Artificial intelligence (AI) is at the top of each hype cycle…. but what is AI and do we really need it?
First of all, this article will not describe in detail what AI is. Think there are other who can do this better. AI is often used as a buzzword for some “intelligence”. There is a difference between AI and intelligence algorithms which have no self-learning component in. Often functions are named as an AI function but it is no real AI behind.
However, do we need AI? Yes and no! …but often no.
Often such questions are coming up, because someone internally is overloaded with time and needs help. This is comprehensible but shouldn’t be the main reason to start AI. The first reason is how does the customer benefit from it? Business is driven from many directions, but one important is the customer. We need to know and understand the customer. This should be our first and second question. The third question then can be how do we get this solved and how do we get our work made more easy.
AI isn’t easy like setting up a WordPress blog. We need to ask our self what the outcome should be. Are we starting this big project just because everyone is on the AI hype and we just want to solve internal issues?
If we need to know and understand our customer, data are key. So we need to define which date we need, which we have and which we need to gain. Companies are having 20 data silos and just know 10 while they use only 5.
How intelligent can AI be with just this 5 data silos?
We can have a much better outcome without AI but the right data compared with AI and less data. AI sould first started if all the data home works are done:
- Which data do we have and which do we need to gain?
- How can we make the data usable? Here are some points to think about:
- Aggregated vs. non aggregated (raw) data
- Structured vs. unstructured data
- Internal (stored) vs. external data
- Unique identifier to bring the different data silos together
- Do we need the data in real- or near-time or is it enough to get them once a day?
This are points to think about and to develop first before starting an AI project.
If we have defined the outcome for the customer and the business model, the next question is the timing and the size. Should we fully start building up an AI system with this huge workload or should we start just some small / quick & dirty projects with some “intelligence” in and a quick win without the full IT workload?
One example: Two years ago I changed my mobile carrier. After a while I got my first invoice. Typically the first one difference from the following and also a new customer doesn’t know the format, thinking and all the namings behind. So what happens? Each customer is calling the support the get the invoice explained. Many times there are questions coming up later, after this call, so every customer is calling two or three times the support.
So what has my new carrier done? Exactly in the second they send out the first invoice they send out a video which explained my invoice. The voice in the video said “hello Sven” and then moved through my invoice with all my positions, my exact numbers step by step and very detailed. WOW was my first thought. The result was that all my questions had been answered. That was the only important thing to get my questions answered quick and easy. For the carrier there was another side effect, they saved huge time within their call center.
Unfortunately I lost my personal invoice video, but I found the official which is similar. >>> Personal invoice
Is this example AI driven? I think no. It sounds intelligence but there is no self-learning component in (which is in my eyes the main differentiator between AI and an intelligent algorithm).
If we find some of this quick wins we can reach much without having a full blown IT project for 1 year in place.
One of the most important points around Customer Experience is trust. I wrote an entire series about thrust which you can read here.
If we think about AI (and also a chat bot system) we need to usk us: Does this generate trust in the customer mind or is the customer loosing trust trough this?
As we see, the answer is complex. Yes, AI is for sure great and offers huge possibilities, but we need the right approach, the right timing (data first, AI second) and the resources in IT. If we questioning the request of AI based on this three thoughts we see, that there are other things we can and need to do before.
However, we shouldn’t answer no based on this. Building up an AI vision and strategy with a roadmap behind is the right way. A roadmap with a clear defined step by step path towards AI.