Despite artificial intelligence seeming instantly everywhere since the public launch of Chat GPT in November 2022, AI’s rise has actually been under way for decades.
James Aden, a Chattanooga-based vice president with CGI, one of the world’s largest IT and business consulting firms, talked about AI’s evolution at the first of a new “Masterclass Series” of learning opportunities for working professionals. The series is jointly presented by the University of Tennessee at Chattanooga Center for Professional Education (CPE) and the Chattanooga Technology Council, or ChaTech. The topic for the inaugural Wednesday session: “Understanding AI Foundations.”
The widely accepted “start” of AI is often described as when English mathematician and computer science pioneer Alan Turing asked in a scholarly paper in 1950, “Can machines think?” followed in 1956 by a summer workshop at Dartmouth College to focus on ideas about “thinking machines” and which chose the name “artificial intelligence” for the project.
For decades, Aden said, AI was an idea without the technology to make it happen. He described Turbo Tax software, first popularized about 30 years ago, as an AI “forerunner,” in which the user interacts with the program via a series of guided questions.
Advances in technology since then have increasingly yielded more opportunities for humans to serve in oversight rather than in “task” roles.
“We’ve been trying to make this happen for a long time and are finally getting closer,” Aden said.
He summed up five main capabilities of generative AI today this way:
- Generation—Responding to a prompt and making “something from nothing” – such as generating a list.
- Production—Taking information and attempting to “synthesize core concepts.”
- Transformation—Taking existing content and changing it to something completely different. Audio AI for musicians, for example. “There’s a Frank Sinatra AI, and I was able to listen to Frank Sinatra sing Snoop Dogg’s ‘Gin and Juice,’” Aden said.
- Knowledge access—All search engines have some kind of AI built into them now, “regardless of how helpful they may or may not be,” he said.
- Emergent behaviors—The result of giving an AI platform context “to understand the world and then asking it to solve problems we haven’t solved,” Aden said, citing an example of an AI-designed athletic shoe considered to have better met performance requirements than previous designs (by humans).
Aden also noted the growing role of generative AI today in business helping with “what used to be judgment calls.”
As an example, he cited “computer vision”—feeding images or video footage of hard-to-access structures captured by a drone, such as power generation systems, into AI for an opinion or recommendation on maintenance or structural integrity concerns.
Following Aden’s lightning-fast trip from the information age to the intelligence age, two UTC faculty members briefly addressed ethical issues and AI. Dr. Chandra Ward, an assistant professor of sociology and director of community engagement for the UTC Center for Urban Informatics and Progress, and Dr. Shahnewaz Karim Sakib, an assistant professor of computer science, took on the topic of potential bias in technology.
While technology is often assumed to be neutral, Ward noted, its creation by humans risks inherent human bias built in. “When we think about technology,” she said, “we often think about efficiencies it can bring, and prioritizing that, but if we overlook the element of potential biases, there’s a risk of greater inefficiencies down the road.”
Sakib, described as having research interests at the intersection of cybersecurity and reliable AI, took several questions from members of the capacity crowd ranging from AI “hallucinations”—information obviously inaccurate or not based on reality, to false citations by AI of attribution for information.
“Yes, I have had AI cite academic papers as the source of information provided, only to go and look up those papers and find they didn’t exist,” he said. “There are many strategies being applied to eliminate that from happening and, while problems persist, including preventing hallucinations from persisting, there is work under way to address it.”
* * *
Wednesday’s event featuring Aden, Ward and Sakib was the first in a four-part monthly series of learning opportunities targeting working professionals and organized by “CHAIN,” or Chattanooga’s AI Network, a joint initiative of UTC CPE and ChaTech. According to CHAIN, the series is intended to support the growth and application of AI across industries in Chattanooga.
Next in the AI series from CHAIN:
The Future of AI in the Workplace
11:30 a.m. to 1 p.m. – Sept. 25
More information and register: HERE