Marc Watkins
@marcwatkins.bsky.social
Assistant Director of Innovation at the University of Mississippi training faculty in AI literacy. Teaching, #OER, #OpenPedagogy, #AIED marcwatkins.org
Beyond the many moral and ethical objections I raised about using AI to grade, here's a concrete one: AI grading will never be secure because of prompt injection. Simply hiding a phrase within an assignment is all it takes to bypass even the most advanced AI models.
October 12, 2025 at 3:16 PM
Beyond the many moral and ethical objections I raised about using AI to grade, here's a concrete one: AI grading will never be secure because of prompt injection. Simply hiding a phrase within an assignment is all it takes to bypass even the most advanced AI models.
Some teachers are using AI to grade and few are talking about it. Scantrons are one thing, but automating grading, feedback, and assessing written work with AI poses massive ethical challenges and should not become the default or go to choice in education.
👇 in comments
👇 in comments
October 10, 2025 at 11:42 AM
Some teachers are using AI to grade and few are talking about it. Scantrons are one thing, but automating grading, feedback, and assessing written work with AI poses massive ethical challenges and should not become the default or go to choice in education.
👇 in comments
👇 in comments
I guess I won’t be posting here much longer. Being in a state that’s instituted age verification laws causing social media companies like Bluesky to restrict access isn’t an isolated event—it’s where the internet is headed. I’ll use a VPN for now, but that’s not going to be sustainable.
August 22, 2025 at 9:51 PM
I guess I won’t be posting here much longer. Being in a state that’s instituted age verification laws causing social media companies like Bluesky to restrict access isn’t an isolated event—it’s where the internet is headed. I’ll use a VPN for now, but that’s not going to be sustainable.
Faculty guidelines and frameworks for using generative AI with their students are often unclear and vague. The institutional silence around faculty AI usage risks sending a message of indifference instead of charting a clear path forward. I created a VALUES framework for faculty to consider.
August 22, 2025 at 12:55 PM
Faculty guidelines and frameworks for using generative AI with their students are often unclear and vague. The institutional silence around faculty AI usage risks sending a message of indifference instead of charting a clear path forward. I created a VALUES framework for faculty to consider.
Another semester approaches and faculty face a super charged version of ChatGPT. I want to talk about value and meaning with my students, not necessarily AI. Here are some ideas to do just that. I also have a session submitted to SXSW Edu about AI aware teaching that I’d love for you to vote for. 👇
August 10, 2025 at 1:18 PM
Another semester approaches and faculty face a super charged version of ChatGPT. I want to talk about value and meaning with my students, not necessarily AI. Here are some ideas to do just that. I also have a session submitted to SXSW Edu about AI aware teaching that I’d love for you to vote for. 👇
Will Microsoft update the education version of Copilot with GPT-5? This was formerly known as Bing Chat Enterprise and hasn't been updated with a new model since GPT-4 launched in 2023. That would give data protection w/ GPT-5 to a lot of schools.
August 9, 2025 at 1:58 PM
Will Microsoft update the education version of Copilot with GPT-5? This was formerly known as Bing Chat Enterprise and hasn't been updated with a new model since GPT-4 launched in 2023. That would give data protection w/ GPT-5 to a lot of schools.
I stand corrected. They simply moved it to the "+" feature. I missed it under the . . . more section
August 7, 2025 at 10:32 PM
I stand corrected. They simply moved it to the "+" feature. I missed it under the . . . more section
RIP ChatGPT's Study Mode. It lasted ten days as a feature before GPT-5's launch removed it. Now students will need to ask the model to not complete tasks for them manually. The point was a ⏹️ button students could use to create a socratic tutor, a space where the model wouldn't offload learning.
August 7, 2025 at 7:17 PM
RIP ChatGPT's Study Mode. It lasted ten days as a feature before GPT-5's launch removed it. Now students will need to ask the model to not complete tasks for them manually. The point was a ⏹️ button students could use to create a socratic tutor, a space where the model wouldn't offload learning.
But when you attach a file and ask it very specific questions, it does a fairly decent job of grounding its responses in the data. Here's a summary of the Walton Foundation report on AI saving teachers six hours a week. It took several minutes to run this so I'm not sure how practical that is.
4/4
4/4
August 7, 2025 at 3:15 AM
But when you attach a file and ask it very specific questions, it does a fairly decent job of grounding its responses in the data. Here's a summary of the Walton Foundation report on AI saving teachers six hours a week. It took several minutes to run this so I'm not sure how practical that is.
4/4
4/4
Without search, the model just makes up material. Wildly so. Asking for a list of publications from me and none of these are real. We're back to the good old days of early GPT-3 and ChatGPT.
3/4
3/4
August 7, 2025 at 3:15 AM
Without search, the model just makes up material. Wildly so. Asking for a list of publications from me and none of these are real. We're back to the good old days of early GPT-3 and ChatGPT.
3/4
3/4
I cannot overstate how bad hallucinations are without RAG. Asking the model for the current AI outlook on my campus produced no factual information, which isn't surprising given the model's size.
2/4
2/4
August 7, 2025 at 3:15 AM
I cannot overstate how bad hallucinations are without RAG. Asking the model for the current AI outlook on my campus produced no factual information, which isn't surprising given the model's size.
2/4
2/4
Giving oos 20b a second look with LM Studio, which is much better at running it. You can toggle the reasoning settings and this does improve the output. Medium setting seems to be the sweet spot, but on high you have to raise the context limit to get the best output to count Rs in strawberry.
1/4
1/4
August 7, 2025 at 3:15 AM
Giving oos 20b a second look with LM Studio, which is much better at running it. You can toggle the reasoning settings and this does improve the output. Medium setting seems to be the sweet spot, but on high you have to raise the context limit to get the best output to count Rs in strawberry.
1/4
1/4
3️⃣ Fail: I asked it to code a game on climate change. The model began searching the web then got stuck in an endless cycle of web search and reasoning. We never got to tools to code or refusal, just a loop in search. After 20+ minutes I ended the chat without a result with a much lower battery.
3/3
3/3
August 6, 2025 at 2:57 AM
3️⃣ Fail: I asked it to code a game on climate change. The model began searching the web then got stuck in an endless cycle of web search and reasoning. We never got to tools to code or refusal, just a loop in search. After 20+ minutes I ended the chat without a result with a much lower battery.
3/3
3/3
2️⃣ Partial success: I used search and asked about the current situation in Gaza. Multiple errors. The model took 8 minutes to come up with a response and gave me a single sentence response about death toll, not a summary of the overall situation:
2/3
2/3
August 6, 2025 at 2:57 AM
2️⃣ Partial success: I used search and asked about the current situation in Gaza. Multiple errors. The model took 8 minutes to come up with a response and gave me a single sentence response about death toll, not a summary of the overall situation:
2/3
2/3
I have a Macbook Pro/ M4 48 gbs and ran GPT-oss 20b via ollama and these were the results on three tests for reasoning and search. My overall impression is the usefulness of the model is very limited on a personal device.
1️⃣ Fail: the strawberry test
1/3
1️⃣ Fail: the strawberry test
1/3
August 6, 2025 at 2:57 AM
I have a Macbook Pro/ M4 48 gbs and ran GPT-oss 20b via ollama and these were the results on three tests for reasoning and search. My overall impression is the usefulness of the model is very limited on a personal device.
1️⃣ Fail: the strawberry test
1/3
1️⃣ Fail: the strawberry test
1/3
The new GPT-oss open weights models are technically impressive, but are they practical? Both suffer from higher hallucination rates compared to proprietary models, don't ground results to improve errors through search, and you need 80 gigs of memory on your device to run the largest model locally.
August 5, 2025 at 8:09 PM
The new GPT-oss open weights models are technically impressive, but are they practical? Both suffer from higher hallucination rates compared to proprietary models, don't ground results to improve errors through search, and you need 80 gigs of memory on your device to run the largest model locally.
The new “Study Mode” for ChatGPT turns it into a Socratic tutor and is just a prompt. The implication is OpenAI could have released this years ago—a version of ChatGPT that wouldn’t complete your homework for you.
Let that sink in. OAI could have nerfed ChatGPT with a few well organized phrases.
Let that sink in. OAI could have nerfed ChatGPT with a few well organized phrases.
July 30, 2025 at 3:41 AM
The new “Study Mode” for ChatGPT turns it into a Socratic tutor and is just a prompt. The implication is OpenAI could have released this years ago—a version of ChatGPT that wouldn’t complete your homework for you.
Let that sink in. OAI could have nerfed ChatGPT with a few well organized phrases.
Let that sink in. OAI could have nerfed ChatGPT with a few well organized phrases.
As the fall semester approaches, we’re bound to hear more talk about AI policies, what new skills students will need, and how to curb AI misuse. We’re also likely to hear renewed calls for faculty to redesign assessments. A year ago I wrote a series about how AI impacted learning and revisited it.
July 29, 2025 at 1:37 PM
As the fall semester approaches, we’re bound to hear more talk about AI policies, what new skills students will need, and how to curb AI misuse. We’re also likely to hear renewed calls for faculty to redesign assessments. A year ago I wrote a series about how AI impacted learning and revisited it.
Google Labs has an AI detector they are rolling out called SynthID. Big caveat that it only works with Google generated text, video, audio, etc. It makes you wonder why all the other AI platforms haven't adopted watermarking. It's almost like Google is trying to send a message 🤏
May 20, 2025 at 8:36 PM
Google Labs has an AI detector they are rolling out called SynthID. Big caveat that it only works with Google generated text, video, audio, etc. It makes you wonder why all the other AI platforms haven't adopted watermarking. It's almost like Google is trying to send a message 🤏
The NYTs piece about faculty using AI undisclosed with students fails to connect this troubling behavior with the exploitation of contingent faculty. People inevitably seek shortcuts when their working conditions are untenable. GenAI didn't create that problem.
May 14, 2025 at 5:09 PM
The NYTs piece about faculty using AI undisclosed with students fails to connect this troubling behavior with the exploitation of contingent faculty. People inevitably seek shortcuts when their working conditions are untenable. GenAI didn't create that problem.
Latest dispatch from me. Students now have free access to premium AI through Google Gemini, ChatGPT Plus, and xAI's Super Grok and were not talking about it
Burnett:[E]veryone seems intent on pretending that the most significant revolution in the world of thought in the past century isn’t happening
Burnett:[E]veryone seems intent on pretending that the most significant revolution in the world of thought in the past century isn’t happening
April 27, 2025 at 1:46 PM
Latest dispatch from me. Students now have free access to premium AI through Google Gemini, ChatGPT Plus, and xAI's Super Grok and were not talking about it
Burnett:[E]veryone seems intent on pretending that the most significant revolution in the world of thought in the past century isn’t happening
Burnett:[E]veryone seems intent on pretending that the most significant revolution in the world of thought in the past century isn’t happening
Google's AI Studio released an interesting and I think useful feature called apps that anyone can access for free using an Gmail account. Apps allow users to build and use the full multimodal features found in Gemini. Some of these playable demos are really impressive:
April 25, 2025 at 12:26 AM
Google's AI Studio released an interesting and I think useful feature called apps that anyone can access for free using an Gmail account. Apps allow users to build and use the full multimodal features found in Gemini. Some of these playable demos are really impressive:
The war AI firms wage over students is intensifying. Anthropic and OpenAI offer free versions for a few months of access to their AI tools, but Google has stolen the show. Google is giving away a year of free access to Google One AI Premium to all college students. $240 worth of access for free.
April 19, 2025 at 2:08 AM
The war AI firms wage over students is intensifying. Anthropic and OpenAI offer free versions for a few months of access to their AI tools, but Google has stolen the show. Google is giving away a year of free access to Google One AI Premium to all college students. $240 worth of access for free.
I will not be speaking at the Digital Universities US conference. The Times Higher Education Events staff rescheduled the conference less than a month out and expected speakers they solicited to foot the bill and scramble to cancel hotel rooms and rebook flights. Extraordinarily unprofessional!
April 10, 2025 at 3:06 PM
I will not be speaking at the Digital Universities US conference. The Times Higher Education Events staff rescheduled the conference less than a month out and expected speakers they solicited to foot the bill and scramble to cancel hotel rooms and rebook flights. Extraordinarily unprofessional!