How can organisations ensure that the adoption of artificial intelligence will drive the desired business outcomes? Kalyan Kumar, Corporate Vice President and CTO at HCL Technologies, provides his insight 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) and machine learning.
Training artificial intelligence (AI), for example, takes time – it needs to be shown hundreds of images of a certain object just to learn how to recognize it. It is, therefore, important for businesses to set realistic timelines that make it clear how long it might take for AI to be fully implemented and start making a tangible difference within an enterprise. Organizations need this artificial intelligence (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 artificial intelligence (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,” Kalyan Kumar, Corporate Vice President and CTO, HCL Technologies.
“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 technologies 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 AI technologies need 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 realize 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, organizations 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.
“Once the model is in place, organizations will need to perform AI model validation exercises using validation datasets, which will further help in retuning the model. For this to be effective, it’s important to define various performance measures that will help organizations analyze the success of their models.”
“Finally, organizations should continuously monitor the validity of their AI model as circumstances and conditions change, and reconfigure it as required. There are several factors that could require AI models to be regularly updated.”
