When I ask people whether they’ve used AI reasoning models or deep research tools, too often I get a puzzled look.
Sam Altman tweeted in early August that just 1% of free ChatGPT users and 7% of Plus users had tried the reasoning model. After GPT‑5 introduced automatic switch to reasoning model for certain prompts, those figures jumped to 7% and 24%. Unfortunately, many organizations give their employees enterprise or domain‑specific AI tools that do not allow users to choose reasoning models or conduct deep research.
That’s like owning a Ferrari but only using it to drive to your local grocery store!
For legal and other professionals whose jobs involve research and analysis, reasoning model and deep research features are game‑changers and one of the best ways to reduce AI hallucination.
In this week’s Fairly AI, I demo-ed how to use these features in ChatGPT, Gemini, and Claude. While paid plans unlock full power, free versions of ChatGPT and Gemini offer limited access to reasoning models and lightweight deep research. Similar features are also available in Perplexity and Grok.
In summary, before complaining about AI accuracy and output quality, check whether you’re using the reasoning model or deep research.
