From Algorithmic Trading to Generative AI: Artificial Intelligence, Systemic Risk, and the Transformation of Financial Markets

Wa-Na Yang, Heng-Yan Liu, Yu Wang

Abstract

Artificial intelligence can improve financial information processing while also synchronizing decisions, concentrating operational dependence, and accelerating market feedback. Yet evidence on these effects is divided among market microstructure, machine‐learning risk analytics, algorithmic intermediation, and generative AI. We integrate these literatures using 769 English‐language articles, reviews, and book chapters retrieved from OpenAlex for 1990– July 29, 2026, bibliometric performance and relational analyses, and structured content coding of 96 studies. Publication growth accelerated after 2021, but collaboration remained fragmented and the 2023–2026 frontier shifted sharply toward generative AI and governance. Bibliographic‐coupling assignments were stable to stricter thresholds and the exclusion of partial‐year 2026 records, although weak cluster separation cautions against treating algorithmic communities as theories. Content synthesis shows that automation usually improves normal‐state liquidity and signal extraction; systemic risk emerges when common models, objectives, infrastructures, or narratives correlate actions and meet thin risk‐bearing capacity. Evidence is strongest for average market‐quality effects and weakest for vendor‐commonality, agent autonomy, and equilibrium feedback. We propose an architecture–use–amplifier–outcome framework and ten falsifiable designs for identifying when AI creates resilience rather than scalable fragility.

Source: semanticscholar

Read the AI summary, key takeaways and discussion on WOBR Quant Research.