@brandon_edu @OfficialLoganK

If you aren't using #DeepSeekR1 with #Cursor, what are you even doing? WOW.

@karpathy Earlier in 2024 I wrote about how RAG and fine-tuning are also versions of what educational researchers call "instructional design."

This x1,000,000. I'm looking at you, #open, #OER, #openpedagogy, etc. From @Ahmed_Lachheb et al.,

The US Copyright Office has released a new report on the copyrightability of works created using generative AI. For people working with #OER, these bullets from the Executive Summary will be of particular interest: - Questions of copyrightability and AI can be resolved pursuant to existing law, without the need for legislative change. - Copyright does not extend to purely AI-generated material, or material where there is insufficient human control over the expressive elements. - Based on the functioning of current generally available technology, prompts do not alone provide sufficient control. This reaffirms that the material we create with generative AI is NOT ELIGIBLE for copyright protection, meaning it is in the public domain. Meaning it is OER. h/t to @cgreen for the pointer to the new report.

@wasified It's part of the prompt that gets passed to the LLM. It just makes it easier for the LLM to tell what's what within the prompt.

RT @hud_zah: in a couple weeks, i built a nuclear fusor in my bedroom – with zero hardware experience the secret? Claude sonnet 3.5 + proj…

25 years ago, when I was a young firebrand, I didn't appreciate it either. But many long and deep conversations with Gary Lopez, who was an incredible mentor and way more generous and patient with me than I deserved, helped me finally understand. Regardless of whether your org is a non-profit or a for-profit, revenue earned from related services IS going to be a critical part of your #OER sustainability model. And if you're not willing to take that step - offering and charging for services - you're simply not going to be sustainable over the long-term. Full stop. Everyone who dedicates themselves to doing OER work full-time eventually learns this lesson - whether they want to or not. But people who work with OER as a small part of a different full-time job are never forced to learn this lesson, and have the luxury of criticizing from the sidelines. What should worry all of us is when people who haven't learned the lesson (either directly or vicariously) start driving departmental, institutional, system-wide, state-wide, or national policies and strategies. For all the good they're really, truly, and sincerely trying to do, they end up shooting all of us in the collective foot as you say. Sadly, I think last month's CCC announcement falls into this category. While the quote at the top talks about "making textbooks and other instructional materials affordable and accessible to all of our students," which is completely compatible with OER sustainability, the new policies are to include language that is VERY different: "developing and implementing degrees that don’t require students to pay for textbooks." This has become commonplace in the OER community - switching back and forth between "affordable" and "free" as if they're the same thing. They're NOT. Policies and strategies requiring "free" completely undermine the sustainability models of orgs who create and maintain OER, while policies and strategies requiring "affordable" create enough space for the orgs to survive and even thrive.

A new article out today ranks arXiv's most-cited #AI papers from January 1, 2023 - September 30, 2024. It's a greatest hits list that includes technical reports on open models like Qwen, Llama, Phi, and Mistral; new model architectures like Mamba, and fine-tuning techniques like DPO and QLoRA. If you're trying to stay up to date, check this list to see what you might have missed.

New essay - Where #OpenEducation Meets #GenerativeAI: OELMs. "Over the next several years course materials will likely shift from formats that look more like traditional textbooks toward formats that look more like large language models (LLMs) and other generative AI tools. The shift to these tools, which comes with the risks like those described above, threatens to erode important progress toward affordability, access, and equity made by the open education movement. Understanding that LLMs are course materials can help us think more clearly about what the future of course materials might look like and how and why open continues to be important going forward."

RT @mckaywrigley: Google Gemini 2.0 realtime AI is insane. Watch me turn it into a live code tutor just by sharing my screen and talking t…

RT @marktenenholtz: Great post. The value of chat models isn’t necessarily the initial questions but the ability to follow up infinitely.…

New research shows that when LLMs collaborate, they write significantly better feedback on student work. #genAI #education #AIinEducation

The 3rd UNESCO World OER Congress, titled "Digital Public Goods: Open Solutions and AI for Inclusive Access to Knowledge," is convening in Dubai next week. It may seem like a small thing, but I am deeply encouraged to see the convening title acknowledge that OER are public goods. (This is a soapbox I've been on since 2018.) People often talk about OER being part of a "knowledge commons," but this is unhelpful when it comes to thinking about the sustainability of OER. OER lack the most critical hallmarks of a knowledge commons - the overwhelming majority of OER are not collectively created or maintained, or governed by a community. (Can you name one, other than Wikipedia, that is?). If all that's required to be considered part of a knowledge commons is being openly licensed, and if the definition of OER is literally just "being openly licensed," then it's easy to see how thinking about OER as part of a knowledge commons doesn't provide us with any additional tools for thinking about their sustainability. However, OER fully meet the definition of public goods (i.e., OER are both non-excludable and non-rivalrous). And when you think of OER as public goods you actually get useful tools for thinking about their sustainability. For example, OER are subject to the free rider problem. That is, the reason we don't have all the OER we need is because people don't have financial incentives to create OER (since people generally don't pay for things they can get for free). This is a classic market failure, and OER overcome the free rider problem the same way other public goods do - through interventions by governments, philanthropy, and others who provide the funding necessary to make sure they get created. Thinking about OER as public goods will be an incredibly helpful step toward more productive conversations about their sustainability. I hope more people will follow UNESCO's lead here. #OER #sustainability #oercongress

Earlier this week @Intuit @QuickBooks closed my checking account without warning and without explanation and is holding my money captive. This is absolutely unacceptable behavior and must be corrected immediately. Please RT to help me get their attention.

RT @motherduck: We put a LLM in SQL and also show you the power of SLMs (small language models) in the MotherDuck data warehouse. https:…