Firms need to take a ground-up approach and get staff on board before they embark on AI projects, a FinovateEurope panelist advised.
The statement was made by Dr. Anandhi Dhukaram, founder and CEO of AI consultancy and training firm Esdha, at the ‘AI In Action – Examining User Cases Of AI Transforming The Future Of Finance’ panel.
Dhukaram reported that the large majority of AI projects fail, and that this is generally due to the people rather than technology side of the initiatives. She emphasised the importance of “starting with the right knowledge,” ensuring that employees are given the right training, are a part of the development process and, crucially, trust the AI.
In order to build that trust, Jean-Pierre Sleiman, head of digital operations at neobank N26, stated that education is key. Staff must understand what AI can and cannot do, with use cases identified and clearly outlined.
Dhukaram agreed, advising companies to “start slow and experiment”, and stressed the importance of failure in the growth and development process. ”Don’t treat AI as a goldrush,” she warned, reaffirming the importance of finding valuable use cases before beginning a project.
Joel Perlman, co-founder?and?senior managing partner at OakNorth, raised the point that many distrust AI out of fear of being replaced, particularly in the face of generative AI. Again returning to the importance of education, he advocated for “open conversations” about what AI will mean for departments.
On the topic of generative AI and its impact on human jobs, Dhukaram argued that it could be used to offer more personalised and efficient customer experiences. While it is not suitable in all cases, generative AI could be used in low-stakes roles that require general and less emotional activities.
Sleiman added that continual auditing is essential to AI projects, with mechanisms put in place to ensure that the technology remains relevant and effective. He added that this process is often overlooked, due to the fact that the majority of projects fail before they reach this stage.
Closing the panel, the speakers were asked whether they thought AI, as it stands, is an ‘infant’, ‘toddler’ or ‘adult’. Perlman answered with the former, while Dhukaram argued that it is dependent on the organisation. Sleiman believed it to be a ‘toddler’, concluding that in order for the technology to continue to develop, there needs to be collaboration across the industry to identify use cases and find “safe and effective” ways to employ AI.
Image: Kaikoro/stock.adobe.com 