The definition of competition, value and performance is going to be irretrievably changed for companies that can scale AI.
Artificial Intelligence is really happening – and it has big implications.
There’s now a collective fascination with the potential for scaling AI in both public and private sectors, and, in the last year, it has gone from being an interesting topic to front and centre of every boardroom agenda.
The big technology vendors jumped on-board very quickly, but we’ve seen a further acceleration; there isn’t a corner of the economy – oil and gas, power and utilities, media and health – that hasn’t been enraptured by the topic.
Given the current hype, I thought it was time to offer some practical advice. So if you’re wondering how to apply AI in business, here’s what you should be thinking about:
Take a thoughtful approach
AI doesn’t automate jobs, it automates the tasks that add up to jobs and makes work more meaningful by liberating employees from repetitive tasks.
It’s possible to calculate how at risk a job is and plan for automation by retraining, reskilling and redeploying resources. That’s a far more considered approach than implementing automation and worrying about the implications later.
Unlike software that simply executes a command, AI ‘learns’ to perform certain tasks. The risk is that AI systems acquire some kind of subtle bias that may not be obvious. So one of the challenges of AI is to be able to continually monitor and find potential failures in the learning process, so you don’t end up with unwanted outcomes.
If you know that risks exist, you can plan for them by setting up governance systems and monitoring systems.
Balance enthusiasm with pragmatism
Despite the current enthusiasm, many business problems and opportunities do not need AI. The technology introduces complexity and risk, so limiting its use to very specific issues is a good thing.
For example, a simple rules-based chat bot answers customer questions without the need for complex technology and solves a problem without AI. It’s a high value, low cost solution.
The technology behind sophisticated AI experiences is incredibly complex and designing them requires a great degree of skill.
AI is not like IT, even though there’s a lot of technology involved – it’s fundamentally about science, where you start with a problem then come up with a hypothesis. You have to understand what forms of AI might work, which domains to use, what kind of architecture, and, beyond that, the best way to train and optimise the network.
The companies that are doing AI well don’t apply an IT model – they apply a continuous innovation model that requires cultural and leadership change.
Set realistic expectations
If you don’t understand how transformative AI can be, you’re less likely to apply it appropriately and achieve value from investment. The single biggest risk is doing nothing. Any notion that AI is not real, is not happening or is all hype, is misguided.
Where realism is needed is in the level of effort required to get value from AI in business. Most vendors undersell the amount of effort it takes to turn AI into value, so there can be unrealistic expectations of the time involved and the potential return.
You don’t implement AI, you apply it. If you don’t understand the difference then you are likely to underachieve because the returns you expect will be in line with those from a mature technology. Having said that, parts of the AI spectrum, like facial recognition, are maturing to the point where they can be deployed at scale in a fairly robust, reliable way.
Don’t believe everything you read, but don’t write it off either. The definition of competition, value and performance is going to be irretrievably changed for companies that can scale AI. Those that can’t will not be competitive in any part of their business, not perhaps by tomorrow, but certainly within the next three to five years.
AI is complex and it takes time to develop robust capabilities – so start now.