As of November 30, 2025, this prediction cannot yet be definitively evaluated.

Reasons:

1. Timeframe not reached
   Jason’s claim was explicitly multi‑year: “over the next several years.” From December 8, 2023 to November 30, 2025 is under two years, so we are still inside the forecast window rather than at its end. The fact that AI assistants are growing quickly does not tell us what the steady‑state multiplier versus web search will be in “several” (likely 3–5) years.

2. Lack of reliable, comparable metrics
   The prediction compares:

• “questions asked” to AI systems like Gemini/Bard (and similar assistants), vs.
• total volume of traditional web searches at the time of prediction.

But:

• Major search engines (Google, Bing) do not publish up‑to‑date, precise query counts; public estimates for Google vary widely and are often extrapolated from older disclosures or third‑party traffic tools.
• AI assistant providers (OpenAI, Google Gemini, Anthropic, Microsoft Copilot, etc.) generally report growth qualitatively (e.g., "hundreds of millions of users") or via revenue rather than disclosing total daily/annual query counts that could be cleanly summed and compared to web search volume.
• Third‑party analytics (Similarweb, Sensor Tower, etc.) give relative traffic estimates but not a unified, audited measure of all “AI questions asked” across products.

Because of this, there is no authoritative, quantitative series that would allow us to say with confidence that AI queries are now 10x–100x the baseline volume of web searches in late 2023, or even of comparable magnitude.

3. Current evidence is directionally supportive but not conclusive
   What we can say from public data and reporting is:

• Generative AI traffic and usage (e.g., ChatGPT and other major models) grew extremely rapidly through 2023–2025, indicating that “questions asked” to AI systems have indeed increased a lot.
• However, there is no credible public claim that total AI assistant queries now exceed traditional web search queries by an order of magnitude, and conventional search still appears to handle an enormous share of global information‑seeking behavior.

Given the incomplete timeframe and the absence of robust, directly comparable usage statistics, the only defensible verdict is that the prediction’s correctness is inconclusive (too early to tell) rather than clearly right, wrong, or permanently ambiguous.