
Will AI Replace Actuaries?
AI tools from Akur8, Milliman and others now handle parts of pricing, data prep and first drafts in actuarial work. The evidence points to a profession whose daily tasks are changing while its accountability stays with humans.
In 2024, 80% of actuaries said they feared becoming obsolete, according to survey results summarized by InsuranceIndustry.ai. A year later, Carrier Management reported that 51 percent of actuaries said they were not worried at all, or had not considered the prospect of being replaced by AI.
That swing matters. Actuaries have now worked with these tools, and the panic has cooled. AI is taking over chunks of pricing, data preparation and first drafts. It is not taking over the professional sign-off on a reserve or a rate. Below is what the evidence shows and what to do about it.
What this means for you, depending on where you sit
- Students working toward exams: The math still counts, but pair it with model validation and data skills. The BLS lists 1,500 openings a year on average and a bachelor's degree as the typical entry education, per its employment projections.
- Mid-career pricing and reserving actuaries: Expect AI to handle more of your data prep and documentation. An actuary.info summary of an April 2025 SOA report says that if you use generative AI tools in your actuarial work, you carry the professional responsibility for the output. Your value shifts toward checking and explaining what the tools produce.
- Employers and insurance clients: The Society of Actuaries recommends running AI alongside existing methods before relying on it. State regulators are also watching how insurers govern AI, so build oversight in from the start.
Where AI already does actuarial work today
The clearest deployments come from pricing software vendors and consulting firms. Here is what the evidence documents.
- Akur8 Agents: Akur8 rolled out a product that embeds expert guidance and no-code automation into the daily workflows of actuaries, fintech.global reported. The company says the underlying technology is relied upon by over 3,000 actuaries globally, that customer data is never used to train AI models, and that the product complies with the EU AI Act. A separate fintech.global article says the agents launched in the second quarter of 2026.
- Akur8 platform build-out: In its 2025 annual report, Akur8 says it introduced Rate Repo and Deploy, which connect model development, rate management and production deployment. Akur8 also acquired Matrisk, adding AI-driven filings search and market intelligence.
- Milliman: A Milliman report describes using a generative AI model to produce a first draft of insurance reporting work. The firm has since launched Milliman AI Solutions, a designated practice.
- Aviva: Pricing software vendor hyperexponential says Aviva built 20 pricing models in 9 months using a unified platform. This is a vendor claim, not an independent audit.
- P&C insurers broadly: The same hyperexponential post cites a Novarica survey of 51 North American P&C insurer CIOs in which 59% had implemented machine learning in actuarial processes.
- Reserving and reporting: EY says AI agents support data preparation, documentation assembly, standardized reporting, and elements of reserving and valuation. EY also says generative AI is now in production at many insurers.
Use is wide but shallow. About 80% of actuaries now use AI, yet most report using it for fewer than five hours a week, according to insurtechexpress.
One gap stands out. The evidence we reviewed contains no documented layoffs or hiring cuts of actuaries blamed on AI. The shift so far shows up in tools and workflows, not in pink slips.
Field note: The Society of Actuaries puts the safest starting point plainly: "Introducing AI as a parallel analysis offers a practical and low risk pathway for adoption." The American Academy of Actuaries makes a similar point for reserving, treating AI output as one input among many that actuaries weigh. If your team is new to this, run the AI estimate next to your chain ladder and study where they disagree.
What the numbers say about actuarial jobs
Every federal projection we found shows actuarial employment growing faster than the average for all jobs. The figures differ because they come from different projection periods.
| Measure | Figure | Source |
|---|---|---|
| Jobs in 2025 | 31,200 | BLS projections |
| Jobs projected for 2035 | 34,100 | BLS projections |
| Projected change, 2025-35 | +9.2% (all occupations: 3.5%) | BLS projections |
| Median annual wage, 2025 | $130,000 | BLS projections |
| Jobs in 2024 | 33,600 | BLS Occupational Outlook Handbook |
| Projected growth, 2024-34 | 22% (7,300 jobs) | BLS Occupational Outlook Handbook |
We report both sets as published and do not try to reconcile them. In either case, the BLS expects more actuaries, not fewer.
The BLS Monthly Labor Review ranks actuaries among the occupations with the highest scores for critical and analytical thinking, with employment projected to grow 21.8 percent from 2024 to 2034.
Exposure studies tell the other half of the story. A table in an arXiv review of AI and jobs lists actuaries with a score of 1.52 among highly exposed occupations, and IQ Career Lab notes that actuaries topped one list of AI-exposed jobs. Microsoft Research found the highest AI applicability scores in groups such as computer and mathematical work, where the BLS files actuaries. High exposure plus strong projected growth is the pattern we would expect for a job where AI changes the tasks while demand for the judgment holds up.
The messy middle: black boxes, benchmarks and blame
The limits are real, and they land on actuaries specifically.
Explainability. A Casualty Actuarial Society presentation warns that many AI models, especially deep learning systems, behave as black boxes. That is a problem for anyone who has to defend a rate to a regulator.
Hidden errors. In a fintech.global interview, an executive argued that errors in actuarial models can stay hidden for years, and that AI still struggles to replicate contextual judgment built through experience.
Uneven accuracy. The INS-ActBench benchmark found that "leading LLMs can already handle formula-based actuarial computation" when the problem supplies standard models. Models with weaker numerical reasoning scored much lower, and on case questions, models often failed to recover the complete set of correct options.
Responsibility. The American Academy of Actuaries states: "AI is a model, and as such, ASOP No. 56 applies." According to actuary.info, the Actuarial Standards Board has not signaled plans for a dedicated AI standard and relies on ASOP Nos. 56, 23 and 41 plus guidance. The CAS AI Primer says safe use of agentic AI depends on governance frameworks.
Regulation. The NAIC issued its model bulletin on insurer AI use in December 2023, and a tracker cited by actuary.info reported nearly half of states had adopted it by March 2025. The NAIC says its AI evaluation tool was being piloted by 12 participating states as of March 2026. In the UK, the IFoA issued an AI risk alert on 6 Sep 2023 and updated its data science and AI guidance in Jul 2025, and it has run joint updates with the FRC on the topic.
Open question: who answers when an AI agent quietly builds a flawed rating variable into a filing? Current standards point back to the actuary.
Our verdict: AI will reshape actuarial work, not erase it
We think the actuaries at risk are those whose days consist mostly of data cleaning, routine reporting and rerunning standard models. Vendors and consultancies are automating exactly that work.
The profession as a whole looks secure to us. Paul Sweeting said in his 2025 IFoA Presidential Address that "AI is not going to replace actuaries", because actuaries make sure AI works as it should. The NAIC says AI is "more likely to support human workers than replace them" in insurance. Add BLS growth projections and professional standards that keep responsibility with the actuary, and our read is that the job survives with a heavier tilt toward validation, governance and explanation.
Questions people ask
Will AI replace actuaries?
Our read is no, not as a profession. The BLS projects actuarial jobs growing from 31,200 in 2025 to 34,100 in 2035, and professional standards keep responsibility for AI output with the actuary.
How much do actuaries use AI today?
About 80% of actuaries now use AI, according to insurtechexpress, yet most report using it for fewer than five hours a week. EY says AI supports tasks like data preparation, documentation assembly and standardized reporting.
Are actuaries still worried about AI?
Less than before. In 2024, 80% of actuaries expressed fear of becoming obsolete, while a later survey found 51 percent of actuaries were not worried at all or had not considered being replaced by AI.
Can AI do actuarial math correctly?
Partly. The INS-ActBench benchmark found leading LLMs can handle formula-based actuarial computation when given standard models, but models with weaker numerical reasoning scored much lower and often missed parts of case answers.
Sources
Every number, quote and example on this page was checked by script against the sources below before publishing: quotes must match the source word for word, and figures must appear in a source or in BLS data. Our verdict is editorial judgment. Spotted an error? Tell us.
- U.S. Bureau of Labor Statistics: Employment Projections, Actuaries (15-2011)
- Will AI take over actuary jobs? 2025 future of actuaries - hyperexponential blog (hyperexponential.com)
- The CAS AI Primer: Practical Guidance for Actuaries By the CAS Artificial (casact.org)
- Underwriter, Actuary Fears of AI Drop; Work Needed on Collaboration - Carrier Management (carriermanagement.com)
- AI Insights Dec 26 2025 - InsuranceIndustry.ai (insuranceindustry.ai)
- AI in actuarial functions: how insurers transform operations (ey.com)
- Insurers Prioritize AI While Actuarial Standards Lag Behind (actuary.info)
- Actuaries (bls.gov)
- Skills savvy: a skills-based look at occupational data : Monthly Labor Review : U.S. Bureau of Labor Statistics (bls.gov)
- Professionalism in a Dynamic Environment The Challenge of AI Presentation to: (casact.org)
- Actuarial & Algorithmic Accountability: Setting Ethical Standards for AI - Actuary.org (actuary.org)
- Actuarial Standards Board Enters Busiest ASOP Cycle in a Generation (actuary.info)
- Accenture Survey Expects AI Underwriting Adoption to Reach 70% (actuary.info)
- Akur8 2025 Annual Report (akur8.com)
- Akur8 launches AI agents to streamline actuarial pricing (fintech.global)
- MILLIMAN REPORT An application of generative AI in the insurance industry (media.milliman.com)
- Can actuaries trust AI models? (fintech.global)
- How Akur8 is building an end-to-end actuarial platform for the next era of insurance (fintech.global)
- Milliman Launches AI Solutions Practice (joplinglobe.com)
- Working with AI: Measuring the Applicability of Generative AI to Occupations - Microsoft Research (microsoft.com)
- Actuaries as data stewards in the AI era | Institute and Faculty of Actuaries Blog (blog.actuaries.org.uk)
- Joint IFoA and FRC regulatory update: guidance relevant to AI, data science, and machine learning | Institute and Faculty of Actuaries (actuaries.org.uk)
- IFoA Presidential Address 2025 by Paul Sweeting | British Actuarial Journal | Cambridge Core (cambridge.org)
- INS-ActBench: A Comprehensive Benchmark for Assessing Professional Actuarial Capability of Large Language Models (arxiv.org)
- AI and jobs. A review of theory, estimates, and evidence (arxiv.org)
- AI-Exposed Jobs Rank Highest in Reasoning | IQ Career Lab (iqcareerlab.com)
- Actuarial AI 2026: Right Data, Right Time. How Insurers Are Turning Weeks of Data Prep Into Minutes - insurtechexpress (insurtechexpress.com)
- May 2026 (soa.org)
- Insurance Topics (content.naic.org)
- AI Use Cases in Insurance and Pension (actuary.org)