A few years after its initial boom, artificial intelligence (AI) remains a huge fashionable word in the Fintech industry, because each company examines a new way of integrating technology into its infrastructure to gain a competitive advantage. Explore how they do this in 2025, Finch times highlights some of the greatest themes of AI in February.
Regulations are a great subject of discussion in the AI world, different countries adopting different approaches to make the technology police. However, even a company fully in accordance with the best intentions can be a failure with AI. But from the point of view of financial decision-making, how does failure have an impact on the decision-making process? Do companies depend too much on technology and find themselves lost after a failure? We hear industry experts to discover it.
AI monitoring to ensure that failures can be treated early
Maya Mikhailov,, CEO of Savvi IAThe company helping organizations to deploy AI, notes the various ways in which AI can fail a company in the decision -making process. She explains how the simple implementation of AI is not enough for technology to work constantly at its best – it must be constantly monitored.
“There are several types of failures with regard to automatic learning in financial decision -making – biases due to quality problems in underlying data sets, data drift due to a lack recycling models and aberrant scenarios such as “ Black Swan ” events.
“The most basic failure is that if the model is formed on a bad set of historical data which contains coded prejudices – these are not necessarily social prejudices, they can also be a bad decision by people who become Coded in the data and then reflected in the model.
“In addition, sometimes the models fail due to data drift – when the historical models on which they are formed no longer apply or no longer change. For example, if a model is constructed to predict loans delinquency and interest rates are starting to increase or lower, the historical model no longer reflects reality. The model can start to see increasing errors in its ability to predict delinquency if it is not recycled on these new changing conditions.
“Finally, the models are fighting with things they have never seen before, think that Covid. Black swan events often cause failures because there were no data on which to train.
“In a well-built AI system, back tests, railings and continuous recycling are essential to prevent failure or correction errors. Of all types of AI, the ML is the most established and commonly used financial decision -making so that companies are better equipped with the management of ML results and failures. »»
Excessive dependence can be expensive
According to Jacques FrancisCEO Management of paradigmal assetsThe asset management company, one of the biggest problems that AI failures can have on a company is to empty resources. Exploring how it can be avoided, he says: “Sometimes, even the most intelligent AI can be wrong – like when your computer stops in a game.
“The erroneous financial decisions of artificial intelligence can be expensive and cause great stress. Forgetting that people have to keep an eye on things, I saw companies become too dependent only on AI. This is why at Paradigme, we combine intelligent people and smart technologies. It’s like managing a superhero team where each member has special skills. We see that artificial intelligence helps us but does not take over.
“Although this can be exciting to apply AI in finance, we are still cautious to balance technology with a wise and old -fashioned human judgment. Finally, even the robots want a friend. »»
Excluding honest customers
AI has the potential to do the customer experience Incredibly simple and pleasant. However, from a loans point of view, if AI is used badly, those who deserve a loan may not be entitled to one. Yaacov Martinco-founder and CEO JifitiThe integrated loan platform, explains how humans must supervise technology to ensure that customers never lose for any offer.
“When AI fails in financial decision -making for loans to consumers and businesses, the consequences can be significant, which has an impact on all stakeholders. Although loans fueled by AI have the potential to accelerate credit assessments, improve risk management and personalize loan offers, these advantages may include risks if they do not supervise correctly and s ‘They are too linked to banks and lenders.
“Although AI applies much wider data parameters, accelerated processes, are more advanced than traditional algorithms, and” teaches “itself according to past performance models, it runs the risk of functioning as a “ Black box ‘decisions, leading to decision -making failures.
“His dependence on historical data models and the lack of subjective” human “surveillance can strengthen biases, potentially refusing credit to deserving people. Landers give too much confidence in AI without monitoring and the risk of appropriate regulation exposing borrowers to confidentiality problems and unjust loans.
“The regulations are crucial to protect transparency, fairness and data security, and provide controls and counterweights. In addition, to ensure that credit supply is in fact in accordance with the principles of the lender and avoid colossal classes, there is a final need for periodic sampling by a human.
“As AI becomes more widespread in loans, financial institutions must avoid complacency and prioritize ethical implementation.”
Understanding a trip with AI does not need to be done alone
Vikas Sharmasenior vice-president and chief of practice for benches and capital markets ExlThe digital acceleration partner highlights an enormous point that companies must understand even before applying AI: becoming an expert in technology does not occur overnight, so to ensure that failures Avoid, companies should try to associate experts.
“The risks associated with the failure of the AI in financial decision -making are far too serious not to take into account the guarantees and the controls governing. These risks include, without limiting themselves, the impact of customer financing, regulatory risk, reputation damage and operational challenges. Without reliable controls and an evolutionary framework, smaller failures can cascade to cause systemic instability and significant financial losses.
“While the financial industry rushes to incorporate AI into their processes and products, the fintechs are at the forefront of this change. Fintechs are constantly experiencing to gain the data difference they have with their large bank peers – and the advent of AI promises to be the final solution.
“Our experience at ES suggests that most of the fintechs should launch their AI initiatives with a partner company that specializes in the evaluation, design and implementation of scalable IA routes. The first step to implement these roadmaps is to configure clear railings and define an AI framework with loop humans. The integration of human surveillance into each critical decision point increases responsibility and reduces possible failures.
“After all, companies realize that they use this innovative technology to develop their membership base and improve customer satisfaction – which will be affected if strong governance controls are missing.”
Robust frameworks
Brand DearDirector of industry banking solutions to FinchosThe company offering a low code approach to help others scan, also noted the different types of failures that can occur when fintechs are based too much on AI and shared its solution. He explains: “The likely consequences of AI failures in decision -making raise important concerns concerning the overcoming of these technologies. For example, there is a disturbing possibility that certain companies become dependent on AI systems without maintaining robust human surveillance.
“Some financial institutions have reduced their human risk management teams, creating potential gaps in monitoring AI systems and dangerous unique failures.
“Automation bias is also a risk in financial decision -making, which makes humans trust the decisions generated by computer despite contradictions with their judgments, which means that obvious mistakes are not disputed because ‘They come from AI or traditional internal systems.
“In response to these increased risks, financial institutions must develop more robust executives to manage the deployments of AI, including better test protocols and clearer lighter structures. Regulatory organizations are increasingly focused on AI governance in financial institutions, recognizing the systemic risks of exceeding these technologies which can lead to new transparency and human surveillance requirements in financial decisions focused on AI.
“In the end, the key is to find the right balance between taking advantage of the increasing AI capabilities while maintaining sufficient human surveillance to prevent potential failures. Financial institutions should consider AI as a tool to improve human decision -making, and not replace it entirely. »»
