The startup DoNotPay, which calls itself “the world’s first robot lawyer,” is introducing a new AI-powered chatbot that can assist you in settling accounts and terminating subscriptions without having to speak with customer support.
The chatbot successfully negotiates a discount on a Comcast internet bill using Xfinity’s live chat in a demo of the programme provided by DoNotPay CEO Joshua Browder. The bot uses the account information the consumer has provided to contact a customer service agent and request a better rate. The agent answers by promising to deduct $10 from the customer’s monthly internet bill in response to the chatbot’s legal threats and complaints about Xfinity’s services.
This application expands on the numerous useful services DoNotPay already provides, namely allowing users to create and submit templates to different organisations, assisting them in filing complaints, cancelling subscriptions, disputing parking tickets, and more. Additionally, it helps users protect their photographs from facial recognition searches and highlights the key provisions of a terms of service agreement using machine learning. The AI chatbot being used by DoNotPay to communicate with a representative in real time is, nevertheless, a first.
In an interview with Browder states, “For the past five years, we’ve primarily been using rules-based systems, and what I mean by that is templates. “We’ve trained our AI to be like a robot lawyer for consumers, and I imagine that the disputes that we can manage now have gone up dramatically since we can handle scenarios where you can respond instead of just providing one template,” the researcher said.
With the exception of one snag, where the tool states “[enter email address]” rather than supplying the user’s actual email, DoNotPay’s bot answers Xfinity’s questions in a sufficiently human-like manner throughout the whole exchange. The bot seems overly polite since it uses “thank-yous” a lot, but Browder tells The Verge that DoNotPay will clean up some of its responses before it goes public.
However, once the chatbot is made available to all users, Browder claims that in this specific instance, the AI “exaggerated the Internet disruptions, similar to how a client would.” The final version “won’t allow for embellishment of facts,” according to Browder. I can never state for myself when I interact with a customer care agent that “but it will still be forceful, citing legislation and making an emotional appeal.”
The ChatGPT chatbot, which a tonne of people have been experimenting with to produce thorough (and occasionally absurd) responses, is built on top of OpenAI’s GPT-3 API, which forms the foundation of DoNotPay’s bot. Although the tool from DoNotPay is intended for a specific goal, Browder appears to see this as a chance to increase the range of duties it can handle, such as negotiating a credit report or speaking with a representative to terminate a customer’s subscription.
According to Browder, the chatbot won’t start making things up if it doesn’t know the answer to a certain inquiry. When in doubt, “it will just stop in its tracks and seek the user for aid,” according to Browder. In order to save consumers from having to watch the tool on their computer, the business is experimenting with ways to notify users whenever this occurs. According to Browder, individuals may eventually reply to the AI’s texts through text messaging so that it can carry on its “conversation.”
