Recent watermarked generation algorithms inject detectable signatures during language generation to facilitate post-hoc detection. While token-level watermarks are vulnerable to paraphrase attacks, SemStamp (Hou et al., 2023) applies watermark on the semantic representation of sentences and demonstrates promising robustness. SemStamp employs locality-sensitive hashing (LSH) to partition the semantic space with arbitrary hyperplanes, which results in a suboptimal tradeoff between robustness and speed. We propose k-SemStamp, a simple yet effective enhancement of SemStamp, utilizing k-means clustering as an alternative of LSH to partition the embedding space with awareness of inherent semantic structure. Experimental results indicate that k-SemStamp saliently improves its robustness and sampling efficiency while preserving the generation quality, advancing a more effective tool for machine-generated text detection.

Traditional recruiting methods have typically had a poor track record at matching candidates with employers. San Francisco-based startup Scoutible is betting its AI-based gaming solution can do better. Most seasoned hiring managers know the sinking feeling that comes with realizing within months of onboarding that a new professional is ill-suited to the role. The pressing work that prompted the hire in the first place may stall, eliciting outcry from stakeholders and frustrating colleagues charged with picking up the slack. Meanwhile, the prospect of letting the employee go and starting the search anew creates even more headaches--not to mention added expense.

Hiring someone who turns out to be a bad fit can be costly: Unhappy employees cost the U.S. economy between $450 billion and $550 billion in lost productivity each year, according to research firm Gallup. And replacing a full-time worker can cost up to twice the employee's salary. While working on a project at Harvard Law School, Angela Antony found herself immersed in statistics like those. "If you look across the economy, about 46 percent of hires leave within 18 months. That's despite all the time, resources, and billions of dollars spent trying to effectively hire," Antony says.