SAN ANTONIO – “Hey, chatbot, what can you assist me with?”
Artificial intelligence, or AI, is a key tool many companies have implemented or will implement into their workflows. Some of the tech world’s well-known CEOs say AI is revolutionary, but how could you incorporate this technology into your everyday life?
Chatbots are one way AI has made its way into our lives by being integrated into existing programs, through customer service uses or through apps like Open AI’s ChatGPT, Google’s Gemini or Microsoft’s Copilot.
The possibilities with chatbots seem endless. Some examples include using the tool to answer questions and assist with homework and even more complex tasks, like helping a customer with an issue when there’s not a human around to help after hours.
“When people are thinking about chatbots today, they’re usually thinking about large language models like ChatGPT or maybe Gemini. and Claude and others –– these closed source models,” said Dr. Anthony Rios, an assistant professor of information systems and cybersecurity at UTSA. “They basically have these huge statistical models that have been trained on basically all over the internet, and they’re trained to learn to predict the next word, given some few words as input.”
Chatbots seem to know everything you can imagine, and that’s all by design.
“Once you have these models trained, they can act as an artificial agent or a chatbot where you can kind of converse with them about different topics,” Rios said.
KSAT put ChatGPT to the test. We asked it to recommend a brunch idea for picky eaters and places to visit in San Antonio, and even asked it to draft a forgiving email to our boss!
Rios said this technology is becoming more integrated into everyday workflows, such as writing and brainstorming. However, some companies are also using it as front-end tools to interact with customers.
“You can use this agent –– you supply them with a lot of background knowledge via prompt, and then you can allow users to interact with that,” Rios said. “Now, with that said, you have to be very, very careful with allowing front-end users to actually interact with these chatbots.”
Companies must implement safeguards to prevent chatbots from doing things they’re not designed to do, like returning responses with offensive language or ordering items a customer may have never requested.
Chatbots potentially make life easier for users, but, like humans, they also make mistakes. One example could be a chatbot being used in a medical setting.
“A very common use case these days is, ‘How should I prepare for a surgery?’ or ‘How should I make sure that I take care of myself after surgery or after a certain procedure?’ If I’m not speaking the exact way that that language model expects me to speak, well, then there could be harmful outcomes because of that,” Rios said.
Rios said many limitations have already been put in place for models used in health care or those dealing with sensitive data. However, it’s still imperative that a company disclose any use of AI from an ethical perspective.
When it comes to using chatbots, Rios said consumers need to develop data-centric mindsets to determine if this type of technology is useful for them and verify that the information they’re receiving is accurate and fact-checked.