AI’s impact in investing: Value vs. hype

Key takeaways

• We see generative AI driving real efficiency and research acceleration, though the value differs widely across use cases.
• Differentiation increasingly depends on proprietary data, creative integration, and disciplined governance, rather than access to the same tools.
• We believe investors need to separate true research enhancements from basic AI-driven efficiencies.

Generative AI is moving beyond experimentation and into everyday integration. Broad surveys covering more than 42,000 business leaders show positive return on investment, and asset managers are reporting similar benefits.

For investment teams, the early gains have largely come from efficiency. Quantitative researchers now use large language models (LLMs) to extract data, speed up code development and back-test. Fundamental teams apply the same tools to summarize earnings calls and filings and assist with first-pass valuation modelling. These changes matter. They have significantly accelerated the research process and reduced time spent on routine tasks.

But because these types of tools are now widely available, they no longer set one manager apart from another. The baseline has moved. In short, these efficiency gains help, but they are not differentiating.

In our opinion, differentiation emerges when AI is used to strengthen human judgment rather than replace it. We see several areas where firms taking this approach are generating real value. Some examples include:

• Proprietary data pipelines. The advantage here comes not from having AI tools, but from the data behind them and the methods used to turn that data into insight.
  
  
• Creative model application. Some teams are moving beyond standard text models—using network-based signals, synthetic scenario generation, or forward-looking cash-flow analytics. These techniques build on existing strengths rather than copying common tools.
  
  
• Always-on awareness. Shifting from point-in-time updates to continuous monitoring gives analysts a more dynamic read on commentary, market reactions, and portfolio signals.

These shifts align with the evolution of Russell Investments’ own manager research process. Our METRiQ tool now auto-generates visuals, tests a strategy’s structural elements, and uses generative AI to identify core strategy characteristics and watchpoints. Importantly, tools like these do not replace human judgment. They help focus attention where it matters—identifying “diamonds in the rough” through machine-learning-based screens and flagging subtle shifts in a manager’s stated edge or portfolio 

As AI becomes a standard part of investment workflows, a new challenge is emerging: separating genuine research enhancement from improvements that come only from generic automation. By generic automation, we mean common AI tasks that now come off the shelf—summaries, first-pass analysis, and routine document processing that speed up work but do not create differentiated insight.

There are a few warning signs that show when an AI-driven improvement is not truly differentiating. For instance, crowded signals and alpha decay are appearing in quant strategies that rely on similar text embeddings or sentiment models. Meanwhile, some research teams are beginning to sound more alike as they lean on common LLM interpretations. There is also a risk of overreliance, where repeated use of the same tools contributes to deskilling among research analysts.

At the same time, AI is helping our manager research analysts pinpoint where genuine skill shows up. New analytics now separate stock-selection decisions from portfolio-construction choices and measure timing effects across industries and factors. By linking these analytics to our deep database of manager records, we can interpret them in the context of each manager’s documented history to pinpoint where value truly comes from. Examples include managers whose security level return levels matter more than hit rates, or whose disciplined entry-exit decisions support long-term outperformance.

In this sense, AI is not the source of alpha. It is the lens that makes sustainable alpha easier to see.

We believe the next phase of AI integration will move from summarization toward situational awareness, supported by systems that track information over time and deliver context rather than content. This reinforces a central belief of ours: AI does not level the playing field but widens it. Managers who combine proprietary inputs, strong governance, and creative application are pulling ahead, while those relying only on common tools risk blending into a crowded middle.

We believe AI will continue to reshape research depth, speed, and structure. The priority for investors is distinguishing true process enhancement from widely available automation. Evaluating data advantage, governance discipline, and integration creativity is becoming as important as assessing traditional investment skill. Ultimately, the real value lies not in the AI tools themselves, but in how managers use them.