Jumping on the AI bandwagon
AI is a huge area. This article focuses on machine learning.
We’ve all been hearing a lot about AI over the past few years and months. As ‘AI’ transforms from esoteric term to ubiquitous buzzword, many companies are beginning to ask how they can ‘do AI’.
There are a couple of important angles to consider here. Can just anyone build a system equipped with some degree of AI (is it an accessible technology?) and, moreover, why should you? Is there genuine use and merit in it? Is there a reasonable prospect of a healthy return on investment?
First though, what is AI?
The dictionary definitions say things like ‘a branch of computer science dealing with the simulation of intelligent behaviour in computers.’ (Merriam-Webster), or, slightly more comprehensively, ‘The theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages.’ (Lexico.com).
Taking a closer look at the examples in the latter definition, it’s decision-making software that really stands out as having a great business case, so that’s what we’ll consider here. (You’re not likely to want to build your own language translation service from square one and, in many cases, things like speech recognition and visual perception have already been skillfully packaged up into prolific services, such as ANPR (automatic number plate recognition)). You are likely, though, to want to build a system that makes your business somehow smarter.
Using decision making as an example then: can a business with no track record in AI, build or commission their own application to empower better-informed, quicker or more reliable decisions?
Short answer is yes, and this is where we might find ourselves sharing some trade secrets and getting kicked out of the magic circle! You see, we don’t actually need to ‘build’ the AI part of your bespoke system for you. Sure, we need to build it into the system, but we don’t need to start from scratch. Pretty much everything we do to create a tailored software platform for your business comes down to curating and selecting the right components, and hand-coding the unique parts, but – critically – not reinventing the wheel.
Services such as AWS (Amazon Web Services) provide developers with access to proven AI modules, such as Amazon Forecast (for making predictions), Amazon Comprehend (for finding insights and relationships in text) and Amazon Personalize (for machine-learning-driven personalised recommendations). All of these are available as established, continuously improving services on AWS. Similarly, there are leading offerings from the likes of IBM (Watson) and Microsoft too.
All we need to do is use the service to interrogate your existing historical data, plus configure a model that matches your business goals, then use the output either in its raw form, or connected to an application.
The key thing to note is that you can ‘do AI’ and all sorts of services (we’re focusing here on machine learning) are out there, ready for developers to ‘plumb-in’.
The remaining question is, should you be doing AI? Is there a compelling case that supports your organisation’s mission?
The answer is yes if:
- You have (or have access to) a meaningful amount of data (such as customer data)
- That dataset is growing all the time (more data = better learning = better outcomes)
- You need to make reliable decisions that have an impact on your bottom line
- You can describe a ‘model’ that combines historical data with other data, to come up with a pertinent formula
- You’re looking to outsmart the odds by achieving something like:
– Better risk management (insurance)
– Better special offers (retail)
– Better budgetary control (government)
– Better recommendations (entertainment & media)
– Better supply chain management
If you want to gain a massive advantage in your field, have an inkling that AI could help, and want to explore if that inkling could become reality, let’s have a chat!