The future of generative AI, a great Chrome extension, and using AI to examine an exam


In this episode, Craig discusses:
- My vision of the future of generative AI
- Harpa - a great AI Chrome extension
- Using Claude to examine an exam
- Should higher ed fear AI?
The highlights of this newsletter are available as a podcast, which is also
called AI Goes to College. You can subscribe to the newsletter and the
podcast at https://www.aigoestocollege.com/. The newsletter is also
available on Substack: (https://aigoestocollege.substack.com/).
Welcome to AI goes to college, the podcast that helps
Craig:
higher education professionals navigate the changes brought on
Craig:
by generative AI. I'm your host, Dr. Craig Van Slyke.
Craig:
The podcast as a companion to the AI goes to college
Craig:
newsletter, you can sign up for the newsletter at AI goes to
Craig:
college.com/newsletter.
Craig:
In this episode, I talk about my view of the future of generative
Craig:
AI. Talk about how I use Claude to check an exam, which turned
Craig:
out pretty well actually. And I also talk about an interesting
Craig:
new AI extension for Chrome, one that I think has a lot of
Craig:
promise. So let's get started.
Craig:
So what's the future of generative AI? It's not been
Craig:
quite two years. So about a year and a half, really, since open
Craig:
AI released chat GPT. And we've come a long way since then. But
Craig:
I think it's time to start thinking about what the future
Craig:
of generative AI is going to look like. I know a lot of
Craig:
people are thinking about this, a lot of people are writing
Craig:
about this. So I'm gonna give you my view, the first thing
Craig:
that I think will happen is that AI will fade into the
Craig:
background, it will go from being this thing that we're
Craig:
talking a lot about, to something that just is it's just
Craig:
there, the chat bots are all well and good. They're very
Craig:
useful. They're very flexible, but they require too much
Craig:
expertise to use, especially for complex topics, or complex
Craig:
tasks.
Craig:
Sasha agrees with me, I don't know if you heard her there in
Craig:
the background. So I really think that what's going to
Craig:
happen is we're going to see these tools, I call them task
Craig:
based AI, that just do something they help you screen your
Craig:
emails, they help you refine your writing, they keep you on
Craig:
top of your task list, you know, whatever it might be.
Craig:
The GPT store that opened up not too long ago is a step in that
Craig:
direction. Custom GPT is can be really useful, especially when
Craig:
they're well designed. If you want to just have some fun,
Craig:
check out one called Cosmic dream, it creates some pretty
Craig:
trippy images, it's a lot of fun to play around with, I created a
Craig:
simple GPT to help with this newsletter, there really isn't
Craig:
very complicated. It just gives Chet GPT some context so that I
Craig:
don't have to enter that context. Every time I want to do
Craig:
something related to the newsletter. Ai enabled tools
Craig:
take this much, much further. For example, I use one called
Craig:
caste magic to create all sorts of material for my podcasts,
Craig:
Otter AI, which I think I've talked about before, is really
Craig:
awesome at transcribing audio summarizing, zooms. Summary
Craig:
feature is another example, Grammarly. The list goes on and
Craig:
on and on. And it's growing every day. So here's a good
Craig:
example. I think one that doesn't exist to my knowledge.
Craig:
What would it be like for those of you who are faculty, if there
Craig:
was a really good AI test generator, you record your
Craig:
lectures, give it access to your notes, or textbook chapters, set
Craig:
some parameters, and click on Go. So, before too long, you'll
Craig:
have an exam with really solid questions ready for printing or
Craig:
uploading to your Learning Management System, or whatever
Craig:
you do. I think those kinds of tools could free higher ed
Craig:
professionals, faculty and staff from a lot of drudgery. It's
Craig:
something I called Moving humans up the stack, which just means
Craig:
that AI can do the boring semi mindless work that so many of us
Craig:
hate freeing us up to do more engaging interesting parts of
Craig:
our job. Maybe you're different, but I hate developing and
Craig:
grading exams. So alright, let's move on. I think the AI enabled
Craig:
tools have almost endless possibilities. So that's step
Craig:
number one in the future of AI, or the first thing I want to
Craig:
mention. My next prediction is that we'll see a significant
Craig:
increase in the use of AI agents, an agent and an AI tool.
Craig:
They're really kind of hard to separate. An agent does
Craig:
something on your behalf. Siri and Alexa are kind of agents.
Craig:
But I'm talking about something different here. I'm talking
Craig:
about things that perform specific tasks as sets of
Craig:
workflows. And I realized I'm being a little bit confusing
Craig:
here because it is very, very messy. So for example, an AI
Craig:
agent can scan your email inbox and classify messages into
Craig:
different topics. Or maybe an AI agent might scan student
Craig:
admissions applications, and extract key data and
Craig:
automatically enter that into an admission system. I'd love a
Craig:
highly capable AI agent that could just grade homework or so
Craig:
alignments.
Craig:
So again, I know that the line between tools and agents is a
Craig:
little bit fuzzy. So let me give you some key characteristics of
Craig:
AI agents, AI agents mostly operate autonomously with
Craig:
limited human intervention. I think that's a big difference
Craig:
between what I was talking about with tools and agents. AI agents
Craig:
are task oriented and typically limited in scope. So they have
Craig:
one narrow thing that they do. AI agents are embedded in
Craig:
workflows. So it's generally not like a chat bot, where you do a
Craig:
bunch of things with it, it's just kind of there. And when
Craig:
it's its turn to do something it doesn't. AI agents are often
Craig:
embedded in application. So you may see it as part of a textbook
Craig:
publishers app or as part of an admission systems app. That sort
Craig:
of thing. Using AI agents doesn't require any real
Craig:
expertise in AI. No prompt engineering. In fact, you may
Craig:
not even know that you're using AI agents. And here's a big one
Craig:
that I'll talk more about later, AI agents can be combined. So
Craig:
these agents will be operating behind the scenes. As I've said
Craig:
a couple of times, you may not even be aware they're running.
Craig:
Although you may trigger the agents use or use its output
Craig:
without really realizing it. When these are all refined, I
Craig:
think these agents are going to be massively valuable. But
Craig:
building and refining them is going to take a lot of effort
Craig:
and considerable resources. That's one reason I see most of
Craig:
these coming from large vendors. All right, so let's talk about
Craig:
my third prediction agent to agent networks. This is where
Craig:
the real power of AI agents will come. From networks of agents
Craig:
working together just like organizations, networks of
Craig:
people can accomplish really great things working together
Craig:
more than they can separately, networks of AI agents will be
Craig:
capable of some amazing things. Imagine if you had an AI agent
Craig:
network that continuously scans your inbox sending messages off
Craig:
to other agents for further processing, your email
Craig:
processing agent recognizes an email from a student who needs
Craig:
advising. So it hands off the message to your advising agent
Craig:
who coordinates with the university's class scheduling
Craig:
agent to find out what relevant classes are being offered in
Craig:
future terms. Then the scheduling agent passes this
Craig:
information back to the advising agent, who figures out what the
Craig:
student should do and sends this information to your email
Craig:
composition agent who writes a response and sends it to the
Craig:
student, maybe after you review it. Except for the review. All
Craig:
of this happens without you doing a thing. That can be
Craig:
pretty sweet. Of course, there are risks, you're putting a lot
Craig:
of trust in the agents. Man, there are places where this
Craig:
could go pretty horribly wrong. But a smoothly working agent
Craig:
network could be a wonderful thing.
Craig:
Okay, there's more to this. But I think you've kind of got the
Craig:
point. So my vision of the future of generative AI as a
Craig:
shift from Ai chatbots to AI embedded tools, the rise of AI
Craig:
agents and networks of AI agents taking on complex workflows. Of
Craig:
course, I might be wrong about all of this AI is to say the
Craig:
least dynamic area. Nobody really knows what the AI
Craig:
landscape will be even a couple of years from now, maybe not
Craig:
even a couple of months from now. It's an exciting time, it's
Craig:
going to be an interesting future. And it has a lot of
Craig:
potential. Maybe to free many of us higher ed professionals from
Craig:
some dreaded, drudgery. least that's what I'm hoping to see.
Craig:
But time will tell.
Craig:
Normally, it takes me quite a bit of time to scan for
Craig:
duplicate questions in questions that are hard to understand. I
Craig:
was reasonably confident of the quality of the questions on this
Craig:
exam, but I needed to double check.
Craig:
I might pass this off to my GA but I really didn't want to do
Craig:
it in this case, because the turnaround time was so quick and
Craig:
that wasn't fair to him. So I decided to use Claude three Opus
Craig:
to do the checking. First thing I did was create a PDF version
Craig:
of the exam when uploaded to Claude. You know, there's some
Craig:
privacy concerns with this, but I wasn't really worried about
Craig:
Claude having my exam questions and there was no personal data
Craig:
involved. So I want to hit
Craig:
that I put in a pretty simple prompt. Please analyze the
Craig:
attached exam for two areas, unclear questions, and duplicate
Craig:
questions.
Craig:
And the newsletter which is available at AI goes to
Craig:
college.com I have a screenshot of clods response course I don't
Craig:
have the actual questions in there, I've kind of blurred
Craig:
that. It gave me a list of unclear questions, but Claude
Craig:
didn't really know some of the context that the students would
Craig:
have. So the questions really weren't, weren't bad. So I
Craig:
didn't make any modifications. I could have given Claude more
Craig:
context, but it really wasn't necessary in this case. So for
Craig:
example, Claude thought a question about the differences
Craig:
between requirements and goals and decision making was unclear.
Craig:
But I'd spent a lot of time on this in class, so I'm sure the
Craig:
students would understand the questions, at least the ones who
Craig:
were paying attention. It didn't identify any duplicate
Craig:
questions, which was good. So I decided since this had gone
Craig:
pretty well, I was going to take it another step. This time I
Craig:
asked Claude to check the exam against the study guide. I give
Craig:
extensive study guides and promise my students that have a
Craig:
concept isn't included in the study guide, it won't be on the
Craig:
exam. Now the questions aren't on the study guide. But the
Craig:
topics are. For example, the study guide might say compare
Craig:
and contrast requirements and goals in the context of decision
Craig:
making. But the exam includes multiple choice question on the
Craig:
differences between requirements and goals. Now, checking the
Craig:
exam against the study guide isn't hard, but it's kind of
Craig:
tedious. So I asked Claude to do it for me. I put in another
Craig:
another pretty simple prompt. Thanks, I'll wait. Thank you i
Craig:
ai. Please compare the exam to the attached study guide. Are
Craig:
there any questions on the exam that relate to topics not on the
Craig:
study guide? Note that the study guide does not contain specific
Craig:
questions for the exam. Rather, it covers topics that students
Craig:
should know. And Claude went through did its thing and raised
Craig:
a few potential mismatches. But after I reviewed these consider
Craig:
these concerning questions, I decided that the questions were
Craig:
fine clods concerns were reasonable. But really, human
Craig:
judgment was required to make the final decision about whether
Craig:
or not the questions recovered in the study guide. Since I was
Craig:
already pretty confident about the exam, I felt entirely
Craig:
comfortable using Claude to do the checking. But I if I were
Craig:
you, I would run a couple of experiments before fully
Craig:
trusting AI with this sort of a task. The big message here is
Craig:
that Claude made a couple of tedious tasks much more
Craig:
palatable. It also saved me about an hour, maybe an hour and
Craig:
a half of time, which was nice at the end of a long day. So if
Craig:
you have some tasks you find mundane and tedious, see if you
Craig:
can find a way to get AI to help free you from that drudgery.
Craig:
That seems to be the word of the day drudgery. But remember, when
Craig:
all of a sudden done, you're still responsible for your work.
Craig:
By the way, if you want some help in figuring out how AI can
Craig:
reduce some pain points, feel free to contact me, Craig at AEI
Craig:
goes to college.com. By the way, I love to talk about this stuff.
Craig:
So I'm happy to come to your campus to put on a workshop give
Craig:
a little talk, we can do something virtually doing this
Craig:
on a number of campuses in the next few months. So if you're
Craig:
interested again, send me an email Craig at AI goes to
Craig:
college.com. Alright, we're gonna wrap up with an
Craig:
interesting new tool called Harpa. That's H ARP a.ai. It's a
Craig:
Chrome extension that integrates AI into your web browser. Harper
Craig:
can use a bunch of different AI models, but the normal chat
Craig:
based interface seems to use GPT 3.5. But you can also connect it
Craig:
to Claude and Gemini. For me, harpist main use is to summarize
Craig:
webpages, but it has some other functions as well, maybe too
Craig:
many, it's a little overwhelming. Here are a few
Craig:
summarizing web pages, generating content, including
Craig:
email messages based on a webpage, monitoring websites for
Craig:
changes, extracting YouTube transcripts, that could be handy
Craig:
extracting data from websites. And there's a whole lot more.
Craig:
All of this goes from a Chrome browser window, which is pretty
Craig:
handy. So it's kind of tough to do here on a podcast. But if you
Craig:
go to the newsletter, which once again is available at AI goes to
Craig:
college.com/newsletter. I run through an example and it works
Craig:
pretty well. Right now there's a lifetime deal on Harpa. I think
Craig:
it's $240, which isn't bad. harpaz pricing structure is kind
Craig:
of odd and a little bit confusing. Most people from what
Craig:
I can tell would probably be fine with what they call their
Craig:
ex membership, or subscription that's their lifetime deal. Or
Craig:
their s one which is a monthly membership. You can go to
Craig:
harpa.ai and find out the pricing for yourself and decide
Craig:
what's what's good for you. All right, well that's it for this
Craig:
time. I hope you got something out of this episode. If there's
Craig:
anything you'd like for me to talk about, feel free to send me
Craig:
some suggestions. Craig at AI goes to college.com And
Craig:
remember, you can subscribe to the newsletter and I goes to
Craig:
college.com/newsletter. All right, I'll talk to you next
Craig:
time. Thank you. Thanks for listening to AI goes to college.
Craig:
If you found this episode useful, you'll love the AI goes
Craig:
to college newsletter. Each edition brings you useful tips,
Craig:
news and insights that you can use to help you figure out what
Craig:
in the world is going on with generative AI and how it's
Craig:
affecting higher ed. Just go to AI goes to college.com to sign
Craig:
up. I won't try to sell you anything and I won't spam you or
Craig:
share your information with anybody else. As an incentive
Craig:
for subscribing. I'll send you the Getting Started with
Craig:
generative AI guide. Even if you're an expert with AI, you'll
Craig:
find the guide useful for helping your less knowledgeable
Craig:
colleagues