CHAMPAIGN, Ill. (WAND) — Artificial Intelligence (AI) is becoming smarter and more powerful each day. The technology is creeping into everyday life, changing the way we work, play and communicate.

WAND News visited the National Center for Supercomputing Applications (NCSA) and the Center for Artificial Intelligence Innovation (CAII) , both housed at the University of Illinois.

"It's improving at a very significant speed. This is really amazing to see how much has happened," Volodymyr Kindratenko told WAND News.

Kindratenko is the assistant director for NCSA and director of CAII as well as an adjunct professor at the U of I.

"This technology has a lot of potential, great potential, and we should not be afraid of using it," Kindratenko explained.

He works with supercomputers each day to understand AI and develop improvements for users across the world.

"When we say AI today, for the most part we really mean these tools that are able to converse with people in human language- and also at a level that it's difficult to differentiate between whether it is a human or is it a machine," Kindratenko added.

Generative AI is artificial intelligence that can write, make images, and create audio and video based on human prompts. But it can also be integrated into technology like spell check, video editing and even scientific research.

"This can actually improve productivity very significantly by taking away simple mundane tasks that people spend a lot of time on, and many of these can be automated with these technologies," Kindratenko said.

AI is already creating change across several industries, allowing workers to focus on creative tasks and let computers do the grunt work.

"So it's to simply some of the tasks that are otherwise are complex and someone has to spend a lot of time implementing them. Also to provide the capability to run things faster to give scientists the ability to run things faster, to give scientists the ability to improve their productivity to give them the ability to do more in a shorter amounts of time," Kindratenko told WAND News.

But he said there are limits to the technology. Right now, most common AI platforms are not able to differentiate factually correct information, based on the data they've mined from the internet.

"There are a couple of big obvious risks with this. One of them is the quality of the models heavily depends on the data they train it on. If the data is polluted with misinformation or incorrect facts- these models don't know any better, they will just repeat them," Kindratenko explained.

The other issue- hallucinations. These happen with AI doesn't have an answer to your question and simply makes something up. Kindratenko said his research is focused on finding ways to prevent this.

"The way we address it is a method called RAG- retrieval augmented generation. Through this technology, we force the live-language model to not invent an answer but actually invent an answer based on a very specific chunk of material," Kindratenko said.

As generative AI continues learning and improving, Kindratenko said it will become and even more powerful tool- but also more dangerous.

"We should look at how to make sure this technology goes in the direction we care about and the direction that the society can benefit from. This is really important to make sure this technology is not misused and is kind of used for the benefits of all- not just for the benefits of some," Kindratenko added.

WAND News will continue our deep dive into artificial intelligence each night this week at 6 p.m.

Copyright 2024. WANDTV. All Rights Reserved.