How Granny Made AI deployment Look Silly. Short Rant About Real Intelligence

Google is stuffing powerful new AI tools into tons of its existing products and launching a slew of new ones. Experts point out it’s a high-risk strategy, given that AI language models have numerous flaws with no known fixes. But this is Google. We, marketers working across operations of all possible sizes, are exposed to far more trivial risks while jumping on the AI train. We can, for example, make ourselves and our company look really silly.

In the world of AI, we have leaders, and …

Google’s actions seem understandable, though risky, as Technologyreview, the voice of MIT, observed. It is a part of Google’s now long-lasting effort to become the world’s leader in AI solutions. Their technology is proprietary. They now strive not to simply deploy, but to invent cutting-edge solutions that will, perhaps, establish a new standard for AI models in the future to come. And their AI will connect all their services in a meaningful way at some point, providing value to the company and customers alike. Or some of their mighty competitors will achieve this goal before them, which is now more than likely.

To put it bluntly, Google has to move fast, which will inevitably lead to breaking things (in a bad way), at least sometimes. Their fight for a position in the world of AI is, most probably, a fight for survival. On the other hand..

… we have those guys

There are a multitude of companies in AI heat right now, and if you are reading this, chances are that your company is one of them as well. And some of them do have deeply thought-out plans for AI adoption, their needs assessed, resources calculated, processes ready to include AI commitment to produce value out of it.

Because, as in case of every asset in business, the reason for deploying AI is value, right?

Other companies are “those guys”. Their actions make little sense, deployments are chaotic, and they seem completely unprepared for the technologies they are using, having no AI-ready processes designed, so the value of said AI cannot be digested by the system. Do they gain anything jumping on the AI hype-train?

Their customers’ stories tell the tale.

The Granny’s Story

My former landlady, for example, is, like we’re all in Poland, a customer of the state-owned service provider. For the sake of the story, we call her Mary – the company’s name is not relevant ;). The lady is in her 90’s, a retired doctor, with her mind still lively, and taking stade-owned companies very seriously.

So, once upon our meeting, she told me how she tried to make some things clear with the company. She called them, of course, because the internet, due to sight problems, is off-limit for her.

Immediately a communication ordeal began, the Mary was sent from desk to desk with her case and no one seemed to be the right person to handle it. Eventually, she managed to connect with a nice lady who sorted things out, in line with her expectations.

A moment later her phone rang.

– Thank you for talking to our consultant. – a nice, male voice rang on the other end of the phone – Please rate the quality of our service.

Such a request has immediately put my landlady on her toes. Of course, she wasn’t happy with this whole desk to desk sending, but the nice lady who finally sorted it out for her did a great job. Did I mention that this lady takes state-owned companies very seriously? She believed that an honest opinion about the quality of service was, in a sense, her duty, but on the other hand she did not want to harm the nice lady in any way. So she decided to describe the matter in detail.

– At first I couldn’t get along with anyone, but then the nice lady I connected with took care of everything and was very helpful, very helpful!

– Please rate the quality of service using a scale from 1 to 5. – After a moment of silence, the same pleasant voice asked.

Mary felt confused, but she still didn’t give up.

– So at the beginning the service was 1, but later this nice lady was 5! – she emphasized once again.

– Please rate the quality of service using a scale from 1 to 5 – the pleasant voice continued to insist.

– Well, I assessed…

A moment of silence.

– YOU ARE TALKING TO ARTIFICIAL INTELLIGENCE! – came from the other end of the phone.

The old lady turned pale and, as she told me, her hands began to shake.

– But I don’t know how to talk to artificial intelligence – she said, extremely embarrassed.

– Please rate…

Mary hung up. Embarrassed and still unsure whether she had accidentally harmed the helpful assistant.

So let’s look for the value

Was there anything to assess really?

Why do you even put post-communication CS assessment in place? To check, if the practice deviates from the designed process and quality, and if so, how far. The lady called with nothing extraordinary, yet still she met with a wall and was sent all over the company. This suggests there is not even a proper process in place, no template to relate to, when applying this scale. Ohh, the scale …

From 1 to 5

The problem Mary called with was a really mundane one, yet still the communication was complicated. There are multitude of problems possible within this service that are far more complex. Is the simple scale from 1 to 5 adequate to assess the job of Customer Service in this situation?

Was it customer-centric?

This is a giant, state-owned company, the monopolist. Their customers are basically most of the citizens, all around the spectrum of age and technological advancement. The younger ones will most likely use the internet, website or apps, to communicate with the company, while those that will choose the phone, are most likely the older ones, often disconnected from modern solutions. For various reasons they crave human contact and human understanding. Are chatbots, in their present state, able to fulfill the needs of such customers, statistically a good chunk of their database?

So what data did the company collect in this situation? An abandoned assessment process. Does it tell anything to the company, provide actionable data?

If anything, it says that existing CS assessment is rubbish, and should be redesigned from scratch.

Did its deployment generate any cost to the company? Or maybe it was free …?
Was the customer satisified?

“Get AI ed” does not mean “Set your brain on idle”!

In this epic battle between Artificial Intelligence and Real Stupidity the latter won. The money was spent, the customer left confused, the data shows inadequacy of the solution to existing structure, processes, and target customer base.

We can now call the chatbot inefficient and put all the blame on it.

Unless we want to drive some real value and deliver it to the company and our customers as well. In this case Real Stupidity – shallow-minded attitude towards technology, laziness, thinking about company’s and customers’ needs in terms of buzzwords and chasing them unprepared – goes out the window.

To utilize the chances brought to the table by Artificial Intelligence, the Real Intelligence is needed to:

Understand various types of AI models, their strengths and weaknesses.
Assess correctly if our own processes are organized on the level fit to include AI’s contribution and the incredible data and performance boost that comes with it.
Choose the right AI model to fit precisely selected tasks.
Deploy consciously, not all over the place.

Because, unlike Google, you don’t need to be a leader in AI. You are to be a leader in sales.