Can we trust a company's decisions to artificial intelligence? Interview with Andor Havasi and Sándor Gáspár

Gergely Szertics now focuses on the insurance sector, and also dives into the topic in the context of AI-based customer service solutions for large companies. with Andor Havasi, Analytics BU Director at Stratis, and Sándor Gáspár, Head of Data Solutions Competence Center at Stratis.

In recent years, artificial intelligence has revolutionised and accelerated workflows in many areas of the life of insurance companies and customer services. And AI-based solutions are evolving at a pace that promises to help us in more and more areas in the future. In any case, it's worth going through a few important questions before adopting the new technology.

Where is the point at which traditional statistical analysis is no longer enough and AI needs to be called in to help make business forecasts for large companies?

Statistical analyses can be used to build predictive systems from large volumes of data, but the complexity of the interrelationship of many variables and parameters beyond human comprehension is worth modelling with artificial intelligence. In this way, we can make much more accurate predictions.

What are the first steps towards the implementation of data-driven operations and AI-based solutions?

First and foremost, consider the infrastructure available to collect data and the business value that can be extracted from it. Customers are often unaware of the valuable da3Html sectionaddexpandmore-dots Embedelt tartalomENta assets they are sitting on, simply because they are unable to exploit the potential it holds.

It's also worth mapping out exactly what business problems we want to solve for our company. We should not want to implement AI in general, but rather select the target processes we want to improve and optimise in our business operations.

What is the biggest difference between Chat GPT vs. AI-based language models used in large ent4Html sectionaddexpandmore-dots Embedelt tartalomENerprises?

AI is in a sense like a first job starting assistant. It needs to be trained and familiarised with the company's data to become an effective employee. Chat GPT is a general language model. It has been taught to answer certain questions, but it is not familiar with the information of a large company: It does not know what its policies and procedures are, nor what decisions are made. You need to train it in these first, and then it will be able to make automatic decisions.

What makes AI an effective employee?

It does not need to be told what grounds to decide on and it does not need to be taught rules. It simply needs to be shown the information on which we base decisions and the outcome of those decisions. It will then be able to see the connections and make sound and informed decisions based on that information.

In our discussion, we will explore other fascinating issues:

• What do data assets mean?

• How can we make forecasts from the data we have?

• What are the long-standing application areas of language-based artificial intelligence?

• What work processes can be automated?

• For how long have we used AI-based technologies?

• In which direction is the market evolving?

Click play to find out everything!