Kathy Gibson reports from Qlik AI Reality Tour – Artificial intelligence (AI) is here to stay. We’ve moved beyond the hype and companies are now ready to use AI in real business scenarios.
This is according to James Fisher, chief strategy officer at Qlik, who points out that it is still early days, with just 25% of companies having a formalised AI strategy – and just 18% of them realising value.
“But you cannot wait; you have to embrace this road. It will be transformative in ways we have never even imagined.”
Indeed, 97% of executives believe GenAI will transform their enterprises and business.
But AI is a complex landscape and the path to get to a result is unclear. Ninety percent of organisations don’t know how to get started and don’t have a best use case. At the same time, 23% of early adopters have seen negative consequences due to inaccuracy.
New regulations are also increasing anxiety, and 50% of organisations have actually abandoned their AI efforts.
“To be good at AI, you need to play the long game,” says Fisher. “You have got to do the work.”
He outlines three best practices that he believes will help companies achieve success in their AI journeys.
The first best practice is to diversify your path – to mix things up to reduce the risk of failure.
“Do not put all your eggs in one basket, to help reduce the risk of failure,” Fisher says. “Think of all the functions and use cases, and think how you can leverage all the different AI models.
“This will help you focus on key areas, and create tailored solutions that will let you get to value quickly.
“Bear in mind that if you use GenAI in isolated instance, you will get limited value.”
At the same time, GenAI might not be the solution to everything.
“Despite the hype, it is not all about GenAI,” Fisher says. “If you use GenAI in the right way, as part of a broader AI strategy, you can help create momentum.”
There is still a huge amount of untouched value to be had from traditional AI, which is still vastly underused. “You could get better value by taking a broader view.”
A key part of diversifying the path is that more than just AI experts need to be involved. To be successful, organisations should leverage human-centred skills as well as technology-centred skills. “Both are critical for success,” Fisher says.
The second best practice is to walk before you run.
This means getting data ready for AI, following six principles of diversity with no bias, with data that is timely, accurate, secure, discoverable and consumable.
At the same time, companies are urged to start taking advantage of AI in existing applications for automated insights, intelligent alerting, predictive analytics, natural language and content generation.
Fisher cautions that smaller models can deliver faster value, with less risk and complexity that a large language model. They are faster to train, deploy and run, less expensive, easier to integrate into existing systems, and can be designed to have specialist capabilities for better results.
The third best practice is don’t skip steps, but establish a responsible AI framework.
This includes an AI policy that outlines the organisation’s principles, guidelines and standards; data governance with systems and process to ensure quality, integrity and security; a governance committee that oversees ethical use, regulatory compliance, risk management and strategic alignment; and vendor control to hold vendors accountable for their AI practices.
Unstructured data is a big untapped opportunity in many organisations, with 80% of the world’s data existing in an unstructured format. “Your AI models need all the right contextual data to generation content and provide answers that are relevant and accurate,” Fisher says.
The final step is to build an ecosystem that you can lean on. Fisher explains that this includes internal players like AI engineers, AI strategists, data scientists, AI power users and domain specialists. These internal resources need to work with expert players like a AI council, a customer advisory board, technology partners, an AI ethics committee and peer groups.
“We don’t know here IA is going to take us,” he says. “It will change, so don’t get locked into one platform. The ecosystem you enable around AI will be critically important.”