Artificial Intelligence isn’t what it’s all cracked up to be
In the world of analytics and predictive modeling, many executives and consultants get excited about artificial intelligence. They want to be first to the market, implement it to beat their competition and win the war for the consumer dollar with it.
But, the reality of it is that humans do much of the word behind artificial intelligence. Yup, that’s the truth. In our world of text analytics, in which our competitors dumb down their data so their proprietary platforms can provide the information executives need, humans often must step in to read the data and make the necessary judgment calls on the data.
Here’s an example of how it works. Let’s say Company A is analyzing a text from a customer that reads:
“I was so frustrated when I entered the store. I had already contacted customer service online and they were of no help. I had tried calling the 800 number and I could not even understand them, so I eventually just went to the store. After nearly an hour, I finally got my phone fixed and was on my way. Please find a better way to service your customers correctly the first time.”
Now, let’s run some Natural Language Processing on that text to remove the stop words.
We’re left with:
“I frustrated I entered store. I already contacted customer service online help. I tried calling 800 number I understand, I eventually store. After nearly, I finally fixed . Please better service customers correctly.”
Now, let’s do some quick stemming – in which we reduce the word down to its root word, and lemmentazation – in which we reduce the word down to the common dictionary meaning.
Now, we have:
“I frustrat I enter store . I alreadi contact custom servic onlin help . I tri call 800 number I understand , I eventu store . After nearli , I final fix . Pleas better servic custom correctli.”
These three basic text analytics processes are needed in order to get the data ready for the “artificial intelligence” machine to analyze the text.
Now, can you tell me if this is a satisfied customer or not?
Neither can I.
So, as part of the artificial intelligence process, this would get put into a bucket for humans to read and decide if the customer is satisfied or not. I wonder if they will put it in the right bucket? I wonder if one person on the team would call the customer satisfied and another say it’s an unsatisfied customer. I wonder how many mistakes are made?
It doesn’t have to be this way.
Decooda is different. Our Cognitive Intelligence Machine analyzes the organic data, structured or unstructured, with a very high degree of precision and recall to ensure the best results.
We’re so confident in our analysis, that we will run the data from your current CX platform provider through our Cognitive Intelligence Machine and improve the results.
Executives should ask their current CX platform provider:
- Based on the data, what should I do next?
- What will the impact be?
- Do I have confidence in the findings?
And, if you don’t like the answers, Decooda can give you a new perspective.
Comments are closed.