Don’t dismiss humans in an AI world

In a future where artificial intelligence (AI) becomes embedded into the financial services industry, human intelligence will be essential in bringing out the best in this area, according to Societe Generale’s UK Chief Digital Officer, Sohail Raja.

“The successes we have had so far in building API (application programme interface) solutions through SG Markets, our business-to-business client interface, have been the result of answering a specific client need,” said Sohail, speaking at the AI Summit in London last week. It is after all people who make the business decisions and build strategies for their businesses, even if the machines are executing in some cases.

Sohail’s comments were echoed by his fellow panellists who discussed technology’s need for human input to enable best business outcomes. The panel, featuring representatives from Microsoft, UBS and Lloyds of London, agreed that machines will struggle to tackle the more esoteric subjects surrounding the use of AI in financial services, like trying to analyse whether the decisions they make are ‘right’, or deciding what the ethical or societal impact is. They discussed the likely emergence of a new layer of personnel whose job it is to build and manage the governance around AI in financial services.

It is the people who understand the data that make the solutions work, not the machines themselves, explained Sohail. “We put data science experts within our project teams. In the bank, we have some very strong mathematicians to work on quants. They work alongside data scientists to enhance their models,” he added.

“Having the human aspect will continue to be important going forward, particularly as we are trying to fine tune these models,” said Sohail. “These models could be, as an example, 99.3% accurate. It is that 0.7% that is going to be the challenge. We are still going to need that human aspect - human intelligence - to bring that out.”