The following panel discussion was moderated by Debi Bell-Hosking, featuring Dr. Janet Bastiman, Chief Data Scientist at Napier AI; Stuart McDowell, UK CIO at Societe Generale; and David Tracy, Head of Data Products at Smart Data Foundry.
The focus of the session was to debunk five myths surrounding AI, facilitated through audience participation via interactive polls and commentary.
Myth 1: Human-in-the-loop is not a crucial role in financial services.
The first myth discussed was the belief that human-in-the-loop (HITL) processes will not be essential in the future. An overwhelming 84% of the audience disagreed with this notion, as did the panelists. Dr. Bastiman stated, "We will need human-in-the-loop going forward. Many regions are beginning to implement regulations in financial services, which necessitates human decision-making." She referenced an earlier speaker’s perspective, suggesting we should think of it as "AI in the human process."
McDowell shared insights from his trading experience, emphasizing that while algorithmic trading and machine learning were initially seen as threats to human roles, they ultimately augment rather than replace human traders, who remain accountable to regulators and clients.
Tracy noted that while low-risk and non-invasive automation may reduce human involvement, understanding context and nuance remains crucial. An audience member from Swift commented that as trust in AI grows, the need for human involvement may diminish.
Myth 2: Soft skills will become unimportant in financial services.
The discussion highlighted crucial soft skills such as critical thinking, interpersonal skills, negotiation, and collaboration. Dr. Bastiman argued against the idea that these skills will diminish in importance, sharing an example from the financial crime sector where critical thinking is vital for identifying suspicious transaction patterns.
McDowell added that effective management relies on emotional intelligence (EQ) rather than just intellectual capability (IQ), as people prefer human interaction over impersonal evaluations.
An audience member from Santander emphasized the need for AI decisions to be reviewed through a human perspective, while another from S&P Global highlighted the importance of inquisitiveness and asking the right questions to utilize AI effectively in the future. Tracy acknowledged that while soft skills will remain important, their nature may evolve.
Myth 3: AI isn’t human and doesn’t have bias.
The panel unanimously disagreed with the belief that AI is free from bias. McDowell pointed out that bias in AI stems from the training data used. Dr. Bastiman elaborated that mathematical proofs reveal that unbalanced data sets lead to biased models. Tracy expressed concern about human bias influencing AI, suggesting that our diligence in checking AI outputs may vary.
Myth 4: Sensitive data is safer with advancing technology than with humans.
The audience and panelists had differing views on whether technology offers greater safety for sensitive data than human handling. Dr. Bastiman noted that users often compromise security measures, such as creating weak passwords. She cautioned that with advancements like quantum computing, maintaining data protection will become increasingly challenging.
Myth 5: Regulation can’t keep up with AI and enforce the role of the human.
The final myth tackled was the assertion that regulation cannot keep pace with AI. A significant 87% of the audience agreed with this statement. Tracy expressed concern over this perception, while McDowell pointed out that regulatory frameworks in financial services are evolving. He emphasized that regulatory principles apply equally, whether an interaction involves a human or generative AI.
Dr. Bastiman concluded by acknowledging that although there are times regulation may lag, it has the potential to adapt, provided it avoids being overly specific about technology, allowing it to remain relevant as the landscape changes.