Are You Talking To A Bot Or A Human On That Website?

Are You Talking To A Bot Or A Human On That Website?

Artificial intelligence is poised to disrupt many of the industries and jobs new college graduates are entering. One particular domain in which AI has already been implemented extensively is customer service. Whether you’re shopping, accessing your bank account, or playing video games, every time you’re online, there’s a good chance you’ll interact with a chatbot. 

This makes customer service an interesting field in which to study what it looks like for humans to work alongside AI. And in the coming years, these lessons will become more relevant as AI begins to disrupt fields such as marketing, software engineering, and much more.

During my 2020-2021 gap year, I had the opportunity to intern with Netomi, a company that specializes in building these types of AI chatbots for companies across verticals in travel and hospitality, gaming, eCommerce, and much more. For me, it was particularly interesting to learn about both the capabilities and limitations of these AI chatbots and what parts of the customer service experience require humans and AI to work in tandem. It’s also exciting to see how AI can create a new future of work. Across all careers, AI has the potential to free up humans from the repetitive and boring tasks that contribute to burnout, and allow individuals to focus on the higher leverage work that intellectually stimulates them.

How Chatbots Work

Most chatbots work the same way. They identify keywords in the question and use a flowchart to match the question to an answer in a database. It’s a simple input and output transaction. The benefit of these chatbots is they save time for customers and businesses. 

A few features that set apart mediocre chatbots from the best chatbots is NLU (natural language understanding for human-like conversation), sentiment analysis , and human escalation algorithm . That means the chatbox has the ability for the chatbot to detect what the customer’s next step action should be and  knows when to involve a human.

The chatbot AI’s ultimate objective is to skip human intervention entirely by answering simple questions that are asked most often (what are your hours, do you accept returns, do you deliver, etc.) They save time so human customer service specialists can spend more of their energy answering more complicated questions.

A recent quote from a webinar my company held with the Head of Customer Service from one of our customer companies sums up the impact of AI:

“When we implemented [AI] we were drowning in [customer service] tickets. A lot of them were repetitive answers. It was a simple matter of pasting in the template and sending it to answer these inquiries. When we implemented [AI] it was exciting for the team because it gave them a chance to stop feeling like a robot. They were able to feel like humans and they were able to answer more complex situations and issues the customer had … Now they’re able to focus and feel like they’re customer service specialists and do all the complex research. [The customer service team] loved it.” 

How Can You Tell if You’re Talking to a Chatbot?

While today’s chatbots are quite advanced and can be indistinguishable from humans, there are still certain scenarios they don’t handle as well. Here are a few ways to tell you’re speaking to an AI chatbot and not a human.

Your Question Is Complicated or Multi-Layered

If you ask a question which is unique to your particular situation or is multi-part and you get a comprehensive response, most likely you aren’t talking to a chatbot. Because AI chatbots rely on a database of answers, they can only cover common scenarios. Additionally, if your question consists of multiple parts, it can be difficult for the chatbot to parse out the individual questions within your request and respond appropriately.

It Keeps Responding to Different Questions With the Same Answer 

If you keep getting the same responses to slightly different versions of a similar set of questions, you’re probably talking to a chatbot. For example, if you’re asking different questions about an online order and repeatedly get the same answer, even when the questions are about different aspects of the order (such as: it keeps telling you the delivery date when you ask about who the deliverer is), you’re interacting with a chatbot. The issue chatbots encounter in these types of scenarios is they can’t gauge the right “intent” behind your question. The chatbot doesn’t understand which specific piece of information you’re searching for. 

Context Matters

For the Tinder fiends out there, if Jessica from 0.4 miles away looks suspiciously like a famous celebrity, you’re talking to a chatbot. That’s an obvious example, but a good way to find out if it’s a chatbot is to keep it talking until you get to the sales pitch. A real person on a dating site probably isn’t going to try to get you to send them gift cards. Most companies will be a lot more transparent with their chatbots. 

Short Term Memory 

Less advanced AI chatbots are also unable to remember information you provide earlier in the conversation. If you ask a different unrelated question, these chatbots will begin asking for your basic information again. That’s why sometimes you encounter situations where you’re asked for your email five separate times by the same “person.” In these cases, you’re speaking with a chatbot. 

Just Ask It

Most chatbots will tell you if they’re a bot, especially on an eCommerce site. They’re not trying to fool anyone. Companies frequently personalize their chatbots to make them more humanlike, but there’s no harm in asking if you’re dealing with a chatbot. 

Why Should You Care?

The best chatbot on the market will answer a user’s questions and, if it can’t, will switch them seamlessly to a real person. It doesn’t matter when they’re all trying to reach the same goal: get our questions answered as quickly as possible. AI is less about replacing humans and more about augmenting them.

In customer service, AI chatbots serve as the first line of defense, resolving straightforward customer inquiries without human intervention. That way, real customer service agents can use their time more effectively and help customers with more complex requests. In the end, customers are happy they get their questions answered faster, and customer service agents are happy they get to work on questions that require more advanced use of their skills. 

For students at Swarthmore looking for their first jobs, it is important to recognize that AI automation may play a large role in their careers depending on the particular field. 
Rather than fearing being replaced, students should look for opportunities to work with AI and recognize that AI frees them up to focus on solving the hard problems we’re facing today.

The opinions expressed in this article are solely the author’s and do not reflect the views of The Phoenix Editorial Board. 

Leave a Reply

Your email address will not be published. Required fields are marked *

The Phoenix