From Cloud to IoT, the majority of the big technology adoptions of the last few years have been ‘plug and play’ – to some extent, they work out of the box. This is not the case with artificial intelligence (AI).
Training AI, for example, takes time – it needs to be shown hundreds of images of a certain object just to learn how to recognise it. It is, therefore, important for businesses to set realistic timescales that make it clear how long it might take for AI to be fully implemented and start making a tangible difference within an enterprise. Organisations need this AI roadmap to succeed in implementing the technology across their business, while equipping their human workforce to work effectively alongside it.
The AI roadmap
“It’s critical that the people driving AI adoption within an enterprise remain realistic about what it is capable of, and the time it can take to start making a positive impact,” says Kalyan Kumar, Corporate Vice President and CTO at HCL Technologies (who is often seen sporting an ‘I ❤ AI’ badge on his lapel!).
AI has huge ‘potential’ to increase corporate profitability
Obtaining value from AI will demand recognition and action from the top of the company; therefore, the benefits must be made tangible to the C-suite and a roadmap is essential. Read here
“While enthusiasm about AI is great, over-promising could open up a chasm between expectations of what it will deliver, and the short-term realities. Taking the time to set out a roadmap for AI can play a crucial role in narrowing the gap between expectations and reality within an enterprise.”
AI, as mentioned, is not like the other technologies to have been consumed by the enterprise in recent years. The technology needs time, and this can often be frustrating for stakeholders wanting to see immediate results. Like so many technology-related business or transformation strategies, artificial intelligence is a marathon, not a sprint.
Business leaders need to understand this, “and realise that the adoption of AI and machine learning — for introducing potential new business models, enabling process automation and achieving cost efficiencies — is a journey,” explains Kumar.
The AI journey
To establish a focused AI practice, every business leader should ask their teams to begin by identifying and defining the business problem and challenges that could potentially be solved using AI technologies, according to Kumar.
The next stage is “to validate the existence of systems and processes that can help in capturing the data required for training new AI models, then ensure that data quality is at optimum levels, to build unbiased and accurate models.” he explains.
The AI model
“When it comes to building the AI model, organisations should look to leverage the open source AI models that are available, or, build their own custom AI models. It is critical to get continuous feedback from the business and domain experts throughout this process, which will be invaluable in building the right model,” says Kumar.
Artificial intelligence: Transforming the insurance industry
“Once the model is in place, organisations will need to perform AI model validation exercises using validation datasets, which will further help in re-tuning the model. For this to be effective, it’s important to define various performance measures that will help organisations to analyse the success of their models.”
“Finally, organisations should continuously monitor the validity of their AI model as circumstances and conditions change, and re-configure it as required. There are several factors that could require AI models to be regularly updated.”