You can relive AI specialist Luc Julia’s fascinating talk on the subject:
What Luc Julia reminds us is that artificial intelligence is primarily used to make tools for very specific tasks.
In the insurance sector, artificial intelligence (AI) has found its main uses in the following areas:
Actuarial skills: AI can allow insurers to facilitate actuarial analysis on very large but necessarily coherent sources.
Fraud detection: Insurers can use AI to detect potential fraud patterns. Machine learning algorithms can analyze the behaviors of benefit claimants and health care providers to identify anomalies and fraud indicators. However, this detection still needs to be corroborated to date by human analysts.
Improved customer service: Chatbots and virtual agents can be used but very carefully. Indeed, their responses may engage the insurer, broker or manager who provides the service. It is therefore necessary to strictly limit the scope of AI’s possibilities in its responses.
Disaster prevention: In the future, by analyzing data from IoT (Internet of Things) sensors installed in cars, homes or industrial equipment, insurers will be able to identify and prevent the risks of disasters. For example, by monitoring sensor data in homes, insurers can detect potential water leaks or fires before they cause significant damage.
For Smalltox, our job is to organize complex data flows in a safe way. The whole difficulty is to design algorithms that are :
– Safe to set up: this involves the use of easily reproducible models whose behavior can be tested automatically before use ;
– Complete: which anticipate all situations with overflows towards human treatment when necessary ;
– Protected by safety nets: ensure that any problem to be addressed is brought to the surface as long as it is not resolved.
AIs are not yet able to do this. Therefore, we continue to use our brains. 🙂
