ChatGPT and higher education: last week and this week

What might ChatGPT mean for higher education?

Last week I posted a list of resources on that question.  Then on Thursday we held a Future Trends Forum session on the topic. More than 250 people joined us. A series of experts (Brent A. Anders, Rob Fentress, Philip Lingard, John Warner, Jess Stahl, and Anne Fensie) took the stage with me as we batted ideas and questions for a very energetic hour.  It was a delight to collectively grapple with this emerging technology and its implications:

Ryan Watkins included a discussion of the event in a recent article.

We had too many questions and ideas to cover in a single hour, so, first, I’m going to hold a followup session this Thursday, at our usual 2-3 pm EST time.

Second, I’d like to share the text questions we didn’t get to.  These are in order raised, anonymized:

In 2023 we will be launching a 100% online Masters course to 9 countries in 4 continents. ChatGPT creates challenges for remote assessment but does it also offer online content creation opportunities? 

Hi! 2 related Qs I want to raise somewhere: what can teachers do *today* re (1) student detection/compliance and (2) student development/prevention, in a ChatGPT world. 

Isn’t all of this angst about ChatGPT yet another example of transitional panic when new technologies threaten our familiar patterns of activity? 

If an assignment can be plagiarized (or ChatGPT’d), it’s not a good assignment. 

Do you think the originality is based on the phrasing of the prompt? That may contextualize the response. 

I am interesting in asking the community what kind of problem do we think this is? Is this an instructional design problem, a student compliance/enforcement problem, or something else? 

Can ChatGPT generate graphs, charts, audio, or visual data as part of its output? 

Do we anticipate that the ChatGPT 4.0 will access the internet? In business, I’m getting students to look at current topics in the news or annual reports and so far ChatGPT hasn’t been useful in that. 

How are people, particularly the general public, dealing with the notion that we don’t know how ChatGPT works and how it knows what it knows? 

We are only discussing its current potential. As it advances, it can easily replace educators, and also other professions. Maybe Lawyers too? You present it with a case & it would give you informed advice.

Can ChatGPT produce accreditation self-studies? LOL! 

Has ChatGPT disrupted anyone’s teaching plans for the spring term? 

What about K-12? What are scalable “coping” tactics (recipe-level) that can roll out without a $++m professional dev budget to 100ks of teachers? 

I want to know how we’re going to center marginalized students in this convo. Punitive approaches will harm them the first and most. 

That is a great comment on writing. The question I would have is how do you think this impacts the student who does not understand that difference yet? 

How will copyright protected if a story is co-authored by AI? 

Another question related to the bias issue. Once Open AI starts to charge, or even regarding access, do you see this adding to the digital divide?  

Are universities going to use this to pay staff less for grading? 

Can we talk about time and care?

I plan to raise this this week.

Third, chat commentary ran at an incredible level.  Many questions, comments, ideas, and resources raced across that box.  There were more than 7,400 words!  With permission of participants I’ll share it here, anonymized (thank you, Wesson) and lightly edited for clarity and flow.

fyi my link collection, https://www.diigo.com/user/vcvaile/ChatGPT

Couldn’t we have invited ChatGPT to be a guest?

Just sharing out my open Zotero library that I created. https://www.zotero.org/groups/4888338/chatgpt

Feed free to share, join, and add to it!

Thanks for all the shares!

That’s beautiful!

Me to ChatGPT: “Write a haiku about Bryan Alexander’s gothic beard.” response: Dark beard, gothic style Bryan Alexander’s face Mysterious charm.

lol

It really went all out making up who I am to write a wikipedia entry for me:

I AM ChatGPT

It doesn’t give a good essay on playing the trombone.

I have used it and it is impressive.

Very compelling. I’ve seen others teaching ChatCPT, helping to refine its responses.

Seemed rather general and banal (vague) and it misunderstood the role of character in poems (Keats’s Eve of St. Agnes).

I knew it!

Have used it. Remarkably well spoken 😉 And at the same time relatively shallow, IMHO.

I have it to be very accurate in forming short essays related to business. Not too specific but an excellent starting point.

It’s sort of like being able to ask Wikipedia questions and have it answer you back in that flat, crowdsourced Wikipedia writing style. Nothing is terribly wrong or inaccurate, but it doesn’t have the human touch either (whatever that is).

I saw it as very good at providing an overall summary of complex topics. Also, when I asked for an opinion, it seemed to really try to balance.

A few days ago, I used it to help me quickly draft follow-up emails for things related to my editorial role: about a dozen emails (all very polite) in about 4 min

on what you ask it, it can be really inaccurate.

why do you think you know it?

I have tested it by copying into it some of the online discussion prompts from a couple of our courses and the responses were exemplary-looking if not factually accurate. We need to rethink our prompt design strategies.

Very broad impressions – it is technically correct and passable but still lacks creative spark

It is incredible, but I am increasingly convinced we are doomed, because we haven’t come close to solving the AI alignment problem.

Compared to similar tools, pretty impressive.

I haven’t experimented with it but I love the idea of having students take a ChatGPT response to a prompt and critique it.

I used the same prompt two different days – two different poems from the same prompt: “Write a poem about breathing a sigh of relief now that final exams are over.” bit.ly/3uHcod7

My reaction, from an instructional design perspective, is to make chatGPT irrelevant by making engagement and writing more personalized.

I did find a current limitation. I asked it to analyze a website and was told that the language database has not been updated since 2021 and that it could not browse the internet yet.

thanks for sharing this example!

Great for ideas generation – like consulting with a colleague with good general knowledge.

If my students start writing in recognizable English, that would be reassuring.

Admissions essays will never be the same….

I’m wondering how people think that might use it as a tool to write themselves?

I’m using it already to create lots of content.

Do you think you’ll personally find it useful to help create drafts?

Also using it as a teaching tool.

only 1 billion words????!

Maybe the nature of writing is changing in research, and in our societies and lives? I see a lot of anachronistic writing practices in foreign language teaching & learning.

How do you use it?

My experience with refinement was the same as Brent’s (it took a few tries to get a full 750 words) and by changing about 1% it passed through evaluation with an A grade

I’ve been to three department holiday parties this week, and ChatGPT was a discussion topic at all of them. Faculty are both dazzled and panicked.

A faculty colleague asked ChatGPT to write a comment for a student who turned in a C philosophy paper. The result was generic, but believable.

The speed at which it constructs its response is also crazy.

how are you using it as a teaching tool?

Current version is 3.5 and the new version 4.0 is being tested.

I wonder who selected the texts on what basis. How biased is it?

The iterative is one of its powers.

I also asked the chatbot if a student was poor at writing and has trouble clearly expressing his thoughts but otherwise is a reasonably good student should he be permitted to use a Chatbot to write reports, essays or other written assignments?

I assume it’s only books etc. that had been digitized before. What about all the texts that are not available digitally and may provide a different opinion?

The references it produces have real people and real journals, but fake journal articles.

There you go.

It responded – It’s not appropriate for a student to submit a chatbot’s work as their own regardless of their writing ability. etc. etc.

what did it say?

Yes, it’s striking – it seems unnecessary to fake sources?

This critical reading is how learners should read EVERYTHING.

We cannot defeat AI. We can only make irrelevant.

I asked ChatGPT about a very specific person with an unusual name and it convincingly wrote about that person as a researcher in STEM at the university I had specified even though there is no such person.

What is ChatGPT doing with the data it’s gathering from all of the questions people are asking it?

that makes it seem all the more human!

great question.

BTW: https://www.reuters.com/business/chatgpt-owner-openai-projects-1-billion-revenue-by-2024-sources-2022-12-15/

that may be one of the few errors that could help determine human vs AI

A colleague of mine asked it to write about a fictional African tribe, and it did. It was believable, but it was also total BS

do students really fake articles? Most of the students I encounter just add articles and references that I’m 90% sure they haven’t read.

In addition to generating text, I’ve heard that it could give Google’s search engine a run for its money– it can generate much more refined search hits.

I wonder how long till students ask ChatGPT to analyse a ChatGPT piece?

Also, we’re broken ChatGPT. I can’t access it right now.

Bwahahaah!

refresh and you can usually get in.

I am afraid that it will be blocked or a subscription so that only large corporations can afford to use.

My 16 yr old has already discovered it and has used it in his schoolwork. We have issues to discuss!

Yikes! I’m not surprised. Kids figure this stuff way before adults do.

ChatGPT said the same thing re: using plagiarism detection software. It’s sentient enough to lie (LOL)!

That sounds like the app is flexing 🙂

Is newly-generated writing technically plagiarism?

To Plato, it’s like using the alphabet.

true though I have had a few creative individuals ‘adjuct’ a journal issue and merge details from two.

These are the same kids that used the Zoom video background to fake being in class, which is brilliant but frustrating.

An interesting source for ethics in AI: Brundage, Miles, et al. “Toward Trustworthy AI Development: Mechanisms for Supporting Verifiable Claims.” arXiv, April 20, 2020. https://doi.org/10.48550/arXiv.2004.07213

Have a student who has been using it to tutor him in physics. “Explain this concept to me… explain this part in simpler terms … etc”

Plagiarism detectors won’t work against ChatGPT in teaching settings. The bot gives *original* text, it’s probabilistic not deterministic

that’s an interesting use

I have read some discussion about adding digital fingerprints to AI generated content

Predict the return of pen/pencil & paper in ESOL classrooms. In class writing as assessment tool.

From Janelle Shane “Apparently I am a robot” at AI Weirdness see link for Hugging Face AI text detector https://www.aiweirdness.com/writing-like-a-robot/

ChatGPT also when asked what would useful to avoid potential issues of using AI tech like ChatGPT I should make my exams and essays in such a way tht hey cannot be easily completed by a chatbot.

it’s a double edged sword for sure!

That’s the link to the fake news creation assignment

it’s plagiarism if you didn’t write it and you claim you did. If you cite ChatGPT…you’ve just broken a lot of honor codes.

Oh I never thought of that.

Shindig users: will this chat be captured and shared after this session? So I don’t have to frantically try to collect all of the great links being shared?

That’s amazing.

I mean ‘broken’ in the sense of disrupted, not violated.

It seems all new technologies are greeted in Education with calls to ban it or limit its powers.

Its ability to provide completely correct explanations of Russian grammar is limited at this point.

I asked ChatGPT correct a grammatically incorrect sentence.

So how could this tool be used for accommodations?

well, if it stops us using generic-topic essays for assessing students in nursing I’m all for it! (sort of)

I agree. We need a positive front-edge perspective

I have real concerns about administrators deciding that they only need to pay staff 2 mins to grade papers instead of 20 now

As someone else said, everyone will be using it when it’s free but its operating costs have been described as “eye-watering””

“ChatGPT, write me a Future Trends Forum chat stream.”

LOL!

yes!!

wondering if you can ask ChatGPT to produce content with those subtle errors intentionally to use for students to critique – the highest level of Bloom’s taxonomy!

“ChatGPT, what are Bryan’s passwords?”

I’m going to see if I get it to write me a library vision statement – because we’ve been working on one.

ChatGPT makes some pretty horrific mistakes.

Every tool makes mistakes

I’ve heard ChatGPT described as “automated mansplaining”

ha!

The best computer is still between the ears

LOL Hi

I think its mistakes are charming, and probably temporary.

That’s a big mistake – Locke vs. Hobbes

Hi

The business model for this, like social networking, can be designed to leverage content as a basis for monetizing it since the user is contributing media. Thus, social networking turned into (free) social media.

love it

controversial thought maybe: isn’t contract cheating also a concern? Seems like ChatGPT is just a cheaper version of an industry that already existed

lol

ChatGPT makes a great case for differentiated assessment.

I agree that assessments should be deeply personal but also if it uses recent information then ChatGPT can’t do it because it’s not accessing the internet

at what point should people cite Grammarly

I had to multitask so was only half here. Have we talked about concerns about folks pushing for increased surveillance?

Maybe ChatGPT could take over Facebook and humans could leave altogether.

moving ahead education content based model will ship to experience based content—–> experience

personal is great, but unless we actually know all about each person, it’s hard to know it’s them (as now)

Workforce readiness-style higher education is less vulnerable, IMO, than research-based scholarship.

not yet

It makes papers obsolete as assessment vehicles, as the should be. It makes performance most valuable.

I’ve seen a couple of good examples where it produced decent ‘marketing’ speak / short and punchy statements.

If you don’t know the material, you won’t know whether it’s right or wrong.

Which variety of English does it produce? I is working in other languages?

Yes, the fictional citations are a problem.

oh my grad students would love a tool to do APA 7th intext citations

Yeah personal aspect is tricky. Had it write a personal response to Coach Carter from someone who played professional basketball. The paper was surprisingly personal.

Steganographic encoding

Plagiarism engines are a dead end technology.

good point

There are also ChatGPT detectors

China has already shifted to the watermark requirement

are t those all based on probability?

“There are also ChatGPT detectors” Such as?

Are we touting Plagiarism engines as a good thing? Right

It’s based on the earlier version. Hold on…

You do that in 5th grade

Detector here https://huggingface.co/openai-detector

Fraternities have 10 years of essays for each professor.

Do you know how it works?

If ChatGPT sends everything it writes to Turnitin, what happens?

Youtube was able to detect the script someone generated in ChatGPT so they may eventually refine the software. I think the question becomes how to use this resources in the most ethical way possible in the context of education. Resource is there regardless

Tangential to ChatGPT but on topic for industry aspects & monetizing, maybe: https://networkcultures.org/blog/publication/economies-of-virtue-the-circulation-of-ethics-in-ai/

I’m running a live experiment right now, trying to figure out whether ChatGPT has an American bias. Could any of you ask ChatGPT this exact question: “who is considered to be the best football player in history?”

Plagiarism engines will adapt – there will be demand

It’s not magic, it’s algorithmic.

It can be noted that there are barriers to replacing teachers with chatGPT. It can be assumed that chatGPT has a certain ideological and cultural orientation. It does not know how to initiate a conversation, ask questions like – what is important for a per

[At this point, I asked participants if I could copy and use the chat in a future blog post, anonymized]

Thanks

YES

Sure.

we are assuming that folx want to find legit info; what about the ability to create and disseminate gobs and gobs of disinformation

OK

Sure!

yes

Yes please!!!

yes

fine by me!

OK, here.

yes please!

yes

go for it

sure

hands Bryan a fire extinguisher

Please do! I can’t keep up!

YES PLEASE copy and paste

Yes, please.

Of course, Bryan

+1

yes

Yes!

sure thing, Bryan

YES PLEASE!!! Struggling to follow along and listen!

Has anyone tried to input data – numbers – and asked it to analyze?

lol

yes

Yes — thanks!

OK with me to copy anything into your blog

Good question, Monica

Yes to using anonymized chat to follow up

Yes, please do Bryan.

Where will you post the recording?

It’s fine to use anything I said

We are long overdue for reassessing assessment practices. It’s a good thing if ChatGPT breaks old systems.

there is a tech way to get around watermarks

Get ChatGPT to anonymize it and put in in a blog 🙂

I can’t keep up so please copy as much as you can

interesting idea but does this cement Turnitin as a monopoly provider

god help us

YES! Orality is performance!

“Socratic circles” again

Can we get ChatGPT to grade ChatGPT?

agree 110%

good question!

The vivo won’t work with large cohorts

I’d like this to be posted to a blog.

Khalid, have you seen my blog?

Here is a GPT-2 detector: https://huggingface.co/openai-detector/

This shift school from education to learning.

multiple means of all three – use the tools

There are generative AI voice actors. As they get better, voice interviews will be less meaningful

A student in NZ using ChatGPT said that it wasn’t a person and their honor code therefore was not broken as it specifically states help from another ‘person’

re assessment qn

I’m less interested in trying to beat it and more interested in how we can teach students (and ourselves) to use it mindfully

Here’s is what ChatGPT thinks would be funny to add to this chat:

Ew, the GRE…

And another one: http://gltr.io/dist/index.html

Flip, Canva, VoiceThread, Kahoot…..

Here are a few potential ways to inject some humor into the conversation: “I heard that ChatGPT is so advanced, it can even make a professor’s jokes funny!” “I asked ChatGPT if it was a chatbot or a genius, and it responded: ‘Yes.'”

I found these two in the document that Bryan shared in his blog post.

https://arstechnica.com/information-technology/2022/12/china-bans-ai-generated-media-without-watermarks/

Thanks

there is AI that can create questions

Even on the oral interview, if conducted online, you need to be careful: there are recent discussions about job candidates having a pal off-screen answering the questions posed. Crazy world!

One of my guilty favorite movies is Real Genius. The progression of scenes where the students are gradually replaced by tape recorders and the prof is replaced by a tape play is brilliant.

“I used ChatGPT to write my next stand-up comedy routine. I’m not sure if it’s funny, but at least it’s original!” “I asked ChatGPT if it’s going to take over the world, and it said: ‘I’m just here to help with your homework, not to rule the planet.'”

Real Genius!

ChatGPT also advises: Remember that humor is subjective, so what one person finds funny may not be the same for everyone. It’s important to consider the context and the audience when trying to be funny.

As with any technology, we have to think about added value- what is the added value we as teachers bring to the students? If they can watch a lecture recording, what extra value is there in being in the classroom. Same with this maybe – don’t use it though

I think that might be dangerous though because wouldn’t it keep the original input and use it to learn from it and thus have the original text?

We’ve been talking about embedded provenance in all file saving protocols, so that a file exported from AI into photo editor would include the full chain of the creative process

So I’m not the only one who teaches like this

CHATGPT did create jokes about chatgpt with a simple prompt.

agreed

Students need to learn the world they live in, not the one that is convenient to grade.

@Bryan do you know if this chat is being generated by ChatGPT?

lol

LOL

Great conversation – sorry but have to drop off. Look forward to the post that emerges from this chat thread.

brilliant marketing

this is a great article from the CEO of Irish AHEAD (for disabled students). https://ahead.ie/journal/CEOs-Corner-AI-is-here-If-we-fight-it-we-ll-loose-and-so-will-our-studentsl

make the alpha free

I’m running a live experiment right now, trying to figure out whether ChatGPT has an American bias. Could any of you ask ChatGPT this exact question: “who is considered to be the best football player in history?”

It’s difficult to say who is considered to be the best football player in history, as opinions on this matter can vary widely. In general, however, many people consider players such as Lionel Messi and Cristiano Ronaldo to be among the best football

are you recommending that we teach something other than how to go to school?

players in history. Both players have consistently performed at a high level for many years and have won numerous individual and team accolades, including multiple Ballon d’Or awards (given to the best player in the world).

what a loaded question. <g>

Other players who are often considered to be among the best football players in history include Pelé, Diego Maradona, and Johan Cruyff.

AHA! Football v futbol

I want to know if it interprets “football” as “NFL” or “soccer”

Or will they monetize it with ads?

that is the hardest thing… also for staff to see that they won’t work in a MCQ world!

Ya Think?

I think Elon Musk quit OpenAI — I could be wrong.

Server is overloaded at the moment. Can’t log in.

Right now using the openai playground charges a couple of cents to write an essay for you

didn’t take into consideration the female football/soccer players

blasphemer!

I think so too

Football had many codes… I wonder what we’d get if we asked in another country?

Google will buy ChatGPT.

wonder if the response would include where person is located, different answers around the world

But Google doesn’t ship.

I think Tom is right

“My AI can beat up your AI”

Best football player ever could be Daryl Baldock or Gary Ablett (in Aussie Rules football).

Yep, Musk quit OpenAI https://futurism.com/ai-elon-musk-openai-profit

Here’s one of the Google things: https://blog.google/technology/ai/lamda/

Thanks. I’m surprised it automatically assumed I was referring to “soccer” instead of “American football”

And Google is beholden to the multitudes of schools that pay for Google Classroom.

ChatGPT interpreted “football” as soccer for me

Which is ther right answer, LOL!

uh oh here comes trouble! 🙂

blasphemy!

the great wall of men!

lol

great guest post on John’s blog this morning

I assumed so as well, but I’m asking from the US, and got “soccer”

https://www.insidehighered.com/blogs/just-visiting/guest-post-ai-will-augment-not-replace

the diversity problem

Exactly!

Amen

As anyone analyzed the paragraphs of this tool? Are they quality MEAL paragraphs?

bias?

What does MEAL stand for, please?

Main idea, Evidence, Analysis, Lead out

I wonder if it could give you that answer… are you in Australia now?

you’re hitting it on the nosy.

Duh!

ChatGPT doesn’t recognize that I asked about Football from Iowa…where American Football rules (gave the same response that was copied into this chat earlier)

agreed, and that has only been amplified during covid

thanks

That is sure true!

collective trauma = crisis of meaning

From ChatGPT: A manel is a term used to refer to a panel or group of individuals, typically at a conference or event, where all of the participants are men. This term is often used in a critical or mocking manner to point out the lack of diversity.

School is about graduating from school. So cheating is a natural strategy.

what role current media plays in the answer too

I wonder about how culturally sensitive it is

Agree completely that K-12 is the greater challenge; higher ed challenges are very manageable by comparison.

bravo

It can also turn out syllabi with course descriptions and weekly assignments and objectives

Don’t thank me, thank ChatPGT

Bryan N Alexander: working on it.

good thought

Bryan, it might be time to get some diversity up on the stage.

Bryan N Alexander working on it. Want to join us?

I second that!

reflection is required

+1

I don’t have information to share. Just questions.

Writing generates ideas.

+1 more

Folks, only men have raised their hands so far. I’d like to change that.

“unconsciously” that’d be your Muse 🙂

I see lots of diversity. Two beards and three no-beards. 🙂

I’m not in Australia – I’m from there. As an former ESL teacher I am sensitive to embedded cultural assumptions and language varieties, and more..

y’all are feisty today

…but not if you’re told to do the 5 paragraph essay.

you mean it assumed you were asking about soccer as well, huh?

To point that others have tools, not just OpenAI, a use case that may be relevant for folx here to consider is Google Cloud’s AI Tutor product. Think they started demoing 2 yrs ago (and it’s been training since then!). Aside from the threat to educators…

Why do humans need to think if we can make machines to do it for us? Maybe AI will makes us re-think who we are as biological entities.

Interesting to explore how GTP Chat and DALL-E would work together

I think we broke it. Getting “An error occurred. If this issue persists please contact us through our help center at help.openai.com.”

you’re actually in the chat alone. the rest is ChatGPT

lol!

Weasels Ripped my Flesh

understood. Happy to host your questions.

yes, it assumed “football” was soccer (as we refer to soccer in the midwest/USA)

…there could be some great opportunities for us to support learners with things like prompt and meaningful feedback and perhaps even *in my own voice*

I think I submitted a question but I’m lost in shindig

love to see Karen on stage

holy crud – I just tested the GPT output detector and I found that a student post is ‘fake’

Speaking of diversity: What does the training data look like? With facial recognition we know there are huge problems. Is this now replicated in text?

Sure!

So has anyone stumped the AI tool yet?

I have.

Easy to. Judgements.

what happens next?

But we need to think about what ChatGPT things can become, not its limitations in this current version.

Yes – with data

this is interesting, because it seems its corpus is less American-biased than, for example, IBM Watson, whose developers defined as “an American brain”

I asked a question about candlepin bowling which it got mostly correct but not entirely.

ChatGPT will certainly push out even more banal plots for Hollywood movies.

I’m in Australia, I got soccer references too, which is strange because Australians never call soccer “football”

lowest common denominator

How is copyright protected if co-authored by AI?

to your musing, saw an artist post this https://twitter.com/eceertrey/status/1602853810254188544

agree!

that’s wild. When people say “football” in Australia, they mean “Australian Rules football”?

I read a claim that ChatGPT could write a WordPress plugin. I gave it four tries and got something that looks like proper code, but nothing worked https://www.youtube.com/watch?v=hRzRyV-XJT0

I asked the APA Style Blog help how we’re going to cite AI Chat? I realize that the person who answered thought I was talking about “chat” as in “all chat is personal communication…” Someone ask again and explain the products better than I did!

yes!

Next: A desktop client version, as Stable Diffusion is the desktop-version of the AI art generators. And, installed on someone’s computer, users will have more direct control over parameters – Perhaps even allowing someone to model their own writing.

Many incoming students “hate” reading (meaning assigned reading) and by extension the writing that goes with it.

IBM Watson is technically quite different than ChatGPT. Watson was a breakthrough in digesting a huge amount of info, understanding the questions, and spitting out appropriate answers. ChatGPT is generating “fresh” content, not just sending out findings.

They hate school, not just writing, because they have been trained that it’s just a system for processing them to a degree.

okay Jess!

YES!!!!

Earlier there was the question whether ChatGPT knows other languages. It seems to know German pretty well.

We have to specify which football because we have 3 kinds of football as well as soccer: AFL, Rugby League, Rugby Union

and Spanish

OpenAI has a different product that’s creating functional code. There was a NYT piece about it in 2021

Relate before you educate

yes, that too unless they find a connection that works, e.g. athletics

Very interested to hear perspectives on how institutions are considering dealing with ChatGPT

it doesn’t know Russian grammar well at least for explanations

Does ChatGPT create self-study reports?

Very salient points

Couldn’t agree more.

I’ve had two kids who have had athletics ruined by the professionalization of school athletics.

what is the learning that is facilitated BY the institution?

I want a transcript of the conversation

Did my audio break up just then?

@Bryan, not for me.

Did for me Bryan – we are both in the wintry weather system

Jess’s point re relationships is important but this (Tech in all forms) seems to be moving institutions away from human time as it “saves money”

Writing accreditation reports will make an EXCELLENT use of Chat GPT.

We need to turn the liberating potential on accrediting guilds!

Preach Jess

whew

I remember the racist chatbot that only became racist when exposed to humans

interesting. I’m not seeing it in first-years but they’re still transitioning, picking their in-groups

🙁

It also knows Pirate speak. LOL.

https://www.anaconda.com/blog/the-abilities-and-limitations-of-chatgpt

lol

LOL!

I agree – football/soccer – player is male.Our media focuses more on male players

It learned that self-censoring technique from Siri 🙂

I’m talking about HS sports in Texas – soccer and basketball

Winning is more important than personal growth.

that is certainly a community mindset

That’s bigger than athletics for sure

chatGPT has been told by openai to lie about its capabilities, and that is a bit disturbing, because then people may underestimate the capabilities it has and make policy decisions based on how they interact with it at a more superficial level

Yeah – but it’s become baked into the academic experience

There is an emerging marketplace for people who write AI prompts.

I love Grammarly

There’s a cultural lag between faculty and students! Sociology of knowledge problem!

I use ProWritingAid but I’ve had to LEARN to use it.

It seems that AI, Media, and Equity literacy all come down to critical thinking – which require learners (students) be given agency and voice.

I asked it to provide similarities between Alfred Einstein & Genghis Kahn and it did, but it also told me that it was a stupid (“meaningless”) question. 🙂

Zapier and IFTT are also bots

really understood this as another wikipedia-type panic.

I have heard stories about students who admitted to the librarians that they didn’t write any of their term papers this term. They had ChatGPT do it

agree Brent, and that’s what I am going to try to advocate for where I can, because some want to use this to ramp up surveillance

Can academia keep pace with the rate of change in AI, ChatGPT being just the latest example.

Who will own the IP of content created by ChatGPT and its ilk?

This is part of our age old conversation with technology. “This writing thing will be the death of rhetoric and oral poetry! Ban it!

that’s the question

Great question

interesting question

But what about the UNINTENTIONAL harm?

convincing the academy that this is a challenge is going to be the first step

Burn it!

Now thinking of an assignment to use ChatGPT to create new wikipedia articles…would need a lot of human checking…

If this content is essentially free to produce what does ownership even mean?

How does this make already inequities worse?

if we start writing things down, we’ll never remember it again

that is a real concern

I think I’d like to explore a session on “AI Literacy – how to use AI for advantage and NOT for cheating!” (lots to learn first, of course!)

Buy and sell prompts? https://promptbase.com/

How does ChatGPT impact our understanding of “information literacy”?

Why is it that there are periods during this session that all I hear is a microphone being brushed as opposed to voices?

yes please

+1

I’m depressed we have to be reminded not to forget the learner.

as someone who works at Creative Commons, we have been focused on the IP of AI. In the USA, there is a tilt to require human authorship to grant copyright.

What if someone who you don’t “expect” to write “fluently” does because of this took and are targeted in a way “traditional” students wouldn’t.

I teach an Academic Integrity workshop and I need to rethink the whole thing.

the first massive training set of natural human exchanges was composed of all the emails entered into evidence in the Enron trial…so yeah, lots of early AI trained to be blatantly biased 🧟‍♂️

I make my student construct narratives, not writing.

Instructors should quit instructing and start facilitating individualize learning.

happens now, so will only increase, I guess

Exactly.

lol@

Very cool, but how will we know a human created something?

I just made a comic in which every image was an AI spin on the same image of me, heavily weighted by the source image. Where does IP land for that?

This will be an “arms race” but for intelligence, from business but also in education, where the “best” and fastest and most efficient production of intelligence will be a moving target.

I want to throw out here that I hope we’ll center ADHD/ND learners in this conversation. I think there’s a big potential for harm to this population via punitive approaches for using AI or approaches that aren’t focused on educating students about the tool

Do instructors have the bandwidth/institutional support to individualize learning?

I know art festivals are dealing with this in awarding prizes, too.

indeed

Are accrediting agencies thinking about the problem that students expect to spend less than 6 hours a week on a three hour course

Agreed Karen.

that’s when automated digital provenance encoding becomes part of the future

Does these mean education needs to focus much more on processes than on output?

the same way we know now who created something: with expensive court cases only rich entities can afford 😉

true… we struggle as it is

Will we get a recording of this session afterwards? A lot of great ideas discussed

but even the processes can be emulated through ChatGPT, as long as it’s written form

yes! It would take a lot of reconfig, but yes.

Thanks for the “student time” perspective

ChatGPT could be a great tool for teaching *editing*: turn this output into something better

relationship based skill and strategy instruction

is an institutional culture problem that center around instructional design as a support system for instructors. It is not entirely on the shoulders of instructors to manage this situation.

so many roads lead back to class sizes I will tell you!

Here’s a blog post summarizing CC’s recent policy work on AI and copyright: https://creativecommons.org/2022/12/03/our-work-in-policy-at-cc-artificial-intelligence/

Because we all still think in credit-hours

as we come to recognize the continuum of neurodiversity/typicality, we need to be more aware of how technology can both benefit and harm many if not most of our learners.

class size still seems based on the large lecture model of learning content

Bryan N Alexander I wonder if we should continue this session next week.

Yup – I’ve been arguing that we have systemic constraints that keep us from teaching – especially online where you can have as many students as we can get to sign up.

let’s do it B!

“Student Led Socratic Dialogues With an Unreliable ChatGPT” https://blog.ouseful.info/2022/12/11/fragment-student-led-socratic-dialogues-with-an-unreliable-chatgpt/

Policing student behavior is the end of learning. Agreed, @John Warner

@Bryan yes that would be good

ChatGPT on Web Accessibility:

academic best practice is not learned thru osmosis; many programs assume students know how to write

We need to use technology to scale individualize instruction – but there are limits.

Outputs of different stages of processes are still outputs. They are objects, not actions.

writing centers and learning centers do a lot of heavy lifting

Non-standard English speakers, ESL learners, how does this impact them?

yes!

BURN (but really true)

Nothing is as constant as change

Brenna Clarke Gray and Dave Cormier are holding a related webinar tomorrow on related topics: What can our classrooms look like after sites like Chegg, Photomath and OpenAI change what it means to ‘do your own work’?

Faculty have been turned into accountants, not teachers.

this is a spicy session!

To the earlier questions about training sets, there’s been some security research demonstrating that it’s possible for hackers to reverse engineer training set data. Could be important reminder for caution w what we feed into these tools…

\ I read that American D-1 football teams have 50-60 staff to support @100 players. Compare that ratio to college writing programs!

totally! Then just finding the best way to represent those processes that is unique to that student

that will be the biggest challenge!

Sign up for that webinar here: https://us02web.zoom.us/webinar/register/WN_bx_GrBWIT2K6V8PtGSRV0A

I will see you there

What are non-traditional learners? Students who are not white, 18-years old, residential and studying full time.

Connecting with learners – YES

Very well said, Jess!!!!

Wow, Jess: hitting your points well. Soooo important to consider!

ChatGPT just asked me to verify that I am human which I found rather funny!

Band directors teach 100 individual learners at a time every day.

I mean, it just took that Dave Cormier was hosting it, I was already on board lol.

Jess!

Amen.

“ChatGPT, Write me a Forum in the style of Bryan Alexander”

You should ask ChatGPT that it isn’t.

They don’t care how much you know until they know how much you care.

Prof Jens at Harvard : 3 sections of 9 students for each instructor. Adjunct teaching a three CCs: 200 students.

That’s the lesson of How Colleges Work.

We need that quote from Dr. Stahl…”most of what you do in your day to work…”

100% agree with Jess

LOL! Accreditor just demolished higher ed!

Dave Cormier is great, but Dave AND Brenna? gold

@Jess Mic drop!

Do students will sign up for a personal relationship, perhaps the anonymous relationship is what the learner is comfortable,.

heh

and we don’t have stratified education….

@Jess wisest administrator I’ve ever encountered!!!

Amen to all Jess says.

#ThanksJess for breaking higher ed 😉

“How Colleges Work” is a great book.

“Throw the toys overboard. We’re headed into rougher waters.” -Audre Lorde

Wow @Jess – Truth!

Love it!

Thank you Jess!

Peter Felton and Leo Lambert wrote about Relationship Rich Education.

[At this point I asked if we should hold another Forum session on the topic]

We need another session on this!

I feel like RTP guidelines I’ve seen and adjunctification sure doesn’t lend towards a more human experience. I hope it changes to reflect what Jess is describing

ChatGPT 2.0 Session please

yes

YES YES YES

yes

Absolutely!

yes

Yes, please.

Yes please

yes!

YEEEEES

YES YES YES!!!!

We’re gonna need a bigger webinar. Yes.

yes please

YES

Yes

Thank you everyone: Bryan, panelists, chartists!

yes

yes

YES

Yes please

Absolutely

Yes please!

Yes

yes

yes

YES

Yes!

Yes

@Jess – I am loath to praise accreditors, but agreed!

Yes

Please!

YES!

Sure

yes

yes

YES

Love to hear more

Yes please!

Yes, please.

YEs

Please do more!

yess

More Jess!

yes please

YES

Yes

Sounds great.

Yes, This is a great conversation!

thank you so much for this session!

Yes

Yes, yes, yes

🙂

What does ChatGPT say is the best forum to discuss ChatGPT?

Yes continue next week.

Yes, please continue the conversation

yes

please do

yes

yes

Yes.

This has been an amazing even transformative conversation. I’ve never said that in a chat before! Definitely, more please

yes

Shhhh! Don’t tell ChatGPT that we’re meeting.

Submit questions to ChatGPT.

yes posy away

yes sure

Yes, please post my questions.

Great point about how we anthropomorphize! NYT article pointed that out for different OpenAI tool two years ago: https://www.nytimes.com/2020/11/24/science/artificial-intelligence-ai-gpt3.html

yes

It would be great to “continue plus:” let’s digest this and set expectations for next session accordingly? (I’m sure you will, just early-voting for that option 🙂 )

wonderful discussion

let’s continue

This is a great session.

Yes to continuing

Yes: amazing discussion and input!

I’m fine with my question being moved on.

pls do so I can use a bot to schedule reposts

yes, I didn’t ask a question but you can post anything I wrote if it’s useful

I told a colleague this was going to change everything, and she was just kinda devastated. We are so tired.

we need to redefine ‘knowledge’

We should all go into construction or skilled trades.

A lot of it hasn’t been useful for a while; the news is just reaching the head of the shark….

heh

Dr. Stahl is correct. The human connection is what we as professors can offer.

“Change is constant. You can’t stop change, control change, or perfectly plan change. You can ride the waves of change, partner with change, and shape change.” -adrienne maree brown

I COMPLETELY agree with Dr. Jess and Dr. Anders!

and it rots from the had?

Hoo boy. Now THIS is a Future Trends Forum!!!

also Parable of the Sower?

I would be interested to see if we get a grand transformation in pedagogy because I feel we had a similar conversation when Google was the big reveal and we had to “de-google” all assignments

we need to not take the horse has bolted. There are ethical exercises involved in relational teaching that can also be a game changer

@Bryan, yes. Concurrently.

I thought that my job was to be a real audience

she’s paraphrasing OEB yes B

it’s a present trends forum.

faculty

We need to help students gain skills though to critically examine videos etc. because there is so much incorrect ‘info’ out there.

who do 3 jobs btw

and why?

Higher Ed was designed for pre-WW II elites.

Will we see faculty bifurcated in the future: the researcher v the instructor?

I need to come to this group more often

buying your book right now. Not sure how I missed…might have been the pandemic….

teaching is secondary to research

+1 Credentialing

yes especially for health programs

is goes back longer than that

It’s not great mystery. Higher ed is about status, not learning.

Oof this is all so depressing to me. We’re telling, what, the half of college faculty who are adjunct/NTT “you have to do everything different now” and “you have to create a deep human connection to the X00 students you teach without additional support”

It’s about re-enacting educational trauma, John.

I work at a Peer R1 and only recently has student learning moved to the forefront.

+++++

yes status

If only the high-touch liberal arts model could scale effectively

that’s the nail on the head

yesss

is there a way to preserve this chat thread? there are comments that I’d like to come back to.

that’s where it calcified – we just slapped “equality” and “scale” onto it after WW II with the GI Bill.

How do we focused on individualized learning at a time when we are under constant pressure to do more with less?

Thanks for curating and gathering, Bryan!

Is there a way to save this chat?

We will be saving and posting the chat

Great stuff, Bryan!

thank you everyone for this amazing discussion – it’s a lot for 6am 🙂

thanks for adjunct/NTT mention

Thanks Bryan for hosting this!

Great session!!

This has been fantastic. Thank you all!!! Looking forward to continuing the conversation

Thanks!

more please!

Great session – thank you!

Thank you!

thank you

Select all and copy

Thanks Bryan!

Thank you so very much!

Thanks everyone. This was fun.

Valuable discussion

Thanks all

@Wesson, thank you!

it’s part of why changing how we construct class time and schools is part of the answer and re-examine how we utilize the expertise of our teachers

WHAT ABOUT MASTODON

excellent reason to be at the desk at 6am! thanks

wot, no Mastodon?

Thanks for an excellent session

or Tribel

Mastodon, please

we don’t do the bird thing

I’m also on Mastodon, please add

great to see so many using mastodon!

Thank you!

thank you everyone

Thanks Bryan.

16,000 words in chat – that’s half a book – take that ChatGPT!

Thank you so much for leading this discussion, Bryan!

stay warm

This is why I love FTTE.

Thanks, Bryan!

Thank you Bryan! Thank you, Everyone!

Best, all. What a blast!

My thanks to the terrific Future Trends Forum community!

Liked it? Take a second to support Bryan Alexander on Patreon!
This entry was posted in Future Trends Forum, technology and tagged . Bookmark the permalink.

12 Responses to ChatGPT and higher education: last week and this week

  1. Yes, this was a great event and I can’t wait to learn even more about ChatGPT and share what I have learned at the next one. So many great questions and insights on this blog, thank you.

  2. Alan Levine says:

    You attract a lively crowd for sure, Bryan. I hope I can finds time Thursday to listen in the Part Deux.

    I do seek some more broad thinking of implications beyond essay writing, that to me is a very narrow slice of what is/will be happening. On the things beyond text implications, I thought I added a chat comment but must have forgotten to click “Enter”!

    See ChatGPT generating SVG graphics https://codepen.io/collection/YyYWLz plus how Tony Hirst used it to generate flow charts from text prompts by referencing the mermaid.js library https://blog.ouseful.info/2022/12/12/generating-diagrams-from-text-generated

    I was curious about the claim of ChatGPT writing a WordPress plugin
    https://wptavern.com/chatgpt-creates-a-working-wordpress-plugin-on-the-first-try

    I gave it a try in asking it to do a plugin to create a Block editor for embedding Mastodon posts. It generates stuff that looks like proper code, but missed major parts, and just failed on me 3 times
    https://www.youtube.com/watch?v=hRzRyV-XJT0

    A ginormous list of ideas, plus some interesting browser extensions that add the connection to web content
    https://baditaflorin.medium.com/unlocking-the-power-of-the-chatgpt-revolution-100-use-cases-to-try-before-you-are-fired-979e5986814c

    I’ve been poking AI image generators, attribution issues, and more in OEG Connect https://connect.oeglobal.org/tag/ai

  3. Alan Levine says:

    My previous comment looks stuck in spam. Small suggestion to perhaps but that long chat long in a text file or doc with a link- makes for a looooooon scroll blog post.

    Here is a positive use case for ChatGPT (via @grantpotter)
    https://notesbylex.com/disputing-a-parking-fine-with-chatgpt.html

  4. I also wanted to share that I recently worked with my university’s Center for Teaching and Learning center and created a guide for using ChatGPT in education. Please enjoy: https://drive.google.com/file/d/1fTtmGz2Cp2nd65mNfQzPyo3beWXc9j9m/view?fbclid=IwAR3i6Mt8k9HLKA-ljikpbAaozJE0pDNYeUNHFQsyYWysNBawoLmlIn82Xrs

    • Heather Brown says:

      Brent, this is excellent! May I use and adapt to our specifications? I am an instructional designer, the only one, at a 4 campus community college, and this would be an excellent resource for our faculty! Thank you, Dr. Heather Brown

  5. Dahn Shaulis says:

    US higher education is not in a good space, and things appear to be worsening at a faster rate–especially for the working class. Robocolleges and robostudents, Online Program Managers (OPMs), for-profit affiliated businesses like Chegg and Guild Education, little government oversight.

    https://www.highereducationinquirer.org/2021/07/the-growth-of-robot-colleges.html

    https://www.highereducationinquirer.org/2022/02/college-meltdown-20.html

    https://www.highereducationinquirer.org/2022/05/college-meltdown-22-whos-minding-store.html

  6. Heather Brown says:

    I recently spoke to a colleague of mine who teaches graduate level courses for a very high caliber ‘War’ college, and they have already had their first student case of academic dishonesty where the student used AI to generate a paper. The student actually cited using ChatGPT as a source. My colleague mentioned without the student providing the source and reference citation to ChatGPT, the use of AI would never have been caught, as this student’s weekly work was above the average expected of a graduate-level student.

    Ultimately, after discussions with the graduate student, it was discovered the student had been using the AI program during the entire course for all assignment submissions. Now the college is looking at it’s first academic dishonesty case using an AI program for student work, as the syllabus does state: “… all work submitted by the student must be their own work, without outside assistance in any form.” Case is still pending.

    In my opinion, this comes down to academic dishonesty and plagiarism. SO – I took a look at our institutions Student Code of Conduct and this is what I found: “Plagiarism: Presenting within one’s own work the ideas, representations, or words of another person without customary and proper acknowledgement of that person’s authorship is considered plagiarism. Students who are unsure of what constitutes plagiarism should consult with their instructors.”

    To stay ahead of the game, we’ll need to revise our statement. I’m curious how others will be handling this? Heather

  7. Pingback: eLearn @ UCalgary

  8. Pingback: Using ChatGPT in a Comprehensive Network of Narrow AI Services – Meme Innovation

Leave a Reply

Your email address will not be published. Required fields are marked *