COMPARATIVE ANALYSIS OF AI-DRIVEN DECISION SUPPORT SYSTEMS AND TRADITIONAL SPREADSHEETS: EVALUATING ACCURACY AND CONSISTENCY IN BUSINESS INTELLIGENCE

Abstract

The paper uses a design science approach to empirically analyse the relative performance of AI-controlled decision support systems (AI-DSS) compared to traditional spreadsheets for startup decisions.  Approach: Drawing on a mixed-methods approach, survey data from 200 startup founders and their teams of up to three members (n = 400) were analysed; field interviews were conducted with 20 selected keynote participants attending the conference overall, as well as in parallel, one-on-one structured follow-up observational studies; and a code review experiment was also run.  AI-DSS did much better than spreadsheets in important ways: it made decisions 16% more accurately (p≤0.001), it saved 35% more time (p<0.01), and the standard deviation of decisions got a lot better (SD = 1/7 vs SD = 0.033). Observational findings suggest superior data integration strength of AI-DSS compared with traditional DSS in various datasets flexible enough to handle complex environment dynamics.  However, we identified a steeper learning curve for adopting AI-DSS, along with challenges related to data quality, algorithmic transparency, and ethical considerations.  The work extends the literature on AI-DSS by demonstrating how much of a competitive advantage can be realised in high-stakes startups, questioning traditional speed-accuracy tradeoffs in decision-making and roles for AI that support organizational flexibility.  The implications of the findings are also relevant to start-ups' strategic thinking, investor choices, and teaching for entrepreneurship, having a potential impact on how this research intends to support decision-making in turbulent business environments.