> poisoning attacks require a near-constant number of documents regardless of model and training data size
I fear this takeaway could be misinterpreted by non-experts.
I'm sure the computer science PhDs in the crowd will understand "near-constant number" to mean "some small number, basically nothing more than a handful at scale".
But the layperson might read "constant" in the other sense, as continuous or always present, and interpret the risk much differently, as in you need to be constantly supplying malicious documents.
I would urge them to use different terminology.
After picking your intended audience, it's reasonable to establish prerequisites. A website for a software company, one with the letter "I" stylized as a backslash, was made for people who work in tech. Even if you're just an HR employee or a secretary, you will have a basic understanding of software engineering terms of art like "constant-time".
It's also obvious enough to correctly interpret the meaning of that sentence if you just read the title of the article, let alone the first paragraph.
Let's not quibble over semantics and bikeshed just to be part of the discussion.
I don't think they're quibbling over semantics but providing constructive cautionary feedback. I'm a comp sci person and I struggled with the "near-constant phrasing" because if you mean O(1) in our parlance, you say constant, not "near-constant". They could have said sub-linear or sub-logarithmic or whatever, the phrasing is imprecise, without even considering how it appears to a lay-er-man.
Also I'm not a huge fan of defending jargon for the sake of it. Sometimes there are efficiency gains, sure. But the paper here is quite approachable generally speaking. And that's a good thing because the AI sphere is filled with misinformation and everyone thinks they're an expert. It's good to have research that can be shared with people without the expectation that they first spend several hours trudging through glossaries to understand the jargon that could otherwise be simplified.
> Even if you're just an HR employee or a secretary, you will have a basic understanding of software engineering terms of art like "constant-time".
Lol. No.
I had to do a double take for exactly the reason you mention here. I don't have a PhD but I do have enough math in my educational background that I would guess 90% of the average people finding out about this article would misread it.