Yes, you read that right: the speed of AI is vastly OVERrated, not UNDERrated.
Now I can hear you say: “but Zanna, AI developments are going exponentially crazy fast!”
Very true!
But what’s not going exponentially crazy fast?
The large organisations that need to integrate AI in their processes, let alone in their legacy systems.
To not even talk about the privacy restrictions that are in the way of giving true personalized experiences.
Let me give an example.
How many of you recognize stories like these:
You go to a website to find information.
Proactively a chatbot is launched at you.
My learnt response by now (since most of them are so bad) is to be very sceptical yet give it a try.
It’s also part of my profession to stay curious (and positive and hopeful) about the improvement of applications like these.
OK, so I ask a question.
Immediately, based on the response, I know that it’s just FAQ-ing.
It’s responding with an FAQ answer and most of the time that answer is not what I’m looking for.
My journey here tends to be: what’s the fastest way to get the bot to ask me “do you want me to connect you to a live agent?”
Yes please!
PS: often it’s just answering “no” many times that will get the prompt to be connected to an agent, free tip, enjoy it.
Honestly, I’d really love to hear about your positive experiences to prove me wrong.
I’m afraid though that your good examples are the exceptions.
In my experience I’d argue that at least 90% of all chatbots are still FAQ bots who have been levelled up with the elevating AI aura (meaning just given the name AI).
Soooo many chatbots are still a huge drama.
Not because of the quality and speed of development of AI.
But because of the bad design (more to come in the next blog) ánd the restricting connections to older, existing (legacy) systems that most large enterprises are using.
By the way, if you have a supplier coming at you saying that they have a chatbot up and running within a week, plug and play, easy peasy, be very very very sceptical.
You need to feed the (actual) AI model a huge amount of data for it to learn about your context if you want it to be more than just an FAQ spewer.
Side note: I’m still shocked at the level of super bad CRM systems I come across.
Well-known, large global organisations are still having really bad data quality in their CRM.
At the end of 1999, when I did my thesis for that same Business and IT degree, CRM was the new hype.
So when I started working for IT companies in 2000, I assumed that I would come to work at organisations who all had amazing CRM systems.
Guess what, even 25 years later (!) most CRM implementations are still investing millions of dollars while not having the business impact they were looking for.
Salesforce, SAP, the large CRM system suppliers are still to this day sharing with me their struggle how most IT implementations only have a technical view.
It’s really hard for them to convince their clients to look at the business side, both for CRM as well as now also AI related implementations.
Let alone design proper experiences for customers ánd employees before checking the technical needs.
And hint: this is the same bad CRM data that your AI needs to work with!
So no, AI will not be super fast in creating amazing CX for large organisations.
Mostly they tend to be separate pilots or separate tools that solve a tiny part of the experience.
This blog is part of a new blog series by dr. Zanna van der Aa
AI and CX: Demystifying the Hype
