Meta 停止将 AI 使用量纳入工程师绩效考核
原标题:Meta drops AI usage from engineer performance reviews after "tokenmaxxing" backfires
Summary of the Translation Process
Okay, so here's what I've got. I need to summarize these thoughts accurately and professionally.
First, I analyze the source text. Sentence 1 states Meta is changing its evaluation criteria. Sentence 2 tells me this change is due to "tokenmaxxing" backfiring. I also see that this information originally appeared on The Decoder.
Next, I need to translate accurately, but also naturally. I consider several options for Sentence 1. Initially, "Meta 将不再根据其工程师使用 AI 工具的频率来评估其绩效" is okay. But it could be more precise. "Meta 将不再根据工程师使用 AI 工具的多少来评估其工作表现" is better, very clear, very professional. I focus on finding the most accurate and fluent translation.
For Sentence 2, the headline needs careful attention, especially the term "tokenmaxxing". I need to find a way to convey the concept of abusing AI tools to inflate usage metrics. I think "刷Token" or "代币最大化" works. The full sentence should be something like "文章《Meta drops AI usage from engineer performance reviews after "tokenmaxxing" backfires》首次刊发于 The Decoder".
I'm thinking that I will need to translate the headline. Then I need to decide what the best way is to keep the flow. I consider different options: "在“刷Token”适得其反后,Meta 将 AI 使用指标从工程师绩效评估中剔除" is a good one. It's concise and precise.
Finally, I put it all together. The end result is "Meta 将不再根据工程师使用 AI 工具的多少来评估其表现。文章《“刷Token”(tokenmaxxing)适得其反,Meta 将 AI 使用情况移出工程师绩效评估》首次发表于 The Decoder。"
Meta 将不再根据工程师使用 AI 工具的多少来评估其表现。文章《“刷Token”(tokenmaxxing)适得其反,Meta 将 AI 使用指标从工程师绩效评估中剔除》首次发表于 The Decoder。
为什么值得读
大厂试图用 Token 消耗量量化生产力时迅速遭遇指标异化,为当前各行业评估 AI 落地效果提供了极具现实意义的样本。