The world is drowning in chatbots, and it’s not solving problems for the consumer. ChatGPT ushered a new way to interact with LLMs in a chat-style UI, with several limitations, including the inability to train off custom data, and knowledge of events after September of 2021.
The world accepted these tradeoffs for one of the fastest growing consumer services in history, generating over 50m active users per month and reaching one million users in just 5 days of being released.
What followed was quite predictable. Google called a Code Red, a dozen start-ups spawning with “ChatGPT for X”, several a16z blogs detailing this groundbreaking opportunity, and professionals who started to work in AI in 2019 being deemed as the “Thought Leaders.” By no means am I diluting the impact that ChatGPT has on our society. What I am taking a closer look at is the response from the business community and the approach they are taking on this current Generative AI wave.
Just about every consumer/B2B app now has some sort of chat option. My bank, my project management tool, my HRIS, and my to-do list app. I’ve tried each of them to see if I’m able to get some sort of insights or better usage out of the app, to mostly no avail. Most of the time, the LLM is unable to fulfill my request, and actually asks me to contact customer service. I should be able to ask my banking chat friend what are my monthly recurring purchases, as it shows up elsewhere in the app, but alas, I am unable to.
To me, it’s clear that companies are trying to match a solution to a problem. Rather than identifying a clear use case where the powers of LLMs can be leveraged (account data analysis, financial tips, customer service, etc.), there’s an inclination to ship some sort of “catch-all” chat genie with a gradient icon. That leads to consumers testing out solutions to their problems, breaking every rule in User Research 101. When consumers need to ask questions to solve their own problem within your app, that usually means your UX took a step backwards.
All in all, AI is a moving target. What is considered “intelligence” today may not be tomorrow, as the capabilities of generative AI and automation get better at an outstanding rate. It’s crucial to identify problems that the consumer is having from a UX standpoint and then validate that generative AI and a chatbot is the real solution. For many people, it just adds one more layer of complexity to get through to have their critical problems solved by a real person.
Make sure to subscribe to better learn how to identify automation opportunities, and ensure that you’re implementing automation and AI the right way. We will be back to our regularly scheduled projects next issue!