Page 8 - EngineerIT February 2022 UPDATED

ARTIFICIAL INTELLIGENCE

How to set your AI projects up

for success

By Robin Fisher, Senior Area Vice President, Salesforce Emerging Markets*

hen business leaders come to picking the right AI If you are seeking to automate 100% of your customer service
project for their company, this often boils down to queries, put simply, you are setting yourself up for failure. However,
Whaving the right ingredients and knowing how to if 25% of your customer service queries are requests to reset a
combine them. Almost every functional AI system today works password, and you want to automate that and take it off your
through some manner of rules that are encoded in the system and agents’ to-do list, that is a reasonable goal.
effectively hold everything together. Another key question is: Can a human do it? Most of the time
Vital to success, too, is the effective communication of their AI can’t do anything that humans can’t do.
teams as data scientists, and repeatedly asking themselves a series There are two reasons why AI projects tend to - and should -
of questions throughout the development process. have uncomfortably long pilot periods.
We can break AI down into ingredients because there’s a whole First, you need to determine whether it actually works the way
menu of things that it can do. When you have an idea of what they it should. From healthcare diagnosis to movie recommendations,
are, it gives you an idea of what its powers are. the context and importance of AI-powered recommendation will
The first ingredient is “yes” and “no” questions. For example, if vary but ultimately you need to share the explanation so your users
I send you an email, are you going to open it or not? These give you will trust it.
a probability of whether something is going to happen. We ask this Second, you need to measure the value of the AI solution
question at every stage of the AI project. versus baseline – human interaction. Think about automating
The second ingredient is numeric prediction. For example, how customer service queries. If your chatbot can’t answer your
many days is it going to take you to pay your bill? Or how long is it customers’ questions, you will end up with frustrated customers
going to take me to fix this person’s refrigerator? who hate your chatbot and end up talking to a human anyway.
Thirdly, we have classifications. For instance, when you take a
picture of your team, ask, “are there people in this picture?” “How Why we need to stay grounded
many people are in this picture?” There are text classifications, too, We know that understanding and using data sets to inform
which you see when you interact with a chatbot. machine learning systems can solve problems more effectively. Yet
The fourth ingredient is conversions - taking information and intuition also has an important role to play in this process.
translating it from one format to another. This could be voice Take for example building a custom prediction for questions
transcription or translation. such as “Will my customer pay his bill late or not?”
Often data scientists in the AI field have a tendency to think
Where to start about algorithms, or maybe slightly higher level abstractions. What
Starting your AI project journey, the fundamental questions you we really need to do is get into our customers’ heads and express
need to ask are: “What data do we have?” and “What concrete the solution to the problem in terms that they will relate to. This
problems can I solve with it?” means that it’s not just about making a recommendation; it’s
Take, for instance, something that every salesperson tracks as a specifically recommending the part that goes into a project. Also,
natural part of his or her job: categorising a lead by giving it a score it’s not just making a prediction, but specifically answering the
of how likely it is to close. Data sets like these are a key source of question, are you going to pay your bill or not?
truth from which to develop an AI-based project. From here you have to decide: if I make that prediction, I give
People want to do all kinds of things with AI capabilities, but if you a probability of the guy paying late, what are we going to do
you don’t have the data, then you have a problem. about it?
Ultimately teams need to stay grounded when considering
How to get from pilot to rollout what problems they should try to solve with AI and what they have
To get the project from the pilot to rollout stage, the first question on hand that can help them do it. It goes back to the question of
you need to address is the problem you are trying to solve. whether humans can do it. If they can, maybe AI is a great way
Am I trying to prioritise peoples’ time? Am I trying to automate to take that task off a human’s plate to free them up for more
something new? Then you can confirm that you have the data for strategic tasks. n
this project or that you can get it.
The next question you need to ask is: Is this a reasonable goal? *rfisher@salesforce.com

EngineerIT | February 2022 | 6