Goldman Sachs partner warns AI could erode bankers’ reasoning skills, 2026 risk assessment

AI's rapid rise in banking and Goldman Sachs' playbook

Over the past five years, investment banks have accelerated the deployment of large‑language models and generative AI to automate routine tasks, from drafting client pitches to parsing market data. Goldman Sachs, a bellwether for Wall Street, has invested heavily in its internal "AI Lab" and partnered with tech firms to embed models into trade‑execution platforms and risk‑management dashboards.

The firm touts the technology as a way to cut costs, speed up decision‑making and stay competitive against fintech challengers. Recent internal reports claim that AI‑driven tools have shaved up to 30 percent of analyst hours on earnings‑call summarisation and have helped surface hidden arbitrage opportunities in real time.

A senior partner sounds the alarm on reasoning atrophy

During a closed‑door conference in New York, a senior technology partner at Goldman Sachs warned that the same algorithms designed to boost efficiency could unintentionally dull the critical thinking of the next generation of bankers. The partner argued that when AI drafts memos, generates valuation models or suggests hedging strategies, junior staff may accept outputs without probing the underlying assumptions.

According to the partner, the danger lies not in AI making mistakes, but in humans losing the habit of questioning those mistakes. "If you hand a model a spreadsheet and it spits out a recommendation, the temptation is to trust the answer rather than test it," he said, echoing concerns raised by academia about over‑reliance on algorithmic outputs.

Why the warning matters for the financial system

Bankers' reasoning skills are the final safeguard against systemic shocks. In crisis moments—such as the 2020 market crash or the 2023 sovereign debt turmoil—human judgment has historically corrected model‑driven blind spots. If future analysts grow accustomed to accepting AI suggestions without scrutiny, the industry could see a rise in undetected model bias, mis‑priced risk and cascading errors across portfolios.

Moreover, regulatory frameworks still expect banks to demonstrate “sound judgment” in capital‑adequacy calculations and client‑suitability assessments. A workforce that leans heavily on opaque AI black boxes may struggle to meet those standards, inviting stricter oversight or fines from bodies like the SEC and the Basel Committee.

The ripple effect on Africa’s burgeoning finance sector

African banks and fintech startups are watching Wall Street’s AI experiments closely, hoping to leapfrog legacy infrastructure. Nations such as Kenya, Nigeria and South Africa have already piloted AI‑assisted credit scoring and trade‑finance bots. If the concerns raised at Goldman Sachs translate into a broader industry shift, African institutions could face a paradox: the lure of cost‑saving AI versus the need to nurture home‑grown analytical talent.

In markets where data quality is uneven and regulatory guidance on AI is still evolving, a premature hand‑over of decision‑making to algorithms could exacerbate credit‑allocation errors. Conversely, the warning may spur African banks to embed “human‑in‑the‑loop” policies, investing in training programs that keep quantitative reasoning at the core of their culture.

Regulators, academia and the industry respond

U.S. regulators have begun drafting guidance on the use of generative AI in finance, emphasizing model transparency and auditability. The Federal Reserve’s recent discussion paper calls for banks to retain “independent verification” of AI‑generated outputs. In Europe, the European Banking Authority is piloting a sandbox that forces firms to document how AI influences risk‑weighted assets.

Academic circles are echoing the partner’s caution. A 2025 study from the Wharton School found that finance students who relied on AI for case‑study solutions performed 15 percent worse on follow‑up reasoning tests. Meanwhile, industry groups such as the Financial Stability Board are convening working groups to draft best‑practice guidelines that balance innovation with human oversight.

What banks can do to preserve critical thinking

Experts suggest a three‑pronged approach. First, embed “explainability” layers into AI tools so that users see the data sources and assumptions behind each recommendation. Second, redesign onboarding curricula to require junior bankers to reproduce AI‑generated models manually before they can rely on them, reinforcing the mental models that underlie quantitative analysis. Third, create cross‑functional audit teams that periodically challenge AI outputs with stress‑tests and scenario analyses.

For African institutions, the strategy may also involve partnerships with local universities to develop AI‑ethics labs and to source talent that understands both the continent’s data nuances and the limits of global models. By turning the warning into a roadmap, banks can harness AI’s speed while safeguarding the reasoning muscle that has long been the industry’s competitive edge.

Quick Answers

What specific risk did the Goldman Sachs partner highlight about AI in banking?
He warned that over‑reliance on AI could weaken future bankers' reasoning skills, leading them to accept algorithmic outputs without critical scrutiny.

How could AI‑driven reasoning loss affect African banks?
It could increase credit‑allocation errors in markets with limited data quality and push African banks to adopt stricter human‑in‑the‑loop policies to protect decision integrity.

What steps are recommended to keep human judgment strong alongside AI?
Introduce explainable AI, require manual reconstruction of AI models during training, and set up audit teams that routinely challenge algorithmic recommendations.

Source: www.cnbc.com