Tech Trends 2025: Software-Based Efficiency in Actuarial Work
AI is Becoming Invisible Infrastructure
AI is moving into the background – essential, ubiquitous, and largely unnoticed, much like electricity or the internet. In actuarial practice, this means AI will increasingly automate routine processes: from cleansing and structuring raw datasets to generating scenario-based insights on reserves or capital requirements.
70% of organizations are already exploring or deploying large language models (LLMs), and two-thirds report measurable business value from generative AI use so far. This invisible layer will power the tools actuaries already use – without requiring prompts or manual interaction.
From Large to Lean: The Rise of Small Language Models
While LLMs like GPT-4 have broad capabilities, they can be inefficient for narrow (actuarial) use cases. Instead, many companies are adopting Small Language Models (SLMs) – AI models trained on focused, internal datasets.
According to Deloitte, 75% of the enterprises surveyed prefer smaller open-source AI models due to cost-efficiency, enhanced control, and data security. Actuaries could soon rely on SLMs trained on historical data or company-specific policy wording – delivering faster, context-aware outputs without compromising confidentiality.
Agentic AI: From Insight to Execution
A major evolution in 2025 is agentic AI: systems that can act autonomously. Imagine an AI assistant that identifies shifts in longevity trends, updates actuarial tables, and triggers impact analysis – automatically.
By 2027, agentic systems are projected to manage up to 30% of structured decision-making tasks in major enterprises. For actuaries, this means less time compiling data – and more time interpreting results and advising leadership.
AI in Hardware: Local, Private, Powerful
After years of cloud-first thinking, AI is now driving a return to hardware. With AI-enabled chips built into laptops and on-site servers, users can run powerful simulations offline – faster and more securely.
The AI chip market is expected to grow from USD 50 billion in 2024 to USD 400 billion by 2027. This enables enhanced modelling in sensitive environments with strict data controls.
AI-Fueled Core Modernization
Enterprise software is being rebuilt with AI-first architecture – creating intelligent cores that learn and adapt over time. For actuarial systems, this includes self-updating reserving tools, real-time stress testing engines, and dynamic pricing models.
AI-powered modernization can reduce processing times by up to 60%, while boosting auditability and control. However, complexity increases at the system level – making actuaries essential as both users and guardians of these intelligent frameworks.
Data Governance and Security: A Critical Foundation
AI systems require vast amounts of data – and poor data can scale mistakes. Moreover, quantum computing is on the horizon, posing risks to the encryption systems used in finance.
55% of organizations avoided certain AI use cases in 2024 due to unresolved data security issues. Actuaries handling sensitive inputs like health data, underwriting models, or regulatory returns must actively shape governance strategies, from encryption to audit trails.
You can download Deloitte’s Tech Trends 2025 report here.