Leo S. Lo and Victoria Anderson

AI Reskilling in Libraries

When the Dean’s Assistant Gets an AI Assistant

Leo S. Lo is dean of the College of University Libraries and Learning Sciences at the University of New Mexico, email: leolo@unm.edu. Victoria Anderson is administrative assistant at the University of New Mexico, email: vbrown34@unm.edu.

©2024 Leo S. Lo and Victoria Anderson

The rapid advancement of generative AI represents a pivotal moment for libraries. With powerful systems like ChatGPT gaining widespread adoption, libraries face both opportunities and an urgent need to reskill staff. Recent analyses by the Boston Consulting Group predict that within two decades more than 30% of activities across most jobs could be handled by machines, while more than 15% of jobs could be fully automated.1 According to the World Economic Forum’s “The Future of Job Report 2023,” training workers to use AI tools is considered one of the top priorities for many companies.2 A 2023 survey of academic library employees revealed enthusiasm for professional development around generative AI, yet fewer than 7% of respondents currently use premium versions of leading systems.3 This discrepancy risks leaving library workers unprepared for AI-assisted workflows and roles forecast to be pervasive within the next decade.4 As libraries undergo their digital transformation, developing AI literacy among staff at all levels is becoming a matter of competitive survival rather than just innovation.

To proactively spearhead closing this emerging literacy and capability gap, the University of New Mexico College of University Libraries and Learning Sciences launched the pioneering GPT-4 Exploration Program in summer 2023. This initiative provided 10 faculty and staff with funded access to GPT-4, the premium iteration of ChatGPT. By facilitating hands-on experience with leading-edge AI, we sought to cultivate deeper technological literacy and equip our community to reenvision library services, research, and instruction for an AI-assisted era.

Crucially, the program spotlighted reskilling as an urgent priority. As AI systems grow more advanced, all library workers must have opportunities to better understand and use these technologies to enrich their careers. This access gap hinders professional capability and innovation. By detailing the dean’s vision in creating this pilot and the firsthand insights of the dean’s assistant, this article illuminates the profound need to expand AI literacy and upskilling programs across libraries. Equipping staff at all levels with premium generative AI will drive progress in realigning skill sets for the future.

Dean’s Perspective on Designing the Pilot Program—Leo S. Lo

In developing the GPT-4 Exploration Program, we turned to proven adult education practices to maximize reskilling outcomes. Aligning with Malcolm Knowles’ principles,5 we emphasized hands-on, self-directed learning driven by professional needs and interests. Rather than lectures on AI, participants chose personalized projects to actively upskill through practice. This immersive approach spurs more active reskilling.

The program comprises three phases: Introduction and Training (two weeks), Exploration and Experimentation (eight weeks), and Evaluation and Sharing (two weeks). This arc builds community and enriches learning. Training provides baseline knowledge before the heart of exploration. Participants then reconvene in reflection, cementing gains through peer sharing. Central to our vision was an interdisciplinary community advancing together. The cohort model breeds collaboration beyond silos, enriching perspective and buy-in. United by curiosity despite divergent backgrounds and skill levels, participants enhanced both their personal capabilities and our college’s broader culture of innovation.

Among the 10 members from varied roles, my assistant Victoria Anderson brought critical administrative representation. Her participation modeled cross-functional reskilling. Beginning as an AI novice, Victoria flourished over 12 weeks, attaining hard skills and perspective to steward administrative adoption of AI systems. Her journey typified the immense reskilling potential in our untapped workforce once given access and community.

Administrative Assistant’s Experience Using AI—Victoria Anderson

As the dean’s assistant, I eagerly joined the GPT-4 Exploration Program despite my limited prior AI experience. Over 12 weeks, integrating ChatGPT and its premium version, GPT-4, fundamentally upgraded my workflow. I honed skills in using prompts to have GPT-4 assist me with a variety of administrative tasks. For clarity and simplicity throughout this piece, I will refer to both versions of the tool collectively as GPT-4.

The Evolving Role of Administrative Support

Administrative assistants and coordinators play an integral role in the operations of any organization. As institutions evolve to meet strategic priorities, so too must administrative support roles. Increasingly, the responsibilities encompass managing schedules, travel logistics, documentation, correspondence, and critical support services for leadership figures like deans or executives. This multifaceted and dynamic position requires adaptability, critical thinking, and efficiency even as responsibilities shift over time.

Integrating AI by administrative personnel can enhance workflows by automatically handling time-intensive coordination tasks and improving productivity in written communications. However, to unlock this potential, universities and companies must invest in reskilling programs tailored to support staff needs in understanding leading-edge technologies like generative AI. Organizations that equip administrative teams with the latest automation tools and skills training will gain tremendous advantage.

Enhancing Daily Workflow with GPT-4 Integration

One major time saver was using GPT-4 to take meeting minutes. By recording meetings, uploading the audio to a transcription tool, then copying the text into GPT-4, I could prompt it to “create detailed meeting minutes in a bulleted format summarizing the key discussion points and action items from this meeting transcript.” This condensed hours of work into a polished 2–3-page summary showcasing the most relevant details and decisions ready for distribution in less than 30 minutes. I estimate this saved me 3–5 hours per week of manual notetaking and write-up.

However, it is important to note that we were careful in considering what information was provided to the system, as we are still learning about how these AI tools use and potentially retain data. For example, I would always redact all personal details, sensitive information, or proprietary organizational data in the transcript before uploading it to GPT-4. By taking steps around responsible data use, libraries can use the immense time-savings of AI tools like GPT-4 safely and ethically.

GPT-4 also streamlined communications. Whether helping craft emails to reschedule a series of complex meetings for the dean or summarizing lengthy documents into digestible briefing papers (reducing a 12-page report to 2 pages), GPT-4 boosted productivity. For example, to condense a long text, I would prompt “Please summarize the key points from this 12-page report in 2 pages in a bulleted format covering the main recommendations.” Additionally, GPT-4’s integration with tools like Kayak’s travel booking functionality automated logistics coordination. By describing the dean’s upcoming conference travel needs in chat, the Kayak bot would automatically find flight options, saving me hours of searching.

I learned prompting is a skill requiring practice—initial outputs might not match the tone or depth needed. However, each week I gained more proficiency in getting useful content quickly. I also heeded guidance from the dean on responsible data use—we ensured no sensitive information was uploaded to GPT-4 given uncertainties around OpenAI’s data policies.

Challenges and Adjustments

While integrating GPT-4 into my workflows yielded significant time savings and productivity gains, the process was not without some needed adjustments. As I was learning how to best prompt the AI, I noticed some key challenges emerge.

GPT-4’s initial responses would sometimes lack the specificity, creativity, or tone that aligned with my communication style and needs. When asking it to generate emails to reschedule meetings, the language was often overly formal or verbose. Additionally, when seeking input on contributing to strategic goals around open education resources, the examples provided did not demonstrate a deep understanding of my administrative coordinator role.

I realized that mastering use of AI requires learning how to properly frame questions and prompt responses. Through trial and error and guidance from the dean on prompt formulation, I was able to improve results over time. However, it became clear that GPT-4 serves more as an augmentation tool, providing supportive content that almost always requires some degree of revision. The back-and-forth to refine responses also introduced some inefficiencies.

To address these challenges, I focused on shifting my mindset when using GPT-4. Rather than expecting fully formed solutions from the AI, I started to view it as a collaborative tool—asking it to “assist me with” tasks rather than relying on it to complete things independently. This helped set proper expectations and in turn improved my prompts and post-processing of its outputs.

While integrating AI did require conquering a learning curve, doing so has unlocked immense potential for upgrading my productivity and contribution. My experience shows that with the right support and training, administrative professionals at all skill levels can overcome initial challenges and thrive with the assistance of leading-edge AI.

Conclusions and Next Steps

The success of our inaugural GPT-4 Exploration Program confirms the immense value of hands-on AI reskilling initiatives. More than 70% of participants achieved expanded fluency with GPT-4, drastically exceeding baseline proficiency. As participants become ambassadors sharing insights with peers, our broader culture of AI readiness strengthens.

Building on this momentum, we will launch an evolved version of the program focused on an alternate leading AI system. Our consistent vision remains experiential, community-based learning in an emergent technology. Enhanced content on ethics and specialized projects tailored to academic applications will enrich the next iteration.

Additionally, we will catalyze wider adoption by collaborating with other campus departments on AI literacy programs. Inspired by Victoria’s transformative journey, we also advocate concentrated reskilling for administrative staff as an immense opportunity. Victoria’s agility in deploying GPT-4 for coordinating events and communications revealed automation’s sizable efficiency potential in critical support roles.

However, her experience also highlighted AI’s practical limitations regarding creativity and judgment. As powerful as these tools are, responsible implementation in both administrative and academic realms requires recognizing AI as an assistive rather than completely autonomous solution. By transparently conveying both potentials and pitfalls, our expanded programs will empower the wider university community to evaluate and integrate AI with care and confidence.

Our pilot program has evidenced the value of equipping learners at all levels with premium access and training for emergent technologies like GPT-4. We urge institutions globally to prioritize on-ramping communities to AI through hands-on, ethically grounded learning programs proven to unlock incredible capability.

Notes

1. Sagar Goel and Orsolya Kovács-Ondrejkovic, “Reskilling for a Rapidly Changing World,” BCG Global, September 22, 2023, https://www.bcg.com/publications/2023
   /reskilling-workforce-for-future.
2. World Economic Forum, “The Future of Jobs Report 2023,” May 2023, https://www.weforum.org/publications/the-future-of-jobs-report-2023/digest/.
3. Leo S. Lo, “Evaluating AI Literacy in Academic Libraries: A Survey Study with a Focus on U.S. Employees,” University of New Mexico Digital Repository, 2024, https://digitalrepository.unm.edu/ulls_fsp/203.
4. Kweilin Ellingrud, Saurabh Sanghvi, Gurneet Singh Dandona, Anu Madgavkar, Michael Chui, Olivia White, and Paige Hasebe, “Generative AI and the Future of Work in America,” McKinsey & Company, July 26, 2023, https://www.mckinsey.com/mgi/our
   -research/generative-ai-and-the-future-of-work-in-america.
5. Malcolm S. Knowles, Elwood F. Holton III, and Richard A. Swanson, The Adult Learner: The Definitive Classic in Adult Education and Human Resource Development (Milton Park, UK: Routledge, 2014).