Artificial Intelligence (AI) is a big deal. Many say it will transform the future of mankind. But there is no agreement on what that transformation will look like. Risk Managers are used to planning for uncertain futures. Here are four possible futures driven by AI that you can use as a starting point for your planning.
Work is already being transformed by AI. Some surveys tell us that as many as 40% of workers are using AI to assist them in their jobs. But AI is also reshaping jobs and hiring. So that is where we start with these scenarios: The impact on employment. How might AI impact overall employment over the next five to 10 years? That will be the main driver for these four scenarios.
Scenario 1: Everything Works Out
The most pressing concern about AI and employment is that AI will take over work for a large percentage of jobs. But in this scenario, everything works out for the best. While AI takes some jobs, it also creates some new jobs. And the knock-on effects of the money spent on AI and the capabilities that AI creates lead to a boom in new employment in small firms and especially tech-enabled start-ups. With the success of AI, there is plenty of capital available to fund new businesses, and those businesses fuel almost full employment. Implementation of AI and its elimination of jobs is spread out over 10 years, mainly because of a shortage of people with the skills to implement the AI projects. This gradual process is one of the main reasons that things do work out in a more or less balanced fashion.
With AI taking up much of the more repetitive parts of work, workers shift toward roles emphasizing creativity, empathy and connection to the physical world that humans still rule. Work drifts back toward higher flexibility with the help of AI. Many jobs start to involve robust AI-human collaboration. AI will also lower operating costs, which helps more start-up firms to survive.
In this scenario, education and training started to pivot almost immediately as AI started to make a big impact. More training for hands-on jobs and more experiential education meant that new hires were not competing their book learning against AI’s training from billions of books of words on every topic. Instead of trying to keep AI out of education, schools quickly learned how to use AI to accelerate and personalize learning. AI becomes a catalyst for increased productivity and creativity.
This all results in a vast improvement in the mood of the population with steadily growing optimism about the future, reminiscent of the 1950s. Goldman Sachs 2023 prediction of a 7% improvement to GDP driven by AI seems to be happening. AI is seen as a capable and extremely helpful partner in many endeavors. So many new businesses end up as successes that children imagine growing up to be entrepreneurs instead of firemen.
Scenario 2: Demand-Driven Depression
The Great Depression was famously described by JM Keynes as driven by a lack of aggregate demand and in this scenario, that is what causes a new depression almost exactly a hundred years later. As predicted, AI takes on the work of over 20% of the employed population. This happens somewhat more quickly than in the prior scenario and efforts at retraining and adjustments to basic education are delayed until after unemployment passes 10%. This results in a prolonged stagnation of GDP, with more than one year of negative GDP growth. Wealth inequality grows sharply. The start-ups that were the stars of the first scenario do not get funded due to lack of venture capital. The economy gets stuck in a negative feedback loop—lower employment leads to lower demand for goods leads to layoffs, and so on.
The public mood collapses. Fear, resentment and anger are endemic. Consumer demand drops even faster leading to a persistent question, “What were they thinking?” about the business executives who decided on all of the layoffs. This is clearly a self-inflicted wound. AI soaked everything in gasoline, but business leaders lit the matches to turn the economy into a bonfire.
Creative professionals in the entertainment and art worlds will face significant devastation. In this dystopian scenario, machine-made media will be fully accepted as the standard and human made work will be seen as crude and unpolished. This means that “creativity” will have been frozen to things that are derivatives of what was done before the AI takeover.
With government funds going to basic life support programs, there is little left over to pay for retraining. But a massive skills mismatch leaves millions unprepared for the new economy. There is a public movement against AI-manufactured products and eventually the firms that embraced AI. And this time, the first hit by job losses are middle-income white-collar workers, which causes a hollowing out of the middle of the income and wealth distribution.
Scenario 3: Muddle Through
Not too hot, not too cold. In this scenario, AI implementation is slowed a little more than in Scenario 1 and retraining programs are started earlier than in Scenario 2, but sporadically. A two-tier workforce develops with the higher paid tier being the people who work with AI in one way or another and the lower tier would be everyone else. The lower tier provides steady work for the most part, but with the flat trajectory of raises that has been a feature of the employment picture for the lower-paid half of the workforce continuing. The AI-related workers include the highest paid, the people who are directing the adoption of AI applications, the humans-in-the-loop who provide the assurance that the AI systems are working and the AI-augmented workers who are using AI to enhance their work.
This is by no means a smooth transition. Over the next 10 years, the economy experiences several brief recessions driven by bouts of higher unemployment as jobs are temporarily destroyed faster than new jobs are created. The slower adoption of AI means that some of the AI firms that are now doing the multi-billion dollar investing in computing equipment will fail. There are temporary gluts of the data centers as the winners buy up the losers. Write-offs by the investors in those firms mean that capital is not always there for new firms to develop and to create the new jobs that are there in Scenario 1.
Success in this scenario requires making the right choices. What those choices might be is not clear now and probably won’t be clear at the time that the choices have to be made. No doubt, this will be a difficult scenario to navigate. But this middle scenario is probably the most likely of the four scenarios presented here.
One of the areas that will be called upon to change the most will be education and training. Some universities will become known as the place to go to hire the next generation of AI experts, who will be in very high demand. Community college and vocational schools will have a larger role training people for what is left of middle-level jobs. Those jobs will be shifting from mostly managing the layer of workers performing the work to working with AI and the technicians and assurance functions. The World Economic Forum predicts that 60% of workers could need retraining in the next five years.
Artistic and creative work will reach some sort of truce with AI whereby in some situations, AI-created work is seen as the standard, but there is still significant demand for new work by humans. People will just have to get used to the disruption and instability. A steady state Volatile, Uncertain, Complex and Ambiguous (VUCA) environment will become the medium-term norm.
Scenario 4: The Revolution that Didn’t Happen
A 2025 survey by MIT Nanda reported that almost 95% of businesses that attempted to apply AI to a business project did not see any benefit from that effort. Add to that the fact that almost all AI businesses are still operating at a loss in 2025, which could mean that prices for using AI models may need to rise, perhaps significantly. What you get when you add those two aspects of the situation together is our next scenario. Here, the AI revolution just doesn’t happen as advertised at all.
Think of this as the “cold fusion” scenario. AI, like cold fusion, is a great idea, but at least so far, we have not mastered a method for delivering it that ends up being cash positive for both the users, the providers and their investors. Under this scenario, AI users cool to the idea of investing large amounts of company time and money in installing AI at a scale that would “revolutionize” the company. AI investors start to worry and pressure the AI firms to start making money. Enough money that the investors will be able to get a return on their investment at a level that makes sense for the risk that they took and that creates a sustainable business that can have a stock valuation supported by performance, rather than hype, so that they can someday get a return of their investment.
But none of that works out. At least half of the AI firms fail and most of those that continue are the AI businesses that are tied to deep-pocketed firms like Google, Meta and Tesla. Stand alone AI firms will struggle to continue. There will be a market pop like the end of the Dot Com boom in 2001.
With all that, there will not be the disruption to aggregate employment that we see in the other three scenarios. But like the end of the Dot Com era, there will be major disruptions caused by the write-downs of the AI investments. The large cap stock index will likely take a relatively small hit, but small cap indices will be devastated just like they were in 2000 to 2002. Money for investment in other new ventures will become scarce. This leads to some stagnation in the economy.
GDP growth will be reduced directly as the AI development spending ceases and additionally via the wealth effect from the reduction of consumer retirement account balances. The combined impact could be as much as one year’s GDP growth. If the economy was running close to recession before this happening, this will definitely be the tipping point that pushes it over.
Any of the other changes that we see in the other three scenarios never happen, or if they had started to happen, are retracted. Things mostly go back to a pre-AI “normal.”
Impact on the Insurance Industry
The insurance industry will experience major disruptions under all four scenarios. Individual results will vary, but here are some general trends to consider:
Scenario 1 will be very favorable for many insurers. AI adoption will create major cost savings. Insurance company services at all points of contact with consumers will improve, and the booming economy will be a boon to the investment portfolios of insurers. In addition, actuaries working with AI will be able to develop and implement several significant product innovations that spark even greater business growth.
Scenario 2 will hurt insurers substantially. Economic troubles, and especially unemployment, will make insurance hard to afford for many. Claims will surge from the remaining policies, both from anti-selection and from the impacts of the depression. Investment portfolios will be devastated as capital adequacy and liquidity become widespread concerns. Regulators will want to intervene, but their options will be limited.
Scenario 3 will be difficult for insurers to navigate. AI will continue to develop, but insurers will have a hard time holding on to the people who are qualified to implement and maintain that revolution. Some AI systems will grind to a halt for lack of people to fix the latest problem and get them running again. Customers will also be subject to the VUCA job market and their interest in insurance may wax and wane. Products that are highly flexible will be in high demand. And the trend toward temporary freelancers will have a significant impact on insurance demand as well.
Scenario 4 will have the early AI adopters struggling to justify their investment in AI as the cost to continue to operate AI systems rises. Investment portfolios will take a hit with a 5% or greater loss to surplus common. Otherwise, things will generally return to the pre-AI “normal.”
What Next?
You choose. Do you want to just let the future happen to you, or do you want to plan ahead? In these uncertain times, planning probably doesn’t mean trying to guess which scenario is going to happen. It means planning to be able to succeed in multiple scenarios. So, what can you do? Here is a list:
- Decide if you want to use these four scenarios or if you want to develop or obtain something different. There are great benefits to be had from thinking through and deciding that question.
- Work out additional details under each scenario that you need to do your planning. I have found that AI is actually pretty good at populating the details of a scenario if you get it started with your choices for the major themes for the scenarios. Simply feeding this article into AI and asking for whatever details you want can produce interesting results. Make sure to verify that what you get makes sense, though. AI can and does make mistakes.
- With your current plans, identify what will likely happen to your company under each scenario. Then start to work on scenario-specific modifications to your company plans that will improve your results.
- Have a conversation within the management team about the results of this effort. See what their reactions are and which modifications to plans that they would say are off the table, and what can be added to the list.
- Identify early warning signals so that when it starts to become evident that things are trending toward one or another of these scenarios, you can pull out the modifications to plans that were identified and start deciding what to do.
While there is not enough evidence to prove cause and effect, many observers have reported that companies that do scenario planning tend to be more resilient and better prepared for unforeseen changes. I have always thought of scenario planning as an exercise for the adaptability muscles.
Finally, I would suggest that these scenarios are not just theoretical; they are a call to action. By identifying the early warning signals for each path, and by planning proactively, businesses and individuals can position themselves to be resilient and adaptable, regardless of which future unfolds. The story of AI's transformation is not at all predetermined; it is our collective choices that will determine what will take place.
This article is provided for informational and educational purposes only. Neither the Society of Actuaries nor the respective authors’ employers make any endorsement, representation or guarantee with regard to any content, and disclaim any liability in connection with the use or misuse of any information provided herein. This article should not be construed as professional or financial advice. Statements of fact and opinions expressed herein are those of the individual authors and are not necessarily those of the Society of Actuaries or the respective authors’ employers.
Dave Ingram, FSA, CERA is a researcher, writer and part-time consultant on risk and risk management. He serves as a member of the board of the Society of Actuaries (2021–2027). Dave is also an editor of the new SOA Research Institute Actuarial Intelligence Bulletin. Dave can be contacted at daveingram@optonline.net.
Further Reading
Scenario 1
- Microsoft Work Trend Index
- US Chamber of Commerce: The Impact of Technology on U.S. Small Business
- A.I. Might Take Your Job. Here Are 22 New Ones It Could Give You.
Scenario 2
- World Economic Forum – The Future of Jobs Report
- International Labour Organization – Global Employment Trends
Scenario 3
Scenario 4