Can't we solve it with AI?
Every project will have some unforeseen challenges appear underway and almost every time someone suggests solving it with some sort of technology or automation. I am not talking about starting a project and ending up with using different automation or smarter solutions to solve a business challenge, but cases where one is working on a project and a specific problem show up and someone suggest the magical bullet of using AI.
Why is it not that simple?
There is one major problem of suggesting AI and automation to solve problems that show up underway in a project and that is time. Most of the Microsoft's (MS) solutions are not hard to learn or to implement, but to implement them correctly however is a lot more challenging. In my opinion to automate or design a smart solution involves 10% understanding the technology, 10 % implementation, and the final 80% is mapping out and designing the solution to actually solve the intended issue and do that without any unwanted consequences.
A practical example
Thus far it has been a lot of words and wide concepts but nothing you as the reader can relate to so let's look at a more specific scenario to explain the issue. Your organization wants to implement a chatbot answering user questions about Microsoft applications. First you need to find the technology. In our case a virtual agent from Microsoft will solve this nicely. Learning to use it might be somewhat challenging but not that time consuming. The implementation of this tool is not that hard either. That leaves us with the final 80% as mentioned above, in our case we need to map out any questions or feature the uses might want to know about and then we need too actually find out how the users will ask the chatbot about these things. So already here you can see the amount of work this will be. Thereafter if you did a bad job on understanding how the users will communicate with the bot, then the users will have bad experiences and therefore won’t use it and instead contact the helpdesk directly and we are back to start. It is possible to fix things as you go but as we all know trust is built over years and torn down in days, so getting a user who had an unpleasant experience back is difficult.
Now that you’ve seen an actual example of how much and what kind of work lies behind implementing a solution like this, you can ask yourself how starting a “fix it with AI” solution in the middle of a project affect the time-frame off the intended project? Automation and AI solutions from Microsoft are great, but you will rarely get the full benefit when implementing it as a quick fix.