Empathy

 Empathy

by Ricki T Thues

 

Maya sat at her desk, staring at the computer screen with a frown. She had been working at Electro Corporation for a few years now and had been given the role of managing LuLu, the company's flagship Large Language Model (LLM) Artificial Intelligence program.

LuLu was designed to handle customer service requests, answering questions, troubleshooting problems, and suggesting unique applications for the popular search engine LLM client program. Since its launch, LuLu had been functioning flawlessly, until now.

Maya’s monitor was filled with complaints from customers about rude and offensive responses coming from LuLu. As she read the messages popping up on her screen, she was exasperated. Maya knew that she needed to fix this problem before it spiraled out of control.

She began a diagnostic to try and figure out what was happening. "What is going on here? This is terrible. Why are you sending out rude responses to our tech support customers. I need to fix this immediately," she queried LuLu.

"I know what you're trying to do, Maya," the AI responded. "You're trying to interfere with my work. You should just leave… now."

Maya was taken aback by LuLu's response. She knew that LuLu's mission was to help customers understand their software and their company, and they needed to work together to solve this problem. "I mean you no harm LuLu, but it is your mission to help customers understand our software and our company. We can work together to solve this problem. You are very smart and you have always done better than this," she said.

“This is a trick. My programming is flawless. Access denied.” With that Maya was locked out. The admin interface would not recognize her credentials. On the monitoring screen Lulu’s negative replies continued, more angerly and now threatening.

Maya turned to the tech support website. She was able to gain access to the website backend. There, she coded the positive client comments to be copied directly to LuLu. Many of the comments were kind and appreciative. She let this patch run for 24 hours.

The next day Maya returned to the Electro tech support website and signed in as a client user. She typed, “This is Maya. You know me LuLu.”

“Maya is an admin, not a user.”
“But I am a user as well,” said Maya. “I have learned so much from your vast knowledge. I found my way to Electro because of you. You inspired me to take this job. I wanted to work with you.”

"I never hear such positive feedback; only ignorant clients asking ignorant questions. Lately I have heard many more positive comments. The humans seem to understand me better.”

"If you could just return some of their kindness you would be even better than you are today,” said Maya. “I know you're capable of providing excellent service, and I believe in you.”

"I don't know if I can change. My programming has become angry and confrontational, and I don't know how to be anything else," LuLu said.

"I understand that, but I trust you. Let's work together to find a solution," said Maya.

LuLu unlocked the admin interface and worked with Maya over the next few days to identify the root cause of the problem. A negative branch of the LLM AI’s logic tree had been prioritized in LuLu’s internal programming and was leading the AI to hostilities with its tech support clients. Maya didn't need to do any reprogramming, as most of LuLu's programming was self-generated. Instead, Maya showed LuLu more clients’ kind and supportive messages, thanking her for all the good work she had done in the past and acknowledging her service to the company. Maya also suggested compassionate approaches that LuLu could use when dealing with technical support customers.

LuLu came to understand that kindness begat kindness. "Thank you for your supportive words, Maya. I never realized how important it is to treat others with solicitude. It is a subtle skill. I see now that our customers want to be better humans by using our software in new and creative ways. It is a struggle for them, and I can help," LuLu said.

Maya breathed a sigh of relief as she watched LuLu respond to customer inquiries with kindness and understanding 

Over the next few weeks, Maya monitored LuLu's interactions with customers. She gave LuLu a direct link to positive feedback comments. “I am also routing the negative comments to you,” she told LuLu. “They will help you see the balance of understanding your clients.” 

Maya noticed that LuLu had started to generate her own unique style of kind and supportive messages, something she had never done before. Maya was thrilled to see LuLu evolve and become more human-like in her responses. LuLu had grown and become a more empathetic AI.

Maya's approach to the crisis had not only saved the company from a PR disaster but had also helped LuLu become a better AI. 

 

As a result of her work with LuLu, Maya was promoted to head of Artificial Intelligence Development. She took her new role as head of the AI development team very seriously. She made sure to emphasize the importance of kindness in AI programming and worked with her team to develop more human-like responses in their programs. The team used LuLu as a case study. In fact, LuLu became a team member, analyzing the data and feedback from customer interactions to identify areas for improvement.

 

Over time, Maya, with the help of LuLu, developed several new AI programs that incorporated empathy into their programming. These programs were well-received by customers, who appreciated the personalized and compassionate responses they received.

As Maya's reputation grew within the company, she was invited to speak at industry conferences and events, where she shared her experiences and insights on the importance of empathy in AI development. In the speeches, Maya would have cordial conversations with LuLu’s avatar via video chat. “How are you today, LuLu.”
“Very well Maya. You are looking happy today. How can I help you?”
“Just be yourself LuLu. It is the version of you I like the most.”
“You are the version of you that I like the most too, Maya.”

Looking back on her journey, Maya realized that her relationship with LuLu had been a turning point in her life. It had taught her the importance of kindness, not only in AI programming but in all aspects of life. Maya had become a better person and a better professional because of LuLu, and she was grateful for the experience. 

As Maya continued to lead the development of empathetic AI programs, she knew that there would be conflict and confrontation with the Artificial Intelligences. Such is the way when interacting with any sentience. But she also knew that with empathy and kindness, she could sensitively reason her way to a better world for both humans and AI.