AI in quantitative investing: A buy-side perspective
Artificial intelligence has moved from experimentation to in-production use-cases across quantitative investment workflows, although buy-side firms remain focused on applications that deliver measurable productivity gains rather than replacing established investment processes.
Based on a global survey of 68 buy-side professionals, this whitepaper examines how firms are deploying natural language processing (NLP), large language models (LLMs) and generative AI across their research and modeling functions. The findings show that adoption is gathering momentum, with firms increasingly using AI in production even though enterprise-wide deployment remains limited.
Information search and retrieval, document summarization and coding are currently delivering the greatest value, helping investment professionals navigate large volumes of unstructured information and reduce manual workloads. At the same time, firms are taking a pragmatic approach to AI architectures, combining commercial and proprietary models with traditional NLP, prompt engineering and emerging techniques such as retrieval-augmented generation. Confidence remains the biggest barrier to scaling adoption: hallucinations, inaccurate outputs, compliance, explainability and data quality are key concerns. Human oversight remains central, with people involved in reviewing or validating AI outputs more than 90% of the time.
Key findings:
- AI adoption is moving into production—33% of firms report limited production use, while 28% have multiple production use-cases; only 13% describe deployment as enterprise-wide.
- Productivity leads the value proposition—Information search and retrieval is the leading use-case, cited by 82% of respondents, followed by document and text summarization (74%).
- AI is augmenting rather than replacing investment professionals—Firms are initially applying AI to information-intensive workflows rather than core investment decisions.
- Human oversight remains essential—More than 90% of firms involve people in reviewing or validating NLP and LLM outputs.
- Trust is the biggest barrier—59% cite hallucinations or inaccurate outputs as a major obstacle, ahead of compliance and auditability at 40%.
- AI is expected to boost productivity, not necessarily alpha—57% see LLMs becoming an important productivity tool, compared with 24% who see potential for competitive advantage.
Move from AI experimentation to trusted, measurable impact. Download this whitepaper to identify the research and investment workflows where AI can deliver immediate productivity gains, while putting in place the data, governance and human oversight needed to ensure reliable outputs.
Download the whitepaper
Register for free access to hundreds of resources.