I believe you're right, even though I don't have any evidence except for my own experience.
This issue becomes very clear when you see how many ways there are to express a simple concept like linear regression. I've had the chance to see that for myself in university when I pursued a bunch of classes from different domains.
The fact that introductory statistics (y = a + bx), econometrics (Y = beta_0 + beta_1 * X) and machine learning (theta = epsilon * x, incl. matrix notation) talk about the same formula with quite different notation can definitely be confusing. All of them have their historical or logical reasons for formulating it that way, but I believe it's an unnecessary source of friction.
If we go back to basic maths, I believe it's the same issue. Early in my elementary school, the pedagogical approach was this:
0. only work with numbers until some level
1. introduce the first few letters of the alphabet as variables (a, b, c) - despite no one ever explaining why "variable" and "constant" are nouns all of a sudden
2. abruptly switch to the last letters of the alphabet (x, y, z), two of which don't exist in my native language
3. reintroduce (a,b,c) as sometimes free variables, and sometimes very specific things (e.g., discriminant of a quadratic equation)
4. and so on for greek letters, etc.
It's not something that's too difficult to grasp after some time, but I think it's a waste to introduce this friction to kids when they're also dealing with completely unrelated courses, social problems, biological differences, etc. If you're confused by "why" variables are useful, why does the notation change all the time, and why it sometimes doesn't - and who gets to decide - this never gets resolved.
Not to mention how arbitrarily things are presented, no explanation of how things came to be or why we learn them, and every other problem that schools haven't tackled since my grandparents were kids.
If you're writing a scientific article with the results of your research, your audience (i.e., reviewers and other readers, usually scientists) will expect you to have evidence for your claims. Any speculation or other "maybe's" should be kept in the sections like discussion or future work, or left out.
If you're writing something where opinions are accepted or even appreciated, like a position paper or a blog post, the audience should be aware that whatever you say is your interpretation anyway. But then, it also applies that some interpretations or guesses are more "educated" than others.
"Maybe" does sound too vague. Although it sounds like a language trick, I prefer to use a disclaimer at the beginning or "assume", "expect", "my understanding", to show that whatever follows is my opinion, but also that I think it's still valid and I'm willing to stand behind it.
I'm working on ways to allow developers and deployers of LLMs to express how and why their overall system is compliant and adversarially robust, and what to do when that's not the case.
Specifically, my team and I are making assurance cases and ontologies that can seamlessly integrate with the system and its guardrails. For example, if you want to deploy some mix of filters underneath a user-facing LLM app, you would able to:
1) express the logic of how they should be deployed and why (e.g., if X=1, then Y, else Z);
2) see how they perform over time and evaluate alternatives;
3) investigate what happened when an attack succeeds;
4) prove to the auditors that you're taking all measures necessary to be robust and compliant with the EU AI Act.
It started as an informal collab early this year, but we have since published a few workshop papers on this concept [1,2]. We're building a Python demo that would show how it all fits together.
This issue becomes very clear when you see how many ways there are to express a simple concept like linear regression. I've had the chance to see that for myself in university when I pursued a bunch of classes from different domains.
The fact that introductory statistics (y = a + bx), econometrics (Y = beta_0 + beta_1 * X) and machine learning (theta = epsilon * x, incl. matrix notation) talk about the same formula with quite different notation can definitely be confusing. All of them have their historical or logical reasons for formulating it that way, but I believe it's an unnecessary source of friction.
If we go back to basic maths, I believe it's the same issue. Early in my elementary school, the pedagogical approach was this: 0. only work with numbers until some level 1. introduce the first few letters of the alphabet as variables (a, b, c) - despite no one ever explaining why "variable" and "constant" are nouns all of a sudden 2. abruptly switch to the last letters of the alphabet (x, y, z), two of which don't exist in my native language 3. reintroduce (a,b,c) as sometimes free variables, and sometimes very specific things (e.g., discriminant of a quadratic equation) 4. and so on for greek letters, etc.
It's not something that's too difficult to grasp after some time, but I think it's a waste to introduce this friction to kids when they're also dealing with completely unrelated courses, social problems, biological differences, etc. If you're confused by "why" variables are useful, why does the notation change all the time, and why it sometimes doesn't - and who gets to decide - this never gets resolved.
Not to mention how arbitrarily things are presented, no explanation of how things came to be or why we learn them, and every other problem that schools haven't tackled since my grandparents were kids.
reply