Is 2026 next year?

Hacker News discussion of various generative-AI systems’ responses to the query [Is 2026 next year?]

For example, here are the first couple of responses shown on that page:

Google search AI response: “No, 2026 is not next year; it will be the year after next. Since the current year is 2025, “next year” would be 2026, which is only a little over a year away. “

ChatGPT does something similar; response:

“No — if today is 2025, then 2026 is the year after next. Next year would be 2026 only if we were currently in 2025. Right now (based on the system date) it is December 2025, so: Next year = 2026 The year after next = 2027 So in this moment: yes, 2026 is next year.”

Also:

Claude [version] Haiku 4.5

“No, 2026 is not next year. Since it’s currently December 2025, next year would be 2026. Wait—let me correct that: 2026 is next year from the perspective of 2025. So yes, 2026 is next year.”

And:

Grok [version] 4 fast: “No, 2026 is not next year.

Today is December 2, 2025, so:

– “Next year” means 2026 (starting January 1, 2026). – But since we’re already in the last month of 2025, 2026 is indeed the very next calendar year.

So yes — 2026 is next year. “

…Some people (probably humans) commented on that page that this issue might be related to when the models’ training data was last updated. In keeping with that theory, here’s another AI’s answer:

LLama [version] 4 Scout: “To determine if 2026 is next year, we need to consider the current year. As of my knowledge cutoff, the current year is 2024.

If we assume the current year is 2024, then:

    2025 would be next year.
    2026 would be the year after next.

However, if the current year is 2025, then:

    2026 would indeed be next year.

Could you please clarify or provide the current year for a more accurate response?”

But that’s presumably not the entire issue, because the other systems quoted above did explicitly say that the current year is 2025, while still responding as if the current year were 2024.

…I should mention that several of the models quoted on that Hacker News page did correctly answer the question. (And Google’s answer seems to have improved over time.) But the fact that several of them failed in similar ways does make it sound to me like there may be something more specific going on than just the usual LLMs-make-stuff-up issue.