Audioburst Search

Coming up next

Meet Emma, The Immigration Chatbot

AI Today Podcast: Artificial Intelligence Insights, Experts, and Opinion
|
2 months ago

AI Today Podcast #136: ML Model Management and Operations (ML Ops)

AI Today Podcast: Artificial Intelligence Insights, Experts, and Opinion
|
6 d ago

ML Model Management and Operations (ML Ops)

AI Today Podcast: Artificial Intelligence Insights, Experts, and Opinion
|
6 d ago

Automatic TRANSCRIPT

Our guest today. Is Courtney wind chip who the division chief of the Digital Services Division at the office of Citizenship and applicant information services under the US citizen and immigration services. USCIS hike Ernie. Thank you so much for joining us on the podcast today. Thanks for having me. Yeah welcome Courtney and thanks for joining us. Today we'd like to start by having you introduce as yourself to our listeners and tell them a little bit about your background and your current role at US CIS. Sure happy to do so. So as you mentioned. My Name's Courtney I'm with. US citizenship immigration services that we oversee the lawful immigration and naturalization in to the United States so this includes any family employment or Humanitarian Ryan benefits that an individual may apply for in order to enter remain in the United States. My current role is as a digital services division cheap and we we oversee all of the public visual tools that we have which includes very USCIS website English and Spanish as well as websites. Thanks for citizenship. Resources as well as they are five and we are awesome building out account services. This is an area of the web that is cure that it allows for online transactions and other support services that allow individuals to self service and provides a bit more precise information. So that's my permanent role. I am in an acting role as well where I oversee. Data Management in our office of chief data offers great. Well I know one of the reasons why we're having you on this podcasts. Of course because we're focused on artificial intelligence and one of the patterns that we talk a lot about is one of those seven patterns of a I is this conversational pattern and the US deployed employed. Recently one of the most widely used chat bots in the federal government called. Ask Emma and I know that website is very highly heavily traffic traffic website and of course also feels a lot of phone calls. People are definitely very enthusiastic about contact regarding immigration immigration status lettuce so for our listeners. Who may not be familiar with? Ask Emma and what you're doing on the chat bots. Can you tell us a little bit about that and how that came to be yeah. Emma started in December number twenty fifteen and is the first interactive virtual assistant to date. She has successfully responded to more than thirty five million inquiries for more than eleven million users. So we're very very excited about that. And she has a success rate of about ninety three percent in English and ninety percent in Spanish so she has become a very useful tool for many of our applicants and the general public. I give you a sense of her usage versus some of the other things things we receive about fourteen million calls from and all of our digital tools. Together you have about twenty five billion sessions for months so you can see. She is a highly traffic pool that we rely on very heavily and she has grown and expanded her knowledge over the years. So right now she has an. I don't know right that hovers between eight to ten percent English. And we're hoping to get that down even lower over coming years. And as we continue to train her and we also see her as a great tool to collaborate with our external stakeholders as well as internal stakeholders so initially she a trained with our adjudicators our case managers review cases and our contact center staff over time. We had her trained also with the public and now she has a ton of information and then in more recent years we've also integrated her with live chat services so she's continuing to the pills than where she doesn't have goals. We are now able ambition to from live help services. Yes that's a lot of inquiries every month when you said those numbers I was like. Oh my Guinness so I can see how a chap can be very useful and now that you've had 'em for a few years. I'm sure that there's a lot that you've learned and as you said you continue continued to improve and try and get her. I don't know right up. So for other agencies that are looking to deploy chat bots and industry as well for that matter. What lessons have you learned both good and bad from this experience that you can share with our listeners so we obviously learned many many lessons along the way you know? Some of the lessons that I think are important than have have been really key for us are clever. Aiding and Co creating from the outset so working as I mentioned with our internal and external stakeholders the really develop develop tour of that is all and also has allowed us to think about ways in which we can leverage technology as best we can so five years ago how we were using chat bots is very different and the investments. We're making this very different than than some of the technology available now so I think allowing us to find a lot of the tools that are available right now row for better automation and content management. And that's something that we have struggled with over the years and I would recommend the folks that if they can consider that and consider how also how transferable it'll may be based on the technologies they're trying not to lock into length solution Shen but really be able to use a variety of solutions meet their needs and also how that could integrate with other efforts that they have that they've prioritized of the years. Yeah look at Chat bots and conversational systems of all sorts kinds that are built until websites but also texting and messaging and even now the new voice assistance on Alexa and Siri Google home and all those places in Cortana and they may be thinking about the return on investment. Right a lot of folks like well. Why would use the Chat Bot well hopefully to reduce call volume maybe to improve customer support and service maybe to respond to requests? When you're not usually open right the the offices are closed and I think you know there's lots of different ways of thinking about an inmate? It'd be interesting to share from your perspective. You know where and how you know the Chat Bot in specific and maybe more generals barrels having a big impact on the way that you're serving customers and just both in terms like where are you seeing the return on investment now but also looking towards the future are you looking at its future. Roi and maybe ways that you're looking to expand the use of chat bots and AI. In general in the next few years got excellent questions. I think our biggest a school has been how to create more space for our case workers to focus on complex immigration issues news and reviews rather than some of the more basic questions like K.. Status or processing time but we receive on a day-to-day basis so allow Emma and other tools to provide that information to the public and ideally get them. It's timely accurate information as quickly as possible on their timeframe right as opposed to having to rely on when contact centers opened lower case workers to focus on the more complex cases and where needed having to have the in person interaction. So I think that's where we've seen value. I mean it's something I will say is a phrase. We thought Emma would produce call volume. We find that and I think one of the things that we are looking. We've been carefully examining over the last couple of years but particularly I I was saying last six to twelve months is how'd we most effectively use the channels that we have available to the public for inquiries are customer. I'm Craig Channel and that we may in fact start to make some really tough decisions about what channels are available for. What purpose and really start to nudge? The people into certain types of tools or resources so that they can get the best information that we have available and allow again our team. Focus on the case Orca. Though you know it's interesting that you said that Emma didn't actually reduce the call volume. I guess people use her as first line of Defense and ask her certain questions but then at the end of the day maybe they still want that human interaction so have you found that it's reduced call volumes for certain types of questions or is that a goal that you're looking to the have in the coming years that's a goal. We are looking to have in the coming year so right now. She provides some basic information about K status and processing assessing times in particular which are the areas. We get the most inquiries about gloves things like changing your address. But I think what we're trying to do is use a similar functionality. optionality that Emma has used that in a more secure environment authenticated environment