Top AI development trends to watch in 2025

In the fast-moving world of AI development, keeping up with evolving business as well as technology trends is a competitive must. Domino experts offer what they believe are top trends to pay attention to as 2025 unfolds. We’ve organized these trends into four categories:

• Industry-wide
• Life sciences
• Financial services
• Public sector

Industry-wide

• Entering the fall season: After an “AI summer” of several years of significant, $25B+ investments, we’re entering an “AI fall” as organizations struggle to scale their AI implementations and investors and boards start expecting their ROI. This adjustment will likely lead to a year-over-year pullback in funding for GenAI startups in general and a concentration of funding for the select few startups that are actually getting market traction.
• AI is becoming boring (that’s a good thing): It seems like AI emerged as a shiny new toy that automatically solved everything workers struggled to do. With more maturity, it became just another technology that solves targeted problems, requires hard work, extensive skills, and specialized capabilities to deploy. In other words, it grew up. The hottest part of AI going forward will be the boring but valuable topic of AI engineering – how to integrate, operationalize, and govern the ecosystem of technology components needed to make AI solutions work.
• Layoffs will be blamed on AI: Contrary to knee-jerk fears, AI development requires investing in the people, processes, and platforms necessary to make it work. It simply won’t happen in corporate cultures preoccupied with slashing costs. This means that in nations with relatively strong economies, few will lose job opportunities due to the proliferation of AI. It could actually drive job creation as companies seek tailored AI solutions that fulfill specific business use cases. On the other hand, developing economies that host outsourced customer service and back-office processing centers are indeed at risk for significant job losses.

Life sciences

• Accelerated drug discovery: Transitioning from proof-of-concept to practical AI applications will significantly speed drug discovery. These applications generate and test molecular compounds and biological interactions in silico, enhancing the efficiency of early clinical testing and paving the way for innovative treatments.
• Increased diagnostic accuracy: AI-enhanced medical imaging such as anomaly detection will continue to improve patient outcomes. Accuracy levels comparable to those obtained by experienced clinicians will enable quicker and more reliable decisions at the point of care.
• Expanded patient documentation: Large language models (LLMs) will efficiently integrate documentation from scientific manuscripts to clinical trial records and regulatory filings. Organizations will increasingly automate Clinical Study Report (CSR) authoring using data from electronic clinical systems and statistical computing environments (SCE).
• Faster trial recruiting: AI models will harness electronic health records (EHRs) and real-time patient data to quickly identify optimal clinical trial candidates with unmatched accuracy.
• Enhanced AI compliance: Wider adoption of AI-driven compliance platforms will enable real-time monitoring of regulatory adherence across clinical trials and drug manufacturing. This trend will improve how companies ensure compliance and maintain operational integrity, ultimately fostering a more efficient and reliable regulatory landscape.

Financial services

• Analysis of unstructured data: Well-tuned LLMs will use context-aware, domain-specific, and highly accurate classifications and extractions to rapidly perform entity recognition and summarization. In practical terms, what this means is that risk managers can analyze unstructured banking and insurance data such as contracts in seconds rather than weeks.
• Delayed agentic solutions: Implementing agentic solutions such as credit approvals and other operational choices will likely slow due to increasingly stringent regulations and compliance requirements. Institutions will find themselves having to provide detailed answers to questions like, “why and how was this decision made?" However, the delay will be temporary and agentic AI adoption ought to be in full swing by 2026.
• Increased model risk management challenges: We expect to see an escalation of model risk management issues as LLMs and AI coding assistants speed-up complex model development. This surge in model creation, along with increased regulatory requirements, may overwhelm traditional governance workflows. The result will be bottlenecks in approvals, delays in productionizing models, and increasing operational and compliance risks. Banks will need to transform roles, processes, and technologies to scale governance and optimize model pipelines for innovation.
• Greater automation: To reduce costs in the face of profitability pressures, banks will increasingly leverage AI to automate repetitive tasks across compliance, customer service, HR, finance, and risk management. These AI solutions will streamline decision-making, optimize resources, and accelerate workflows, lowering costs while fostering innovation. This rapid adoption of AI will naturally bring increased regulatory scrutiny and operational complexity, but banks with mature model development and MLOps, governance platforms, cloud-first infrastructures, and well-organized data will gain a decisive advantage.

Public sector

• Increasing balkanization: International AI development will continue to become divided, with countries and regions carving out their own paths driven by unique regulations, national interests, and strategic goals. The result will be technological capabilities and standards that vary significantly. While some regions may prioritize ethical and transparent AI, others may focus on strategic dominance, resulting in a competitive environment marked by diverging norms and increased geopolitical tension. To counteract this trend, governments, industry leaders, and international organizations must collaborate through global forums and partnerships to establish unified standards in ethical AI, data security, and transparency. Such efforts can help bridge regional divides, build mutual trust, and reduce the risks of competing AI agendas.
  
• Operational AI in defense: The defense industry will likely see a surge in nine-figure contracts with specialized AI tech providers. This marks a pivotal shift from experimental pilots to full-scale operational AI deployments such as the U.S. Army’s Tactical Intelligence Targeting Access Node (TITAN) Ground Station. Non-traditional software and AI tech providers are transitioning from peripheral innovator roles to central “prime” contractor positions within the defense ecosystem. The rise of these AI primes will modernize defense procurement, accelerating the deployment of advanced AI-driven capabilities across the military, where they are urgently needed to preserve U.S. supremacy on an evolving AI-driven battlefield.
• Ownership of responsible AI shifts from public sector to private: The current administration has triggered major policy shifts likely to lessen the role of the federal government and the public sector in guiding and regulating AI. Governance of AI applications is, of course, a critical pillar in successful AI adoption, but this role will increasingly fall to corporations and private industry organizations.
  

Next steps

Looking to explore more topical trends? RSVP for an upcoming webinar on the future of AI or check out Domino's resource library for more content to binge.