Profile avatar
aewolfai.bsky.social
I'll be posting my thoughts on various AI platforms for teachers and promoting good uses of LLMs. Read my thesis https://repository.usfca.edu/thes/1547/
56 posts 26 followers 16 following
Regular Contributor
Active Commenter
Posts 25 Comments 25

I hate wholesale rejections because I think it ignores creative avenues for learning. If the real question is the ethics, would a more ethically produced AI like Bloom be a better alternative? If it's the pedagogy and "meeting students where they are at," what happens if that's -in part- some AI?

submitted 25 days ago • 0 comments

I disagree with the conclusions while agreeing with almost all of the premises. Let's talk about it if you have the time :) www.linkedin.com/posts/anthon...

submitted 32 days ago • 0 comments

On an entirely related note, I'm tickled by the fact that NN were cracked by modeling minds and deepseek cracked the next step by referencing mentorship in its modeling. #educationAI

submitted 32 days ago • 0 comments

That the work is consistency checks is essentially the same problem every expert and company deals with to craft the best response to a company's day-to-day. We might have a better hint into how to better encode the output if we recognize, model, & automate the process of verifying

submitted 32 days ago • 0 comments

I'm a little hesitant to opt into Deepseek experiments for the time being. The research looks good but these policies are worrisome in the same way red note is worrisome.

submitted 33 days ago • 0 comments

For the laymen trying to get into developing AI or understand its application, this is an incredibly insightful video on building with AI. I also recommend the Anthropic blog post it references in the details. As for me, I'm definitely workflow AI > agent. www.youtube.com/watch?v=tx5O...

submitted 37 days ago • 0 comments

I just heard about "dirty sodas" and this just sounds like 7-11 swamp water and mocktail soda lounges I dreamed of as a kid.

submitted 41 days ago • 1 comment

“LLMs amplify existing security risks and introduce new ones” Even Microsoft sees it. New paper at arXiv, discussed briefly below. open.substack.com/pub/garymarc...

submitted 46 days ago • 4 comments

People too often forget that the legal protections are made by people and are as easily unmade and the first inconvenience www.axios.com/2025/01/10/m...

submitted 50 days ago • 0 comments

The conflict of interest is the worst part about this thread, but I highlight the public school issue because no AI system will stop a fight between students. No AI system will sense the issues with bullying and harassment and intervene. No AI system will set up the environment for learning.

submitted 53 days ago • 0 comments

arstechnica.com/information-... The bias against non-native speakers in chatGPT mirrors the way that ESL learners were systemically targeted when using tutors to help improve their writing. Elmesky's study on alleged "Cultural Mismatch" informed my thesis: doi.org/10.3102/0002...

submitted 54 days ago • 0 comments

Top comments for the post are 1) send articles that say ai detection doesn't work, 2) escalate to the dean, and 3) use ai detectors on unrelated original documents like the teacher's email. This is why the approach is so key: even though the teacher communicated the use of the tool, it's flawed

submitted 55 days ago • 1 comment

Before I forget, this is Alex Kotran's project Aiedu.org These guys have been on this for years starting with Machine Learning bias and and their lesson material is great.

submitted 55 days ago • 0 comments

Still, when I'm about to train teachers, questions should be addressed: How do you draw the line? What exactly is "ethical use"? What skills should we teach? And my fav. inspired by Alex Kotran: what jobs should we be training the next generation to do?

submitted 55 days ago • 0 comments

When it comes to teaching with AI, recognize that it's not AGI and it's certainly not infallible no matter the safe guards. It's better to proactively guide students if they're going to use it anyways. We are long past the age that hushing ourselves will make the problem go away.

submitted 55 days ago • 0 comments

I would have written this study for my MA if I had unlimited resources. It's worth a read if you're doing AI policy for schools

submitted 62 days ago • 1 comment

While appealing, the positive impact carries a lot of weight to explore: Teachers see schools to better students, even liberate them with critical knowledge. Yet students and businesses compare grades to measure *talent* (not even understanding) because who would be excused to blame their teacher?

submitted 63 days ago • 1 comment

So Bluesky apparently uses your device's system time to attach the time to posts. And as OP has demonstrated, this can be exploited.

submitted 106 days ago • 28 comments

AI is a threat to human teachers only if we believe the act of teaching is limited to delivering information and watching kids practice skills. If we value the art of teaching, we can find ways for AI to support and enrich it. #edusky #AIineducation brokenhand.substack.com/p/what-we-le...

submitted 66 days ago • 0 comments

In a world where we expect the most motivating educators to inspire learning in kids regardless of circumstance, what exactly in "agent AI" can cover for the emotional support demanded of teachers and mentors? Spoiler: the answer is that they cannot

submitted 63 days ago • 0 comments

LLM summarizing for outlines can be great for students and chunking tools are similarly helpful. Pre-prompted API feels more and more needed the more I look at the problem

submitted 63 days ago • 0 comments

For being a Comp Sci MA, this shows the framework everyone should follow: ✅ RAG leveraging class material ✅ Explicit the guardrails coded into the prompt Now we just need to add streamlined use. Coders really want LLMs to be a car, when in reality it's still a train. Build rails.

submitted 63 days ago • 1 comment

Gemini aggressively passing the buck for not being able to read my messages will never not be funny.

submitted 65 days ago • 0 comments

The parallels to the OCPL fiasco continue to intensify 🥲 docs.edtechhub.org/lib/9BRCRSSN

submitted 72 days ago • 1 comment

The examples are the most telling and useful: No AI: Skills cannot be evaluated while using GenAI (11) AI Planning: the decision skill being evaluated is separate from the AI inspiration (11-12) AI-Assisted Tasks: Using AI as a lab writeup tool b/c the lab is graded > independent writing 1/2

submitted 72 days ago • 1 comment