El impacto de la IA generativa en el riesgo crediticio

La inteligencia artificial generativa (IA) está transformando el panorama financiero y consolidándose como una herramienta clave para redefinir la forma en que se evalúa el riesgo crediticio. Al superar los modelos lineales tradicionales, esta tecnología permite analizar enormes volúmenes de datos no estructurados, como historiales de pago o comportamientos digitales, y detectar señales sutiles que mejoran la precisión de las predicciones.

Más allá de ofrecer evaluaciones de riesgo más completas y precisas, la IA generativa también agiliza tareas manuales y repetitivas, especialmente en la gestión de documentación corporativa compleja. Esto permite a las entidades financieras operar con mayor rapidez y eficiencia en un entorno cada vez más competitivo.

El vídeo destaca cómo, durante años, los bancos han confiado en modelos de credit scoring basados en relaciones lineales. Sin embargo, la capacidad de la IA generativa para incorporar datos no estructurados incrementa significativamente el poder predictivo de estos modelos.

Los expertos Tiago Rodrigues de Freitas, socio y head of data and analytics Iberia at Oliver Wyman, y Ignasi Barri, Global Head of AI and Data en GFT Technologies, subrayan que, además de mejorar la calidad y la capacidad de discriminación, la IA generativa acelera los procesos manuales, reduce el error humano y optimiza el tiempo de análisis, aunque enfrentan retos como la rápida evolución tecnológica que genera cautela, la necesidad de modelos explicables para ganar confianza y la formación del personal para aprovechar estas soluciones eficazmente.

Como ejemplo práctico de aplicación, los expertos mencionan el Credit Risk Assistant, una herramienta desarrollada conjuntamente por GFT y Oliver Wyman, que permite a los analistas de crédito procesar de manera más eficiente múltiples fuentes de datos en distintos idiomas.

De cara al futuro, ambos expertos muestran una gran confianza en que la inteligencia artificial dejará de ser un mero apoyo a los procesos de crédito para convertirse en el motor principal de su transformación, marcando el inicio de una nueva etapa para el sector financiero.

Ignasi Barri: Tiago, we're hearing a lot lately, around generative AI, but in particular generative AI for credit worthiness assessment. Do you think that this is a potential way to improve that process?

Tiago Rodrigues: Absolutely, banks have relied for decades on credit scoring and rating models that, while reasonably good at discriminating clients, operate through very linear relationships between variables and outputs; what we’re seeing with generative AI is the ability to incorporate unstructured data — such as payment history, transactions, and behavior on Amazon and e-commerce platforms — which can provide subtle signals that generative AI can capture, thereby increasing the predictive power of credit models, improving quality and discrimination, while also accelerating processes related to manual work, especially in the corporate world where extensive documentation must be reviewed.

Ignasi: Benefits always come with certain barriers, so what do you think are the major challenges that banks and financial institutions face when trying to adopt these technologies in their processes?

Tiago: I think the first challenge is the speed of change, which is so rapid that it makes clients defensive and hesitant to fully embrace new technologies, waiting to see a final product even though the process is continuously evolving; the second challenge involves explainability — understanding what the technology delivers — which has improved significantly over the past year through techniques we’ve developed and applied in solutions like the credit risk assistant to help clients interpret outputs, though some resistance remains; and the third challenge is the level of adoption, as the workforce needs time to acquire new skills to understand and work with these solutions, a hurdle that persists today but is likely to diminish over the next two to three years.

Ignasi: To increase adoption, demonstrating prior experience is essential; that’s why GFT and Oliver Wyman created the generative AI-based credit risk assistant — can you tell us more about it?

Tiago: I can. This credit risk assistant is a fantastic example of how generative AI can be rapidly applied to help credit analysts process large volumes of information, such as annual reports, technical statements, and transactional data, extracting valuable signals and insights for client assessment; tasks that were previously done manually, prone to human error, and time-consuming are now streamlined, delivering appreciated results, and notably, despite initial concerns about language barriers, it has been successfully implemented in French, Arabic, Japanese, Spanish, and Portuguese without any issues.

Ignasi: That’s wonderful. Looking ahead, we will see the credit sector completely evolve and transform through the use of AI?

Tiago: I believe we are only seeing the tip of the iceberg, as current use cases like the one I described involve implementing AI within existing processes, but moving forward, these processes will continue to evolve and be reengineered so that generative AI is not just a support tool but the central enabler, and based on the evidence I’ve seen, I am convinced this transformation is inevitable.

Ignasi: So, exciting times ahead, right?

Tiago: Absolutely.

Ignasi: Thank you, Tiago, for sharing these valuable insights and for participating in our quarter.

Tiago: No, thank you for inviting me. As you can see, I am a passionate believer in this topic, and I truly think it will change our future.

Ignasi: Thank you again.

La inteligencia artificial generativa (IA) está transformando el panorama financiero y consolidándose como una herramienta clave para redefinir la forma en que se evalúa el riesgo crediticio. Al superar los modelos lineales tradicionales, esta tecnología permite analizar enormes volúmenes de datos no estructurados, como historiales de pago o comportamientos digitales, y detectar señales sutiles que mejoran la precisión de las predicciones.

Más allá de ofrecer evaluaciones de riesgo más completas y precisas, la IA generativa también agiliza tareas manuales y repetitivas, especialmente en la gestión de documentación corporativa compleja. Esto permite a las entidades financieras operar con mayor rapidez y eficiencia en un entorno cada vez más competitivo.

El vídeo destaca cómo, durante años, los bancos han confiado en modelos de credit scoring basados en relaciones lineales. Sin embargo, la capacidad de la IA generativa para incorporar datos no estructurados incrementa significativamente el poder predictivo de estos modelos.

Los expertos Tiago Rodrigues de Freitas, socio y head of data and analytics Iberia at Oliver Wyman, y Ignasi Barri, Global Head of AI and Data en GFT Technologies, subrayan que, además de mejorar la calidad y la capacidad de discriminación, la IA generativa acelera los procesos manuales, reduce el error humano y optimiza el tiempo de análisis, aunque enfrentan retos como la rápida evolución tecnológica que genera cautela, la necesidad de modelos explicables para ganar confianza y la formación del personal para aprovechar estas soluciones eficazmente.

Como ejemplo práctico de aplicación, los expertos mencionan el Credit Risk Assistant, una herramienta desarrollada conjuntamente por GFT y Oliver Wyman, que permite a los analistas de crédito procesar de manera más eficiente múltiples fuentes de datos en distintos idiomas.

De cara al futuro, ambos expertos muestran una gran confianza en que la inteligencia artificial dejará de ser un mero apoyo a los procesos de crédito para convertirse en el motor principal de su transformación, marcando el inicio de una nueva etapa para el sector financiero.

Ignasi Barri: Tiago, we're hearing a lot lately, around generative AI, but in particular generative AI for credit worthiness assessment. Do you think that this is a potential way to improve that process?

Tiago Rodrigues: Absolutely, banks have relied for decades on credit scoring and rating models that, while reasonably good at discriminating clients, operate through very linear relationships between variables and outputs; what we’re seeing with generative AI is the ability to incorporate unstructured data — such as payment history, transactions, and behavior on Amazon and e-commerce platforms — which can provide subtle signals that generative AI can capture, thereby increasing the predictive power of credit models, improving quality and discrimination, while also accelerating processes related to manual work, especially in the corporate world where extensive documentation must be reviewed.

Ignasi: Benefits always come with certain barriers, so what do you think are the major challenges that banks and financial institutions face when trying to adopt these technologies in their processes?

Tiago: I think the first challenge is the speed of change, which is so rapid that it makes clients defensive and hesitant to fully embrace new technologies, waiting to see a final product even though the process is continuously evolving; the second challenge involves explainability — understanding what the technology delivers — which has improved significantly over the past year through techniques we’ve developed and applied in solutions like the credit risk assistant to help clients interpret outputs, though some resistance remains; and the third challenge is the level of adoption, as the workforce needs time to acquire new skills to understand and work with these solutions, a hurdle that persists today but is likely to diminish over the next two to three years.

Ignasi: To increase adoption, demonstrating prior experience is essential; that’s why GFT and Oliver Wyman created the generative AI-based credit risk assistant — can you tell us more about it?

Tiago: I can. This credit risk assistant is a fantastic example of how generative AI can be rapidly applied to help credit analysts process large volumes of information, such as annual reports, technical statements, and transactional data, extracting valuable signals and insights for client assessment; tasks that were previously done manually, prone to human error, and time-consuming are now streamlined, delivering appreciated results, and notably, despite initial concerns about language barriers, it has been successfully implemented in French, Arabic, Japanese, Spanish, and Portuguese without any issues.

Ignasi: That’s wonderful. Looking ahead, we will see the credit sector completely evolve and transform through the use of AI?

Tiago: I believe we are only seeing the tip of the iceberg, as current use cases like the one I described involve implementing AI within existing processes, but moving forward, these processes will continue to evolve and be reengineered so that generative AI is not just a support tool but the central enabler, and based on the evidence I’ve seen, I am convinced this transformation is inevitable.

Ignasi: So, exciting times ahead, right?

Tiago: Absolutely.

Ignasi: Thank you, Tiago, for sharing these valuable insights and for participating in our quarter.

Tiago: No, thank you for inviting me. As you can see, I am a passionate believer in this topic, and I truly think it will change our future.

Ignasi: Thank you again.