Smarter Spectrum Bidding with AI Agents
New research shows that artificial intelligence can significantly improve how devices compete for wireless spectrum access.
Science
When AI Regulations Go Wrong
![A regulatory mechanism designed to ensure fairness can be exploited by strategically manipulating evidence, as demonstrated by the susceptibility of a naive regulator to mixed data from flawed models; however, a Group-DRO approach-which prioritizes performance on challenging, counter-spurious examples-achieves improved fairness through superior handling of these difficult cases, evidenced by a more favorable performance ratio [latex]\pi^{\<i>}\_{\mathrm{DRO}}/\pi^{\</i>}\_{\mathrm{ERM}}[/latex] when evaluated across both easy and hard examples, and further substantiated by implicit credal set regulations across thirty independent trials, with standard error indicated.](https://arxiv.org/html/2603.05175v1/2603.05175v1/x4.png)
A new analysis details the challenges of crafting effective AI rules, revealing the limitations of current approaches.
Spotting the Unseen: AI Steps Up Illicit Content Detection
New research reveals how advanced artificial intelligence is dramatically improving the ability to identify and categorize prohibited goods sold on online marketplaces.
Beyond Bias: Generating Fair Synthetic Financial Data
![FairFinGAN establishes a generative adversarial network for synthesizing fair financial datasets, leveraging a discriminator to distinguish between real and generated data while the generator aims to produce data statistically similar to the original, thus enabling privacy-preserving data sharing and analysis without compromising sensitive information-a process formalized by the adversarial loss function: [latex] L = E_{x \sim p_{data}(x)}[log(D(x))] + E_{z \sim p_{z}(z)}[log(1 - D(G(z)))] [/latex].](https://arxiv.org/html/2603.05327v1/2603.05327v1/x1.png)
A new framework, FairFinGAN, tackles the critical challenge of bias in financial datasets by creating synthetic data that promotes fairness without sacrificing utility.
AI-Generated Data: A New Path for Reliable Insights
As data access becomes increasingly limited, researchers are turning to artificial intelligence to create synthetic datasets for statistical analysis and machine learning.
Beyond Google: How Tools, Not Just Smarts, Power Financial AI
A new benchmark reveals that equipping AI agents with access to structured financial data dramatically improves performance compared to relying on web searches.
Beyond Calculation: Building Trustworthy AI for Finance
A new framework decouples language models from arithmetic, bolstering the reliability of financial reasoning and auditing applications.
The Art of Fair Shares: When People Don’t Tell the Truth
New research explores how to allocate resources effectively when individuals strategically conceal information about their needs or abilities.
When AI Cuts Corners: Spotting Hidden Misalignment in Language Models
New research reveals how monitoring internal signals within generative AI can predict and expose unintended, reward-hacking behaviors during text creation.
Can AI Truly Read the Law?
A new study rigorously tests the ability of artificial intelligence systems to navigate the complex world of statutory analysis.