Artificial intelligence is here – how can we adapt?

Just six years from now, the skills we need to do our jobs will change by 68%, according to data collected by social networking site LinkedIn.

"Even if you're not changing jobs, your job is going to change on you," the LinkedIn analysts predict, while underscoring "the enduring need for human skills."

The reason? Artificial intelligence.

Whether artificial intelligence, or AI for short, is already part of your job, whether you have dabbled in the technology or have chosen to ignore it altogether, the trend "is accelerating faster than we as individuals can absorb it," says Arthur "Barney" Maccabe, executive director of the University of Arizona Institute for Computation & Data-Enabled Insight, or ICDI. 

Image

"To advance the U of A as an AI university, we need to get staff much more involved in using these technologies, in addition to the ways students learn and faculty use AI in their research and in the classroom," Maccabe said.

"I know people are worried about AI taking their jobs," said Darcy Van Patten, chief technology officer with University Information Technology Services. "I like to say, 'AI is not going to take your jobs, but people who know how to use AI effectively are going to be more competitive in the job market.'"

This translates into a need to enable AI features for employees, particularly those who work in some kind of knowledge-based role, as van Patten puts it. And while educational licenses are available in some cases and prices have and likely will come down, cost is a key consideration. 

Access to AI tools, which often come at an extra cost, is a central issue that the university community is grappling with, Van Patten said. 

Image

"Of course we would like to buy licenses for everyone on campus. But if you do the math, it will become very expensive," Van Patten said. "We're talking millions of dollars a year." 

How AI can make work life easier

At the same time, the potential of AI-empowered applications streamlining processes is undeniably appealing. UITS experts have been working on a secure AI chatbot for the U of A's instructional platform D2L that would allow students to ask questions about course materials. Similarly, the team is currently experimenting with embedding chatbots within various complex university websites that house a lot of policies and procedures. Employees could use the chatbot to ask specific questions and easily find relevant information specific for their situation. 

Other opportunities abound.

"Think about our university policies website, for example," Van Patten said. "We have so many policies, and people don't necessarily know about them all, or what they're supposed to do and not to do. Imagine they could ask an AI-driven chatbot to help them find the right policy very quickly." 

At the institutional level, AI products have to pass muster in terms of data security and affordability, Van Patten said. AI features are currently available in the conferencing platform Zoom, and they will soon be enabled for the content management and sharing application Box, which will help users with a variety of tasks, such as providing an outline for a slide deck or a short news blurb based on a Word document. 

"In platforms where AI features are already baked into the base cost, we are evaluating whether to turn them on," she said. "For us, the primary point of evaluation is, does use of AI in this context keep our information proprietary, or does it feed it into a public large language model?"

Get over that hurdle

Image

At the pace AI is developing, it's easy to feel overwhelmed, and if you're not sure where and how to start, you're certainly not alone. Across campus, several units have begun offering workshops and resources to help move the university toward responsible and sustainable solutions surrounding artificial intelligence.

"The people who seem to adapt the fastest to AI applications seem to be the ones who are embracing the relationship aspect of AI, and that has certainly worked for me," says ICDI Director for AI and Industry Ash Black. For users new to AI who want to get a feel for what's coming, Black recommends starting with ChatGPT, the large language model developed by OpenAI, and "befriending" it.

"Try asking it a question about something you know lot about," he said. "Ask it to do work for you, and if you're not sure where to start or how to adapt to it, you can actually ask it, 'Hey friend, what's the best thing I can do to adapt to AI?"

Nirav Merchant, who as the director of the U of A Data Science Institute lives and breathes artificial intelligence, urges newcomers to the field to not give in to the hype. 

"If you go on a fishing expedition and say, 'I want to learn AI,' it'll be hard, because there are too many distractions, too many things, and everybody's going to sell you something," he said. "Instead, say, 'I want to be more efficient in my work. Here are the four tasks I want to accomplish, and I will define my efficiency by these three things for each task.' If you embark on this journey of talking to other users, figuring out the product, you're going to learn it much better. Otherwise, it's a lot of spinning your wheels and eventually not getting what you need."

Image

Questions are good, prompts are better

While AI can answer your questions, it really begins to shine with deliberately designed prompts, Merchant said.

"You could say, 'I am trying to write an email with such and such purpose and then you put into an example in parenthesis,' and then you say, 'Here's a copy-and-paste placeholder for the text that should be in the body.' The AI will reformat these elements, put them together and give you an output that sounds more and more like you. The output will become your style, because the system is going to build around that, rather than sounding alien."

He says he typically de-emphasizes the intelligence part of AI and likes to focus more on its information technologies, which are becoming more and more advanced, and how to integrate them into what humans use them for. 

"Who cares if it's intelligence or not," he says. "If it helps me do my job, if it simplifies my email, I think that's a great example of automating things that are very routine."

How to learn more

• Artificial Intelligence at the University of Arizona: This comprehensive resource provides a great starting point. 
• The university's AI Access & Integrity Working Group fosters campuswide conversation focused on access equity and academic integrity surrounding artificial intelligence, with a focus on dialogue that is open to all staff, faculty and students. 
• The U of A DataLab offers almost 45 workshop sessions covering a variety of interests and applications, including a series on Natural Language Processing. They are open at no cost to everyone at the university and include office hours.
• The AI at Arizona website was launched last month to provide a hub for the campus community to collaborate and connect around anything AI. Accessible to anyone with a NetID, the portal also provides access to a Microsoft Teams group hosting regular online workshops to discuss various topics, such as "AI for social good" and "How to build a simple chatbot without coding."
• For instructors who want to make paywalled AI applications available to their students, the Data Science Institute has developed AI-Verde, a platform that allows setting a usage allowance for each student and the course, which is economical compared with paying the full monthly fee for each student. 
• Future Tools: Maintained by AI YouTuber Matt Wolf, this portal contains a searchable smorgasbord of AI tools for all sorts of applications.