By Tomás Blas, Head of Property Spain at BHSI
The Traditional Artificial Intelligence (AI) It has been integrated into our lives for a long time. For example, Amazon uses AI algorithms to learn what a person likes and what other people with similar profiles have purchased to provide recommendations for future purchases. Similarly, Netflix’s recommendation engine is powered by AI and uses each subscriber’s viewing history to offer suggestions for future shows.
The insurance sector has also been using traditional AI for several years, from reading underwriting requests and extracting the necessary critical data, to interpret incoming emails about claims and automate the start of processing of these.
Therefore, we can understand traditional AI as the pattern recognition from discrete data sets.
The Generative AIon the contrary, goes much further and is a technology that can create new content, such as images, music and text, without being explicitly programmed to do so. It works by learning patterns and rules from existing examples and then uses that knowledge to generate new content.
The data set available for generative AI is potentially unlimited and can create a huge number of valuable business use cases, in sectors as diverse as medicine, education, transportation, energy or agriculture and, of course, in the insurance sector itself.
With regards to the risk underwriting, an example of using generative AI can be training models with historical loss data so that underwriters can better understand risk factors, identify trend patterns and estimate the probabilities of a loss occurring, which can lead to more accurate underwriting decisions and optimized pricing. Refering to claims managementgenerative AI techniques can play a critical role in fraud detection by identifying anomalies and outliers in claims data.
Although generative AI has enormous potential, there are risks associated with its use, such as the following:
- Disinformation and false content: Generative AI can be used to create realistic but completely made-up content, such as images, videos or text. This raises concerns about the spread of misinformation and false content, as it becomes increasingly difficult to discern between authentic and generated content.
- Data security and privacy: Publicly available generative AI solutions generally do not guarantee data security. Any data shared with these tools would become part of that tool’s ecosystem. This also raises privacy concerns, as personal or sensitive information may be included in the training data set.
- Bias amplification: Again, since generative AI models learn from the data they are trained on, they can also capture biases present in this data. If these are biased in terms of race, gender, or other characteristics, the generated content may show the same biases.
- Lack of transparency and explainability: AI systems can be very complex and opaque, making it difficult to understand how they make decisions or reach certain conclusions. This lack of transparency can make it difficult to refute biased or erroneous AI decisions and hold AI developers accountable for their systems.
- Intellectual property infringement: Generative AI can create content that resembles existing works, raising concerns about intellectual property rights.
- Tampering and falsifications: Generative AI can be used to create “deepfakes,” which are highly realistic manipulated content that falsely depicts people saying or doing things they never actually did. Deepfakes can be used for malicious purposes, such as spreading disinformation, blackmailing, or damaging someone’s reputation.
- Ethical considerations: The use of generative AI raises ethical questions about consent and ownership. For example, when generating human-like content, questions arise about consent and possible unauthorized use of someone’s image. The development of AI also raises other complex ethical considerations, such as what is the nature of consciousness, what is the meaning of intelligence or what are the limits of human intervention in AI systems.
- Displacement of jobs: As AI becomes more sophisticated, it could automate tasks that people currently perform, leading to the loss of certain jobs, although other types of AI-related jobs should also be created.
Addressing these risks requires a combination of technical advances, responsible use and ethical guidelines. It is crucial to develop safeguards, verification methods and regulations to mitigate the negative effects of generative AI and ensure its responsible and beneficial application in society. Likewise, it is important to consider the social and economic impact of AI and develop strategies to help people who could be negatively affected by this technology.
In addition to the potential dangers mentioned above, some experts have also warned of the risk of AI becoming so intelligent that it surpasses human intelligence and capabilities, which could lead to a scenario where AI poses a threat to the humanity. However, other experts believe that this risk is unlikely and that AI will continue to be used for the benefit of humanity.
In BHSI, our approach to generative AI involves informing ourselves about the benefits, opportunities, risks and challenges it poses, ensuring adequate protections and controls are in place, and identifying and creating use cases and pilot projects that have practical business applications. These cases must be well managed so that we gain detailed knowledge and understanding of the scope and expected results. This will help us determine the future protections, policies, and controls we want for these tools in general.
We believe that our main opportunities should be focused on improve overall operational efficiency and in improve underwriting and claims results.
In short, AI in all its aspects already surrounds us and It is our responsibility to try to make the best use of this, striving to understand its potential, its limitations and its risks, which will also evolve over time. It is a very powerful tool that can help us improve in very different areas of our lives and it is here to stay.
To write this article, in addition to our own knowledge and experience, we have also relied on traditional online tools, as well as tools that use generative AI. In any case, what we reflect here is a set of human reflections and thoughts that are totally free of biases derived from their use. Although, can we be completely sure of this?
