You open a website, and in the bottom right or left corner, there’s a bot icon sitting in front of your eyes and it constantly asks, “Hi there, how may I help you?”
It’s not bad initially, but when you actually start asking a question to it, it will hallucinate like crazy even if it is properly trained.
Here’s something from my personal experience.
I used an AI chatbot from a reputed hosting company, and I was having doubts about moving a website from a VPS to a Cloud infrastructure. And the crazy part is, that reputed hosting company had their own migration setup (which I found out later when I was going through their docs), but that “good” AI chatbot simply suggested that I take help from their competitor to achieve this. I laughed out loud.
I am not saying all AI chatbots are bad, but most of them are.
The important thing to do before launching an AI bot is train it, train it, and most importantly train it to all extents. Who knows, your customers may use some words to jailbreak the restrictions you’ve added.
Send it to your team and ask them to test it a lot. Do hundreds of questions and follow-up questions.
If you don’t have much time and you are a solo founder like me, then Reddit is your best bet. Redditors are really good if you are good to them 🙂
I won’t go deep into Reddit here because that itself is an entire topic overall.
And not just hosting companies, every industry is trying to build their own support team with AI chatbots. Real estate, shipping, cloud services, banking, e-commerce, and most importantly even healthcare… I genuinely feel bad for people who take help from an AI bot regarding health-related questions.
Even fast food chains are implementing AI chatbots to fulfill orders, and it is very easy to trick them into giving free meals. You just need to use your brain a little bit more to achieve it. I will not go deep into this.
The real problem is that many companies are rushing to launch AI support systems just because AI is trending right now. Most of them care more about saying “We have AI support” rather than actually making it useful and reliable.
A badly trained AI bot can frustrate customers (like me) more than having no chatbot at all. One wrong answer regarding payments, healthcare, legal information, or technical support can completely destroy customer trust within seconds and of course, your brand reputation.
Anyway, these are all my thoughts regarding AI chatbots, and my final verdict is:
“We are not in a stage where we can fully trust AI chatbots for customer support yet.”
Until next time, stay strong and keep learning.
– Rocky
