end of this text). Here, I will focus on the potential benefits and challenges from an institutional perspective. I will also highlight the use of large language models (LLMs), such as ChatGPT. Remember, however, that AI is a broad topic: specialized models can manage urban infrastructure and traffic, assist with urban planning, analyze large datasets, help prevent crises, and aid in management when a disaster unfortunately strikes, among other things.
AI Perks for Government Institutions (incl. Local)
Just as AI is beneficial for most knowledge-based professions – especially those involving computer-based tasks – it can also be highly valuable for various types of administrative work.
One advantage public institutions have is that much of their output is… public. This means it’s not confidential, and I personally see no reason why it couldn’t be processed by large language models. Thus, based on laws, regulations, decrees, and other legal acts, we can carry out various tasks while leveraging AI. Of course, we must be mindful of potential AI hallucinations and use prompt engineering techniques to ensure the highest quality of responses.
By using legal acts, we can, for instance:
- Search for relevant provisions in legal texts pertaining to a specific topic;
- Interpret a given situation according to the regulations, requesting specific articles to verify the accuracy of the interpretation;
- Draft a response to received correspondence, which should also be reviewed for correctness;
- Prepare templates for documents that meet the requirements of legal acts.
But we can go a few steps further and, supported by legal acts, use AI to semi-automatically (with human oversight) or fully automatically (more likely in the future than today) generate responses to letters or inquiries. I believe everyone would appreciate a shorter case processing time. However, it’s crucial to note: as a public office, we must ensure AI-supported decisions are clear, consistently good, and offer an easy appeals process, handled solely by a human.
Artificial intelligence interprets text well. Supported by sources of knowledge about the functioning of institutions, it could prepare responses to community inquiries, for example, from residents.
AI can also assist in preparing responses to public information requests, particularly in less complex cases. Furthermore, it could be invaluable in crisis situations where quick action is necessary (though it doesn’t absolve anyone from verifying AI-generated content).
AI Advantages for Every Institution
Artificial Intelligence is becoming synonymous with increased office efficiency. Therefore, many computer-based tasks, when assisted by AI, can be completed faster and/or better, even in institutional settings. However, in this text, I won’t cover examples like email drafting, text translation, social media content creation, or other tasks where AI’s usefulness is evident. Let’s dig deeper!
It’s hard for me to point out an institution or municipal company that couldn’t benefit from AI’s potential. However, this doesn’t mean that the process of implementing artificial intelligence will be trivially simple.
Many institutional decisions are made in public or semi-public forums. As such, they are often recorded and/or transcribed. If a meeting is recorded and the recording isn’t classified, we can create a transcript. If a decision-making meeting occurs online, and we can transcribe it live with author attribution, all the better. From this text record, we can do much, for example:
- Summarize the meeting;
- Prepare a ready-to-send memo;
- Develop an update for the website detailing the decisions made;
- Draft an action plan outline based on the resolutions passed;
- Format the text to comply with legal regulations.
This might not be a perfect example (I see some gaps), but perhaps the idea of using ChatGPT or another language model to check the compliance of resolutions with other regulations could be helpful.
AI can also support the planning of various activities, such as public consultations or academic conferences. A well-crafted prompt can even offer creative solutions to tough problems that many institutions face, like executing projects with a limited budget. In this way, AI can improve the community’s perception of offices or schools, portraying them as more efficient and well-organized.
Regarding PR – quite literally this time – AI is excellent at rephrasing content. Regarding media communication, it can help craft more personalized press releases tailored to specific editorial needs. I believe media monitoring companies will increasingly use large language models.
AI opens new perspectives for media relations activities. For instance, it is possible to create press releases tailored to specific types of media.
I’ve already mentioned automation in the context of local and government institutions, where legal texts are involved. But, of course, the possibilities don’t end there. Institutions typically have many processes that can be streamlined and automated using AI, allowing employees to focus on truly important tasks.
AI-supported automation can, for example, aid in digitizing and describing documents, making their content searchable, and using categories or tags to organize files better.
We might not yet frequently experience well-functioning chatbots (at least I don’t see them often). It’s a shame because it’s already possible to create chatbots that can provide valuable and accurate answers to questions. This requires good tools, knowledge, and effort (especially at the beginning, later for maintenance). The obvious benefit is a reduced number of phone calls and emails. By the way, contact forms and email inboxes can also be integrated with AI!
Ideas for Using AI in Specific Contexts
Let’s move on to more specialized applications – still focusing on large language models like ChatGPT.
I think it won’t be an exaggeration to say that the healthcare system in Poland (and for sure not only here) is overburdened and often inefficient. AI offers a potential long-term improvement. By integrating applications for managing clinics or hospitals with LLMs, we could streamline medical documentation. It could be better written and more accessible to patients.
The future, where a significant part of medical documentation work is done by AI and done well, sounds fascinating! I’m rooting for it so that doctors can have more time for patients.
In education, large language models – mainly thanks to students 🙂 – have already found their place. However, teachers, schools, and universities can also benefit from AI applied to the teaching process (and not only there). Consider creating lesson content or assisting in grading assignments.
In court systems, LLMs could help with processing and generating legal documentation. Expecting judges to fully comprehend thousands of pages of case files is very optimistic. Meanwhile, large language models could help create case summaries and extract information necessary for verdicts (with files needing to be in digital form). Naturally, AI could assist in drafting verdict justifications, processing trial transcripts, and handling other administrative tasks.
I deeply believe that I will live to see the day when law is algorithmically organized, making it easier for AI to interpret. This would significantly increase the possibility of quickly resolving court cases, especially those where evaluating motivation or empathy does not play a crucial role.
All the applications described above are just the tip of the iceberg. We can approach AI very creatively, which I try to showcase in training sessions by preparing examples tailored to my audience groups. Training sessions are, therefore, an opportunity for further inspiration.
Now, let’s discuss the potential challenges that institutions considering AI might face.
AI in Institutions Isn’t Just a Rosy Prospect
So far, I’ve devoted quite a bit of space to the positives of the public nature of institutional work. But just as the law mandates making many institutional outputs public, it also introduces many restrictions that can affect AI implementation.
I mentioned the possibility of semi-automated decision-making. I’m not going to dismiss that option – don’t worry. But it might be harder to execute because even if the cases we use AI for are simple, we might not be able to share personal data or case details with commercial AI models. In such cases, an AI model installed on our infrastructure, like an open-source model, may help. However, this typically involves significantly higher costs for deployment and maintenance.
Speaking of costs, AI is still a relatively new technology, and it takes courage and determination to secure funding for such initiatives. I personally advocate for institutions to start with a pilot phase to show initial results. This phase can also help alleviate employee concerns. There are more benefits, and I once again refer to my text on transformation.
Implementing AI requires a certain amount of courage. However, I am convinced that in the long run, the result will be better service for the communities that the institution serves.
Copyright issues regarding AI-generated content could also be somewhat problematic. I have minor concerns here, but it’s something to keep in mind, especially when much institutional work ends up in the public domain.
AI will undoubtedly impact the job market, linking closely with the trend of lifelong learning – continuously updating skills. Support for learning AI may become a critical task for institutions like employment offices/job centres. I dare say that soon, a lack of AI skills in office work could be as exclusionary as not knowing how to use a computer or the internet, which once meant digital exclusion. The sooner institutions acquire the necessary skills, the easier it will be for them to help their communities.
AI will transform the way many institutions operate – it’s only a matter of time. However, what comes out of it will largely be our decision.
I am a strong supporter of Poland. I believe in the spirit of our entrepreneurship and innovation. Time and again, we have proven our ability to thrive in difficult situations. When it comes to implementing AI in institutions, I hope there will be bold decision-makers ready to embrace the future – whether in the work of local governments, state institutions, companies serving local communities, or educational institutions. Improved public service will benefit all of us, and I believe AI can be a valuable aid in this regard.
I invite you to read the article on digital transformation toward AI. I also offer my services for training sessions regarding AI and for the participation of my company, Oxido, in transformational projects.
