Modern AI was born when rules-based software was no longer able to tackle the problems the computing world wanted to solve. Think about it, when you have colossal volumes of data, it isn’t possible to code every condition the program needs to measure.
Most AI concepts and algorithms have been around for a long time but, up until the last few years, they solely remained the tools of statisticians, mathematicians, and data scientists. With the growing availability of AI cloud services and ever-evolving sets of open-source libraries like Scikit-Learn, PyTorch and TensorFlow, AI is becoming more and more democratised and this has led to greater adoption by organisations of all sizes.
How is AI being used?
One of the most common facts (that people still find hard to believe) is that AI is all around us. It exists in our day-to-day lives seamlessly, managing various technologies in real-time. Let’s have a look at a specific example;
UPS – Route Plotting: In supply chain deliveries, every minute and mile matters. UPS uses an AI-powered GPS tool called ORION (On-road Integrated Optimization and Navigation) to create the most efficient routes for its fleet. Customers, drivers and vehicles submit data to the machine, which then uses algorithms to create the most optimal routes. Instead of back-tracking or getting stuck in traffic, ORION helps drivers make their deliveries on time and in the most efficient manner. The routes can even be changed on the go depending on road conditions and other factors. Optimizing delivery routes has a huge impact on all areas of UPS’ business, from saving time and money to reducing emissions and wear and tear on its trucks. With ORION, UPS estimates it can reduce its delivery miles by 100 million. Those savings can add up, especially because UPS predicts that for every mile its drivers cut from their daily routes, the company saves $50 million a year.
This is just one example of an otherwise endless list of how AI is applied in our day-to-day lives.
Why, What and When?
To successfully implement AI, an organisation really needs to understand the answers to these key questions:
Why should we implement AI?
This isn’t a complex answer. AI, like many other things, should be implemented to automate or improve the efficiency and accuracy of existing processes. It should save your valuable resources time so that they can focus on things that humans, not computers, do best. AI is the next logical step in the way we do things and should be treated as any other tool in an organisation’s toolbox.
How about when?
As with any tool, it depends on the problem you are trying to solve. AI is not a panacea; the simplest solution is usually the best one. Whatever the answer is, it should ideally be aligned with your organisational strategy. If an AI solution is chosen, then how it is implemented and used should also be ethical.
What project or area of the business should I start with?
Start with the “low-hanging fruit” as easier projects will often give good returns with minimal effort and will push others within the organisation to start thinking more about their data and potential that can be unlocked. Think along the lines of a chatbot to answer everyday questions or text recognition for scanning documents. Lots of these frameworks have already been developed and allow organisations to dip their toes with minimal cost and effort.
Even some of the more seemingly advanced AI algorithms are getting much easier to implement due to the availability of Cloud services and open-source libraries. For example, trying to recognise data that is deviating from a standard pattern is much simpler than predictive analytics.
Do I need a data platform to facilitate AI?
For some simpler use cases (like chatbots and text recognition), you won’t need to build organisational-wide data platforms to facilitate the use of AI. However, as use cases become more advanced, there is a requirement to have a platform where this data is readily available to be processed by your algorithms.
The majority of AI tools require a fair amount of data in order to find patterns within it. These patterns are what form the basis of artificial knowledge which is used to make predictions or classifications. This is all well and good, but without accurate, clean data, the old saying “garbage in, garbage out” still rings true.
Unfortunately, there is no magic bullet for this problem and data wrangling (cleaning and preparing data) on machine learning projects can typically take up to a whopping 80% of project time. However, if the data is already clean and stored in an easily accessible way, it will really accelerate your project. This is where a robust, scalable data platform comes into play and building this foundation from the outset will save lots of pain down the road. When considering a platform, think about the following:
- What portion of my data is unstructured or semi-structured?
- What portion of my data is structured and already imbued with business context?
- Do I need batch, real-time processing or both?
- What capabilities do I have access to in-house or through a supplier?
- What do I plan to do with the data that has been processed by AI?
The answers to these types of questions will start to form the basis of what your data platform should look like.
At Neueda, we’re extremely passionate about AI and love seeing organisations implement this successfully to achieve truly transformational results. We know that AI is a scary concept but define your use cases, pick your low-hanging fruit and begin to build your foundation from the outset, and you’ll be walking the walk in no time.
Demystifying Data Workshop
Neueda is currently offering a FREE Demystifying Data Workshop, where we will demonstrate the value a Modern Data Platform can bring to your organisation.