Mohammad Nazzal

AI and the Preference Paradox: When Larger Models Think More Like Humans

Large language models do not behave neutrally in economic decisions. In preference tasks, more advanced models become increasingly human-like and irrational. In belief tasks, they become more statistically rational. Simple role priming modestly improves performance, but deeper debiasing fails. AI agents inherit—and selectively correct—human behavioral distortions .

AI and the Adoption Divide: Who Captures the Frontier, Who Falls Behind

A tractable firm-level framework explains why countries grow at the frontier’s rate yet remain persistently behind it. Absorptive capacity, institutions, credit frictions, and technology appropriateness determine relative income levels—not long-run growth. Development becomes a question of incentive design, selection, and adoption speed rather than invention alone.

AI Adoption Is Not Dividing Along Ideology - It Is Sorting Along Structure

Democrats report higher workplace AI use and exposure than Republicans, but differences largely reflect education and occupational sorting rather than ideology. Once controlling for industry and job characteristics, partisan gaps disappear. AI adoption follows structural labor market patterns, suggesting technology diffusion mirrors human capital distribution more than political affiliation.

The Coming Information Crisis: Why AI Is Destabilizing What We Know

Artificial intelligence increases the speed and scale of information distribution, but it also weakens the economic foundations that sustain credible journalism and research. As misinformation becomes cheaper and truth more costly to produce, the information ecosystem faces structural instability, requiring strategic intervention to preserve long-term knowledge integrity.

The Innovation Illusion: Why AI Alone Won’t Transform R&D

Artificial intelligence holds the promise of accelerating scientific discovery, engineering breakthroughs, and productivity across industries. Yet the magnitude of its impact depends on three factors: the share of research tasks AI can perform, how productive AI is at those tasks, and the bottlenecks that limit overall progress. The paper presents a structured model showing that even extremely powerful AI produces limited gains if it operates on only a minority of research tasks. Conversely, broad automation across the research pipeline—far more than incremental improvements—determines whether...

The Missing Data Layer That Makes or Breaks Your SWOT

Traditional SWOT often becomes outdated and superficial without real-time data. Integrating AI and machine learning allows organizations to quantify strategic factors, prioritize risks and opportunities, and refresh analyses regularly, transforming SWOT into a dynamic, credible decision support tool.