Older than you think
Put simply, artificial intelligence is "the performance of tasks by a (computer) system that would normally require human intelligence." Classic examples are problem-solving, learning from experience and complex decision-making. It may be surprising, but the concept is anything but new. For centuries we have imagined artificial beings and systems that can carry out human actions on their own.
The ancient Greeks told of mythical bronze guardians built by people to protect them from enemies, and in the 11th century European courtiers were amazed by so-called 'automatons': mechanical animals or structures that could move on their own. Medieval robots, if you will.
A first scientific application arrives in the 17th century with the invention of the first calculating instrument by Blaise Pascal, who originally named his device 'the Pascaline'. This instrument could autonomously perform various calculations, and so it lies at the basis of the first modern computers some three hundred years later.
We owe the model for those first computers to Alan Turing, who at the same time was thinking about 'machines that can not only carry out tasks on their own, but also display a form of autonomous intelligence'. His theories eventually led an American scientist in the 1950s to capture the whole discipline under the term 'Artificial Intelligence', or AI for short. The rest, as they say, is history.
Applications in the insurance sector
Since the rise of digitalisation in the insurance industry, which has been under way for a good number of years now, AI has already been used for various purposes. Most of those applications, however, sit at the back end of insurance processes.
Take document processing, where documents are automatically read by a system that then fills in the data in structured form. Almost every insurer in Belgium uses this kind of technology to process claims. Those same claims files are, in many cases, also checked for fraud by AI. A more advanced example is one where certain events or actions are predicted by a computer, such as buying behaviour or risk patterns. In AI circles that is called 'predictive modelling'.
Regardless of how AI is deployed by insurers today, the goal has in most cases always been the same: efficiency gains. Until recently, AI technology was seen above all as a way to make processes faster and more effective. That makes sense, because the impact shows up immediately on the bottom line.
"Have you met our chatbot yet?"
Thanks to the success of various AI applications at the back end of insurance processes, companies naturally started experimenting with front-end applications too: 'customer-facing AI', as it is called.
The best-known form is the chatbot. We have all spoken at some point with that friendly virtual assistant that tirelessly staffs the customer service of countless companies and is meant to be the first point of contact for questions or problems. Often the experience was rather disappointing, and chatbots were quickly written off in the insurance sector (and far beyond) as superficial and unreliable. Users and developers alike lost faith, and the chatbots were then often banished to some forgotten corner of the website.
"Users and developers alike lost faith, and the chatbots were often banished to some forgotten corner of the website."
Recently, however, these virtual helpers were brought back to life by the American company OpenAI. In the autumn of 2022 they launched the chatbot of chatbots: ChatGPT. The company, led by CEO Sam Altman, a familiar face in Silicon Valley, astonished friend and foe with the sheer power of their GPT, in full 'Generative Pre-trained Transformer'. They set in motion an irreversible movement that many believe will change the world for good.
More important than fire and electricity
Google boss Sundar Pichai calls it "a technological development that will have more impact than the invention of fire, electricity or the internet". Other technology heavyweights, such as Tesla's Elon Musk, instead call for a development pause for as long as we cannot better understand, and if necessary limit, the real impact and potential dangers of AI.
"Artificial intelligence is a technological development that will have more impact than the invention of fire, electricity or the internet." - Sundar Pichai, CEO Google
But what makes the most recent months so different from, say, the past 50 years? At the heart of it lie the so-called LLMs, in full 'Large Language Models'. A Large Language Model is a computer model that understands and analyses language in order to formulate a human-like answer. Thanks to successful development and a global roll-out, OpenAI managed to accelerate the progress of artificial intelligence at an unprecedented pace. Not so much because their results are ground-breaking, but because the way they were made accessible brought about a real democratisation of what was previously complex, often inaccessible technology.
It is that public accessibility that is now driving a true 'AI gold rush', and that at the same time feeds the fears of a great many people. Thanks to its recent progress, artificial intelligence has already proven several times that it can not only carry out tasks faster than a human, but that it often also achieves better results than its human counterparts ever could.
Strikingly, that competition has not come where it was initially expected. In the past it was predicted that simple and repetitive tasks would be replaced first, and that assignments requiring deeper knowledge or a form of creativity would remain out of reach. That theory has now been swept off the table. In the past few months, AI applications managed to predict the stock market with astonishing precision, give correct legal and medical advice, and even produce impressive art in a matter of seconds.
What now?
We can therefore rightly ask what we can expect from this promising but possibly also ominous technology. A techno-optimist would say this technology will make life a lot easier for everyone: far-reaching automation, unprecedented fast access to knowledge, a robot helper at your side day and night, and so on. Sounds good!
The opposite reasoning can be made just as easily. If patients of GPs in the United States already say they feel better helped by an 'AI doctor', and if it then also turns out that the diagnoses given are more often correct, how do you, as a GP, look at this technology?
"An American study shows that GP patients would rather be, and are better, helped by AI doctors."
Within our own beloved insurance sector, too, we are left with plenty of questions. For most players AI is still a black box. And what is unknown is often also frightening. In the search for answers, a good first step can be to dig into both the possibilities and the risks of the technology. Becoming familiar with, and staying up to date on, AI can serve as a solution. To support brokers in this, WeGroup even set up an entirely new service, in which brokers and their staff are immersed in the world of AI during a true AI workshop at the office, discovering how the technology can already support their daily work today.
Either way, on one thing both the true believers and the doom-mongers already agree: artificial intelligence is here to stay, and it is up to humankind to learn how to deal with its own, smartest creation.
This article was written for more than 60% by an artificially intelligent system. In doing so, the system drew on publicly available knowledge, and it was steered by the human author to think critically about the claims made, the information presented and the overall context of the article you have just read.
