[2023] [deleted by user]

Hey all,

Confirmed with Player Support that they’ve responded to OP on this (think OP is offline for the night) but wanted to respond here too since people have some questions and concerns.

To answer the big questions here: we don’t write responses using AI and if you’re talking to AI when you’re communicating with Riot Support then we’ll tell you.

The team is still sorting through what happened here, but here's some context. We maintain pre-saved phrases and responses that we encourage agents to use as the baseline in responses for common issues. This helps ensure that there’s consistency plus it speeds things up. Agents can adapt these to their own voice and save them in their accounts to respond to players when there's language that works for them. In this case, this individual agent used some AI-generated language as part of their saved response. Writing any part of a response with AI isn't part of our protocol and shouldn't have happened.

Couple points about how we use AI: - we do use chat bots and automated responses in some cases (~15% of tickets) - we're honest about it every time we’re using automated responses (like BlitzBot and our online chat widget) - we don’t use character.ai or have any relationship with them - we don’t use other AI’s like chatgpt to generate responses to players - we don’t allow player tickets to go through any tool outside of our secure network

Hope that helps clear some of this up. There’s a lot of discussion happening inside the team coming out of this and we appreciate OP letting us know.

Yes and no. Easy copy/paste responses are part of our approach to player support but AI-written ones (at least like this) aren’t.

The team is still sorting through what happened here, but to answer the big questions here: we don’t write responses using AI and if you’re talking to AI when you’re communicating with Riot Support then we’ll tell you.

We do maintain pre-saved responses for agents to use as the baseline for their response for common issues. This helps ensure that there’s consistency plus it speeds things up. Agents can adapt these to their own voice and save them in their accounts to respond to players. In this case, this individual agent used some AI-generated language as part of their saved response. Writing any part of a response with AI isn't part of our protocol and shouldn't have happened.

Couple points about how we use AI: - we do use chat bots and automated responses in some cases (~15% of tickets) - we're honest about it every time we’re using automated responses (like BlitzBot and our online chat widget) - we don’t use character.ai or have any relationship with them - we don’t use other AI’s like chatgpt to generate responses to players - we don’t allow player tickets to go through any tool outside of our secure network

Hope that helps clear some of this up. There’s a lot of discussion happening inside the team coming out of this and we appreciate OP letting us know.

I was here first so i like to think chat gpt sounds like me, not the other way around

I gave it a try:

Me: Can you help me write a response to a thread on reddit? I need to respond to questions about whether a video game company is using artificial intelligence as part of its customer support program. The company does use AI for some things, but not in the way the thread suggests.

ChatGPT: Sure! I can help you write a response. Could you provide me with the questions and the suggested way in which the company's AI is being used for customer support, so that I can assist you in crafting an appropriate response?

Me: Sure! The company uses AI to respond to routine inquiries that can be easily addressed (I lost my password, etc.). But most (85% or more) of responses are provided by real human agents, often using previously written statements. The thread alleges that these previously written statements were written using AI, which isn't the case.

ChatGPT: Here's a response you can use for the thread:

"Hey everyone, I wanted to chime in and clarify some information regarding the use of artificial intelligence (AI) in our customer support program at [Company Name]. While it's true that we do incorporate AI into certain aspects of our customer support, it's not used in the way that some have suggested in this thread. To provide efficient and timely responses to routine inquiries, such as password resets and common troubleshooting steps, we employ AI technology. These inquiries can often be addressed with standardized solutions, and AI helps us handle them swiftly. However, it's important to note that the majority of responses, approximately 85% or more, are provided by real human agents. Our team of dedicated support representatives takes pride in personally assisting our customers and ensuring their issues are addressed with care and attention to detail. There seems to be a misconception that the previously written statements used by our human agents were generated by AI. I want to clarify that this is not the case. These statements are crafted by our support team members based on their expertise and experience, allowing them to provide consistent and accurate responses to frequently encountered issues. At [Company Name], we value the human touch in customer support, and our primary focus is to deliver exceptional service to our community. AI is employed to complement our efforts by handling routine inquiries, while our talented human agents handle more complex and specialized customer needs. If you have any specific concerns or questions about our customer support program, please feel free to ask. We're here to address any inquiries you may have. Thank you!"