What are the risks of using AI for investment analysis?

The primary risks of using AI for investment analysis are overconfidence without calibration, hallucinated or fabricated supporting evidence, sycophantic reasoning that confirms rather than challenges the analyst's hypothesis, and data sovereignty exposure when sensitive deal materials are processed through shared infrastructure. Research shows that AI systems trained through standard methods are systematically overconfident and that users cannot reliably distinguish between correct and incorrect AI outputs.

By Scot Thom, Co-Founder & CEO, DiligenceWorks · Updated

Overconfidence and calibration failure

<p>AI systems trained through reinforcement learning from human feedback are systematically overconfident. Research demonstrates that the training process encodes human annotators' bias against uncertainty expressions — annotators consistently prefer confident-sounding answers, which trains the model to present uncertain conclusions with unwarranted certainty.</p> <p>In investment analysis, this means an AI system may present a revenue projection assessment with high confidence even when the underlying evidence is weak or contradictory. The analyst sees a definitive-looking output and has no reliable way to distinguish genuine confidence from trained-in overconfidence.</p>

LLMs are systematically overconfident because RLHF training encodes human annotators' bias against uncertainty expressions. Zhou et al. (2024), arXiv:2401.06730

Sycophantic reasoning in deal evaluation

<p>AI alignment training creates a second, related risk: sycophancy. Systems trained to be helpful learn that agreement is rewarded. When an analyst asks an AI system to evaluate a deal, the system is structurally incentivised to find reasons the deal works rather than reasons it does not.</p> <p>This is the opposite of what due diligence requires. The value of independent analysis comes from its willingness to contradict the prevailing narrative. An AI system that confirms the sponsor's thesis is not performing diligence — it is performing marketing.</p>

Data sovereignty and information leakage

<p>Most commercial AI tools process data through shared cloud infrastructure. For fund managers evaluating proprietary deal flow, this creates a data sovereignty risk: the fund's information advantage — its proprietary analysis, its deal pipeline, its assessment of specific opportunities — may be processed alongside other clients' data on infrastructure the fund does not control.</p> <p>Sovereign, single-tenant deployment eliminates this exposure. When the analysis infrastructure runs on the fund's own systems, in the fund's own jurisdiction, the information boundary is absolute rather than contractual.</p>

See it in action

Book a live demo with your own deal data. Your box. Your data. Your country.

Trusted by Bluzand Group Sovereign deployment No credit card required