I took a close look at CFO Connect’s new report on The State of AI in Finance. The adoption story is strong, but the reality is more grounded—AI is still mostly used for basic tasks, while bigger challenges like fragmented data and output reliability remain. Surprisingly, the report doesn’t meaningfully address hallucinations at all.

The AI Trial Balance Trial: Chat GPT low effort failure, Gemini won't even try and Claude tried hard but failed harder. The results were clear: general-purpose AI still can’t handle even basic accounting reporting reliably.

A real-world EASY test of AI tools for gathering public company financials reveals major accuracy issues. Perplexity and Gemini both failed to produce reliable data, while ChatGPT performed best. A cautionary look at how generative AI handles earnings reports and SEC filings.

With so much hype in the AI world, prediction markets like Polymarket offer a refreshing way to cut through the noise. Instead of opinions, people put real money behind their beliefs — giving us a more grounded snapshot of which AI models are viewed as the top performers. Polymarket currently ranks Google’s Gemini at the top, followed by OpenAI’s ChatGPT and Elon Musk’s Grok. But these standings are based on Arena benchmark scores — not revenue or usability. For business le