by Leslie Lytle, Messenger Staff Writer

“AI should help us think better, not think instead of us,” insisted Mel Oliveros who works with school districts throughout Tennessee and the nation helping students learn how to make meaningful use of the AI platform Coursemojo. In her July 23 Monteagle Sunday School Assembly lecture titled “Beyond the Hype: A Practical Look at AI,” Oliveros tackled questions about how AI works, how to make AI work for us, and what to be wary of. Although unaware of it, many people already use AI. The new frontier, the hot potato both celebrated and feared, is generative AI.

Anyone who has used Siri or Alexa, browsed Netflix movie suggestions, taken Google Maps advice on a different route, or relied on email spam filters operating in the background engages with AI. “AI means artificial intelligence,” Oliveros said. “Think about it as a piece of software, but instead of programmers typing up the rules of what it should be doing, AI platforms have been fed an enormous amount of content and information and it learns patterns to figure out what the next thing could be. It learns by example.”

“The main difference in the past couple of years is that AI is now something we can talk to and interact with,” Oliveros stressed. “That didn’t exist as much in the past. We can reason with it, talk to it, give it follow up questions. This new piece is generative AI. It takes the patterns and creates new things.” Guided by the questions and prompts it is given, AI can write anything from emails to novels to computer programs. Oliveros distinguished between a Google search which produces a list of web pages on the search topic and generative AI response to a search prompt or question. “Generative AI has read all the pages,” Oliveros explained. “When you ask [AI] a question, it has this vast amount of knowledge that will help the machine answer the question.”

But Oliveros cautioned, “because it is just thinking about patterns, it isn’t right 100 percent of the time. It can get the patterns wrong. That’s what the term AI ‘hallucinations’ means. AI tries really hard to give you an answer.” How to avoid the ‘hallucinations’ problem? “AI is also really good at following directions,” Oliveros observed. “If you tell it don’t invent and make things up, it is going to follow the direction. Tell it, ‘If you’re not sure, say so. Don’t guess.’”

Popular free generative AI platforms include ChatGPT, Claude, and Google Gemini. Oliveros’s personal preference is Claude. “Claude feels more natural to me. ChatGPT feels robotic,” Oliveros said, but added, “others feel differently.”

“I’m not a software engineer,” Oliveros conceded. “But I’m good at describing what I want to AI. I’ve used AI to create programs and apps.” However, Oliveros always has skilled coders check her work. She has also used other AI platforms to check work she creates with AI. But Oliveros stressed, “Don’t take AI’s confidence for truth. AI pulls data from everywhere, good and bad sources.” She expects AI to get better at what it does. “It’s far better than it was a year ago,” Oliveros observed.

Another caution from Oliveros, “Free AI platforms take the information you give them and use it to train larger models.” Never give AI passwords, social security numbers, or banking information. Paid business and educational AI platforms are told not to use input data to learn. Users of free platforms can, likewise, adjust their setting to tell AI not to use their information to train larger models.

Asked about data centers and their impact, Oliveros conceded, “The environmental cost is a reality. ‘The cloud’ is servers and microchips. They’ve got to be stored somewhere. It takes a lot of space and energy.” A top concern is the energy needed creates heat and vast amount of water can be needed for cooling. Another concern is noise. “Data centers are loud,” Oliveros said. The noise factor was the issue with a data center locating near the Nashville Zoo.

Asked if it was critical for people to learn to interact with AI, Oliveros said, “In the coming months and years, AI may just be happening in the background.” Her advice to users, “The more you tell it, the better answers you’ll get. Be specific about what the situation is, what matters to you, and what format you want the answer in.” Users could variously ask for a report, chart or map. Citing AI’s application in the classroom, students ask AI questions and give it prompts to solve a problem they are confronted with and are rewarded when they find the solution. Attesting to her personal experience with AI, Oliveros said, “AI has made me a better problem solver.”