 |
Esko Kivisaari
Chairperson AAE Artificial Intelligence – Data Science Working Group |
|
Large Language Models (LLMs) like ChatGPT might finally bring with them the long-awaited benefits of AI. It is not clear what they will mean in practice and how fast their adoption will be.
While LLMs have stolen a major share of publicity AI is also proceeding in other areas. It has been used, e.g., in opening the folding of proteins (previously thought to be an impossible feat) and finding prototypes for new antibiotics (critically helpful with more and more resistant bacteria). Actuaries need to understand the possibilities and threats introduced by AI. We might see in the coming years an ActuaryGPT (Generative Pretrained Transformer) giving us better tools in our day-to-day work. We should also think of unsolved actuarial problems that could be manageable with AI, mirroring the successes with proteins and medicine.
Actuaries need to understand how they can, based on their long experience with complex models, bring their expertise to this novel area, and also warn of the possible dangers with models that might be used when assumptions they are based on are not valid or when the tools are used in areas where their validity has not been tested. Actuaries can also help regulators in safeguarding that new legislation like the AI Act will be successfully implemented. One element of this will be to develop tools in explainable artificial intelligence (XAI).
This is a unique opportunity for actuaries to strengthen their role, broaden their areas of influence, and safeguard responsible use of AI in critical areas. The AAE has recently published its paper on ‘What should an actuary know of AI’. This paper hopefully lays the groundwork for fruitful actuarial engagement also in the age of AI.
Different religions and philosophies have discussed values of different aspects of life. The Hindi concept of Dharma comes close to what in my more familiar tradition is read from 1 Corinthians 13 especially in weddings: ’Charity never faileth: but whether there be prophecies, they shall fail; whether there be tongues, they shall cease; whether there be knowledge, it shall vanish away.’
Data explosion can lure us into the thinking that data and the tools to manipulate it are neutral. They are not. The data answers to questions we pose to it. The answers we are getting depend on what weight we give to different value propositions.
Actuaries are there to serve the common good, to serve our societies. The paper just published tries to help our profession with new tools in the domain of data science. Actuaries need to use them responsibly. One way of doing this is to equate our ambition to serve the common good with charity.
P.S. We had ChatGPT read through the paper mentioned above. Its (his?) answer, which could serve as a recommendation, was: ‘After thoroughly reviewing the document “What should an actuary know about Artificial Intelligence,” I found no typographical or grammatical errors. The document appears to be professionally written and well-edited, adhering to standard writing conventions. Consequently, there are no specific corrections or suggestions regarding typos or grammatical issues to report.’
23 January 2024
This blog is written on a personal title.
Read all the AAE Blogs here!
