This blog post was first published in the June 2023 edition of The Actuary magazine. It’s also part of an ongoing series on how I use AI interns to improve my life. In this case, I’ve used ChatGPT for writing inspiration. The complete prompts and responses are at the bottom of the page.
Artificial Intelligence (AI) and machine learning (ML) techniques have been much hyped over the years. Beyond creating alternatives to generalised linear model (GLM), they have not come close to revolutionising actuarial work. Adoption of AI and ML techniques by the profession has been hampered by a lack of quality data and poor data infrastructure. Worse still, premature guardrails were imposed with good but misguided intentions – overzealous guidelines on AI safety and ethics were not commensurate with the primitive capabilities of the AI models in use.
In recent months, there have been impressive breakthroughs in AI models. You would have heard about ChatGPT and its ilk by now. These generative AI systems are unlike any that has come before – not only do they understand natural language (ChatGPT tells me it understands over 100 languages), they generate text, software code, and images with human-like flair and creativity.
They are now becoming so capable that the fear is that we are ill-prepared for non-human superintelligence. Eminent tech leaders, including Elon Musk and Steve Wozniak, have since signed an open letter calling for a temporary pause in training more advanced AI systems.
Like it or not, the genie is out of the bottle. Microsoft is incorporating the technology in its suite of productivity tools to automate tiresome tasks like responding to emails, writing meeting minutes, and so on. These are low-hanging fruits; before long, AI systems trained on company-specific data will be capable of analysing data, building models, and managing assets and liabilities – the bread and butter of actuaries.
Job displacements will be slow as it is unlikely that advanced AI systems will be widely adopted in the short term. Most large organisations assess new technology with a risk-tinted lens; rightly so because the stakes are high. However, make no mistake, the technology is compelling and its capabilities will improve exponentially. Deploying AI technology will also cost much less than employing a full-time person. The logical conclusion is that the world will need less white-collar knowledge workers. One-man actuarial teams, enabled by AI agents, will be the norm.
Organisations that are not AI-first will cease to be competitive. Before “AI Transformation” initiatives begin in earnest, there is a window of time for actuaries to pivot and upskill. The goal is to be hard to compete with so that we cannot be replaced. I asked ChatGPT to suggest strategies to that end – there is no silver bullet but I found the following to be sound advice:
- Embrace and leverage technology – Just as AI-generated art prompted by professional artists is noticeably better, collaboration between AI and actuaries will create immense value. You need domain expertise (for now at least) to prompt generative AI into giving the most useful and accurate response. ChatGPT may be banned at workplaces due to data privacy concerns but that is not an excuse to not explore, outside of work, how it could be incorporated in your workflow.
- Be good at intersections of multiple fields – Having a firm grasp of financial economics and risk management techniques is no longer sufficient. Frank Redington famously said, “The Actuary who is only an actuary is not an actuary”. While specialisation is important, roles in wider fields will provide valuable learning opportunities. Being able to connect the dots across multiple disciplines – e.g. finance, technology, and marketing – is infinitely more beneficial to your career than being a master in quarterly Solvency II reporting and analysis (which can soon be fully automated).
- Master sales – Snake oil salesmen have made “sales” a dirty word. Being able to sell your vision and get buy-in from your target audience is an important skill. It may not come easy to actuaries who are introverts by nature or are modest to a fault but the skill is learnable with practice.
AI agents work for free and do not ask for holidays. In the face of such competition, it is natural to have concerns about job security. The robots may be rising, but there are ways to thrive. If there is one takeaway, it is that continuing professional development is no longer an annual tick box exercise.
The bottom line: AI is becoming more capable and may soon displace white-collar knowledge workers, actuaries included. To stay a step ahead, actuaries should leverage technology, be good at intersections of multiple fields, and master sales.