Blog Post - Hugh Fletcher, Apr 27 2017

State of the Artificial Intelligence for Businesses

State of the Artificial Intelligence for Businesses

A few years ago, the halls of client organisations rang with the demand for “apps”. No strategy was complete without “an app for that”. Fast forward a few years, and one of the new buzz terms doing the rounds is AI – artificial intelligence – with organisations looking to find an AI solution for many of their problems.

But do people really know what artificial intelligence is? Is it really the answer? And what is the question?

One of the major challenges with the mass market adoption of AI is the term itself. It’s too broad and too generic. Say AI and most of us think of Skynet or Artificial General Intelligence (AGI) to give it its official term. But in reality, the most common and best usage of AI is in circumstances which are very narrow. Sure it’s intelligent. Sure it might start to replace some jobs currently performed by humans. But it’s not the human alternative that some have led us to believe…yet.

AI really isn’t about conscious intelligence – despite what the scaremongering media would have us believe with its tales of imminent AI related takeovers. Mostly, AI’s best implementation is about laser-guided and narrow expertise to solve a specific task.

AI – what is it?

To understand how best to use it, it might be worth understanding what AI actually is. In base terms, it’s a load of computer code. It’s a computer task, an algorithm that solves a problem, and that constantly learns to give the best result. Understand it like this and it’s easier to understand how best to deploy it (it also helps to allay some of those paranoias!)

But like most things, it needs input. And AI’s input is data. The more, and the more specific, the better. AI is a hungry beast. Feed it and nourish it, and it becomes a beast that can be tamed for your usage. Force feed it bad data, and it’s a beast you don’t want to unleash.

And key to AI is that it learns. And it can learn in multiple ways: through data learning, through human supervision, and through self-learning and in some cases, a combination of all these things.

Is Watson AI, and is AI Watson?

Perhaps the most famous AI engine, is IBM’s Watson. Watson is the umbrella term for a load of services provided by IBM that use AI to do a variety of tasks. It is without doubt that IBM has done an exceptional job of branding and promoting its AI solution, and in some cases its services are some of the best-in-class. These include natural language processing (NLP), retrieve and rank, and discovery to name just a few.

What also sets Watson apart is that IBM has very cannily started to integrate its Watson branded AI into its portfolio of products. WebSphere Commerce, IBM’s ecommerce system, has now been rebranded Watson commerce and will come with AI embedded; a sign of things to come and an illustration of how the market penetration of AI capability will grow with or without specific market demand.

So while Watson is certainly a strong exponent of AI, it isn’t the only solution.

What should my business do about AI?

What’s important to remember is that AI can help a business solve problems. But it is the solution to the right problem, and not a solution looking for a problem. Like most tools, used in the right circumstance and it’s invaluable; use it in the wrong scenario and it’s useless. As with most of the best solutions, the right course of action is for businesses to identify issues, tasks and problems that exist within their business, and then decide whether AI can help to solve these - not the other way round.

So where does AI work best?

As mentioned, AI is hungry for data. It works best where there is data readily available or where somebody is already using AI to solve the problem – for instance security, marketing or identity recognition. And it’s important that artificial intelligence isn’t about perfection, it’s about making something better over time. So, consider that when considering what the right solution is.

And what about writing your own AI? Well yes, you can. And in certain circumstances that might be the right course of action – for instance if your organisation has security concerns around its data, or you have a particular problem which isn’t being addressed by anyone else. Or, whisper it quietly, if you just want to show off and say that you’ve developed your own AI.

But it would be commercially prudent, in almost all cases, for the first step to be to see if the AI you need has already been developed by someone else already.

Approach the usage of AI like a normal project: identify your requirements, identify the need for AI and then identify, via your existing digital partners, the AI providers who could help you.

And what of the future?

As we’ve seen with the Internet of Things, whose growth is down more to its inclusion in products, irrespective of whether its customers proactively want or search for it, the expansion of AI is an unstoppable inevitability. Like Nest and its learning thermostat, more and more AI capability will become embedded in products and services until it will be hard to remember a day when we lived without it.

To find out more about AI get in touch with our Global Head of Consultancy & Innovation, Hugh Fletcher.

AI is one of the key themes that Salmon called out at its annual ecommerce event Commerce 2020. Download the 20 for 20 Trends ebook for other key trends to have on your radar.

Interested in more topics like this? Go to What We Think or sign up for our newsletter.