Popular language models have identified the most frequent questions posed by cryptocurrency users in 2025, covering a range of topics from trading strategies to asset protection. The experiment was conducted by the editorial team at Wu Blockchain.
The researchers employed ChatGPT, Grok, Perplexity, Claude, Gemini 3, and DeepSeek. To avoid duplicate data from search results, the models were prohibited from using real-time web searches. AI generated responses based on long-term discussion patterns embedded in their databases.
Model focus areas:
- ChatGPT took on the role of a trader, focusing on identifying the peak of the bull market, seeking “alpha,” and assessing structural risks of projects;
- Grok reflected social media narratives, highlighting the impact of halving, market phase debates, and capital inflow into ETFs;
- Claude adopted a cautious approach, emphasizing market entry timing, project scam checks, and risk management for newcomers;
- Gemini 3 delved into industrial trends, with top queries including RWA, the impact of regulation on personal assets, and competition among L2, Solana, and AI tokens;
- Perplexity and DeepSeek provided more general results akin to a search summary: price forecasts, fraud protection, and regulatory changes.
Analysts observed that Chinese models tend to focus more on “hard” variables such as cycles, halving, and regulation. They also noted a dependency on AI capability: advanced models provide more structured answers, while less sophisticated ones tend to generalize about prices and security.
Overall, all queries converge on three areas: cycle positioning, profit paths, and risk boundaries.
Earlier, a team of humans was outperformed by artificial intelligence in a trading tournament hosted by Aster.