近期，谷歌的AI概述功能在回答用户关于“1995年是否是30年前”的问题时，出现了一系列计算错误，多次将1995年计算为29年前，引发广泛关注和讨论。尽管AI技术发展迅速，但此次事件暴露了其在基本数学运算和事实核查方面仍存在不足。用户在社交媒体上分享了大量谷歌AI给出错误答案的截图，甚至在同一查询下出现“29年前”和“30年前”的矛盾回答。此前，谷歌AI概述已多次出现“信口开河”或事实性错误的情况，例如编造虚假俚语含义、混淆年份或在体育赛事中误读规则。尽管存在这些问题，谷歌仍继续向用户推送AI概述内容，这促使用户在获取信息时应保持警惕，并建议用户在依赖AI答案前进行二次核实。

💡 AI近期在处理基本时间计算时出现失误，谷歌AI概述将1995年错误地计算为29年前，而非实际的30年前，这引起了用户对AI准确性的质疑。

🌐 用户在社交媒体上广泛分享了谷歌AI给出错误答案的截图，甚至在同一查询下出现“29年前”和“30年前”的矛盾回答，显示出AI在一致性和准确性上存在问题。

⚠️ 这并非谷歌AI概述首次出现错误，此前已有编造虚假俚语含义、混淆年份、误读体育规则等情况，表明AI在事实核查和可靠性方面仍需改进。

⚠️ 尽管存在技术上的不足，谷歌仍继续向用户提供AI概述功能，用户在获取信息时应保持审慎态度，建议进行二次核实，以避免被错误信息误导。

In the year 2025, it seems we're hearing frequent warnings about impending doom via artificial intelligence. And that's not without some merit: Heck, AI is causing job loss and filling our social feeds with fairly convincing slop.

But sometimes, the world sends a reminder that our AI tools are a far cry from Schwarzeneggerian domination. For instance, Google's much-maligned AI Overviews is, at times, struggling to decipher if 1995 was, in fact, 30 years ago. For the mathematically challenged, yes it was, because, well, 2025 minus 1995 equals 30. But don't go telling that to Google.

People online noticed Google's AI Overviews seemed to have trouble with this particular query. It became almost a trend of sorts to ask, "Was 1995 30 years ago?" Lots of people posted screenshots appearing to show Google answering with incorrect answers — typically speaking, AI Overviews decided 1995 was 29 years ago, for some reason.

We tested the query at Mashable, and did — at least sometimes — replicate the results of these screenshots and get an incorrect answer from Google's AI Overviews

Here is a screenshot from a Mashable editor.

When I tried to recreate the results, Google, at first, correctly answered the question. But later, it answered both incorrectly and correctly. The AI Overview spat out a garbled response that said 1995 was both 29 years ago and 30 years ago. Mashable has reached out to Google for comment and will update as necessary.

We've covered Google's often confidently wrong AI Overviews at Mashable. There was that trend where it would invent meaning for any made-up idiom. Or when it struggled to get the year right. Or the time it got the NFL rules wrong during a high-profile game, causing a lot of confusion.

Despite these repeated issues and mistakes, Google continues to feed users more AI Overviews. Remember to double-check your sources...or maybe just break out your calculator.