Artificial intelligence is simply the new Renaissance: Pasquarelli

The best way to understand artificial intelligence (AI) is that the idea of a polymath as a human genius, exemplified by Leonardo da Vinci in the late 1400s and early 1500s, has disappeared, according to Walter Pasquarelli, a member of the OECD One AI Group of Experts and of the AI Research Affiliate at Cambridge University and a contributor to The Economist Group.

Speaking to insurers at the Geneva Association’s 50th anniversary conference in Zurich, Pasquarelli said, thanks to the internet and the vast store of knowledge it offers, everyone has become a polymath. Beyond that stage of scientific evolution, he added, we are at a “new frontier” of what AI presents.

AI, put simply, is a “non-human problem-solver”, he said. John McCarthy of Massachusetts Institute of Technology may have coined the phrase “artificial intelligence” in 1956, but Pasquarelli pointed out it was Alan Turing, the British mathematician and computer scientist, who in 1950 first posed the question: “Can machines think?”

“The hype around AI has come in waves over the years, until those waves became shorter and the heights longer,” Pasquarelli said.

One wave formed between 1997, when the Deep Blue IBM chess computer beat Garry Kasparov, and 2015, when AlphaGo became the first computer Go program to beat a human professional Go player.

“AI is actually an umbrella term, under which lies machine learning, meaning a machine might not play chess today but after a million simulations it becomes undefeatable at it. Then there is generative AI, which is AI that is not only able to learn but also to create. Then there are foundation models that are technical elements of that,” Pasquarelli said.

“The point is that AI used to be narrow in its application. What has changed is the generative pre-trained transformer, a type of foundation model that does multiple tasks.”

“AI is a journey. You cannot transform your company within a day, a week or a month. The important thing is to start, and slowly. AI is going to transform the whole economy. It’s happening. It’s not a theory anymore. It’s a reality that is increasingly going to determine business success”
_{Walter Pasquarelli}

The most famous of these is ChatGPT, which between December 2022 and February 2023 became the fastest-growing app of all time, accruing 100 million users.“

The beauty of this free-to-use AI system, Pasquarelli said, is that “it is actually quite useful”. ChatGPT is a “classic example”, he continued, of a type of generative AI called a text-to-text tool, which “democratises” data analysis capabilities. This function can work for the insurance industry, he said, in claims processing, information retrieval, predictive modelling and generating bespoke policies.

Another type is text-to-image, which has interesting applications for marketing, while text-to-speech, like avatars, can make a customer’s experience “feel more real”, he said, adding, “Give it two or three years and I guarantee there’s going to be a Hollywood-grade movie made by someone just from a consumer laptop.”

The future of AI is to serve as “agents” in the way “Siri and Alexa were always meant to be”, by responding to commands such as, “Buy me two concert tickets to see Taylor Swift in two weeks’ time”, Pasquarelli said.

The next development will be for these “servants” to be proactive, such as by responding to the cancellation of a train by rebooking the journey for another convenient time. Citing a recent article from The Economist, “Is the end of the iPhone near?”, Pasquarelli said there will come a point when people are connecting with AI “all the time” because it will “always be assisting us”.

Getting AI-ready

Making a company “AI-ready” requires preparation.

Pasquarelli said: “Identify specific use cases. Don’t just throw a neural network at your whole business model. Have the right data, have the right talent and have the right infrastructure, such as cyber security.

“Be aware that scams may have been AI-generated, and don’t trust detection tools for deep fake because they are highly unreliable. Even OpenAI pulled down its own tool for detecting whether a text was AI-generated because its accuracy was so incredibly low, at 26%.”

Instead, he said to use embedded cryptographic hashes or data provenance, which place a watermark or stamp in the pixel of an image.

To ensure privacy, he warned against putting sensitive client information on ChatGPT. To avoid bias, he stressed the need to audit data with customers involved in that process, “by speaking with them regularly to understand their context and lived experiences”.

Data is where AI begins and ends, he said. “Data is where you really need to put your effort in, to develop systems that are not only good in a way that they don’t contain bias, but systems that are free of bias are also much more trustworthy for your clients.”

He added: “Importantly, put a human in the loop. Always make sure when a decision is made, a person is in there who can basically have the final call or at least review it. This is to avoid any kind of lawsuit, but it generally increases the quality.”

A question AI experts are often asked is, “Am I going to lose my job?” Pasquarelli said: “The answer is AI seems to be producing more jobs rather than actually eliminating them. But all jobs are actually going to disappear.”

The “productivity gain” from using AI will mean “increased competition for workers”, he added, “but overall, the benefits seem to be net positive”. The main positive is that using generative AI “increases job satisfaction, because you can give away some of the tasks that you don’t enjoy doing”.

In adopting AI, the first step a company needs to take is to “demystify” it, by accepting that it is simply “statistics with computing power”, Pasquarelli said. The second step is understanding that the AI strategy is the business strategy, meaning a company ought to identify the use cases where AI can help. The next step is to create a pilot project using AI for a small use case to test the technology “within a safe environment, to build momentum without losing too many resources if something goes wrong”. The final stage is to “weave in” ethics and trust by being able to display that the company has responsible AI practices, which is a source of competitive advantage, he said.

“AI is a journey. You cannot transform your company within a day, a week or a month. The important thing is to start, and slowly.”

He concluded: “AI is going to transform the whole economy. It’s happening. It’s not a theory anymore. It’s a reality that is increasingly going to determine business success.”