00:00:03
hello this is Jim spor with isip and the
00:00:06
previous recording I made I forgot to
00:00:09
share my screen so this time I'm going
00:00:12
to share my screen and I'm going to talk
00:00:15
about two things very very quickly um
00:00:19
and uh I'll upload these documents and
00:00:22
some links and stuff uh as well but the
00:00:25
first thing I want to talk about is the
00:00:28
uh white paper uh
00:00:30
for
00:00:32
um the uh for the course um
00:00:38
and uh also the AI tutor okay so here is
00:00:42
the uh AI driven is of course for
00:00:45
onboarding participants you can use AI
00:00:48
to help generate this paper um I uh you
00:00:53
know there's the authors the date how to
00:00:55
site it you know need an abstract
00:00:58
there's an AI generator abstract by the
00:01:00
way um what I do is I outline the paper
00:01:04
they all all white papers have the same
00:01:06
structure there's an introduction in the
00:01:08
problem you're going to solve and I just
00:01:09
give instructions to the AI like
00:01:11
describe this describe that describe
00:01:13
this so section by section I ask
00:01:16
different AIS to to build this up and um
00:01:19
again as always when I'm starting stuff
00:01:22
I try to have a um a a browser window
00:01:27
with like you know a chat GP PT and
00:01:30
Claude and Gemini and U Microsoft Bing
00:01:35
uh are all um all open the four that I
00:01:40
use and actually there's another one
00:01:42
that I think is quite good called duck
00:01:45
uh duck duck uh AI chat um that is
00:01:50
relatively new I'm just starting to
00:01:52
explore it but you can see um it's got
00:01:55
different models so you can choose among
00:01:58
some open- Source models uh mistl is
00:02:00
open source and llama of course from
00:02:02
meta's open source and then there's mini
00:02:05
models from some others that are open
00:02:07
source so they they do uh you know a
00:02:10
model selection kind of thing so so
00:02:12
maybe I'll add duck AI so there'll be
00:02:14
like five AI tools anyway um it's one of
00:02:18
the deliverables for the isip course is
00:02:20
a white paper so I want to help you get
00:02:24
started on that you shouldn't wait till
00:02:25
the last minute to do the white paper so
00:02:28
I'm going to upload this and you can use
00:02:30
this as a starting point um you can see
00:02:33
it's got the outline what sections it
00:02:36
will be descriptions of what goes in the
00:02:38
sections not perfect but just to help
00:02:40
you get started uh you know you want to
00:02:43
know what the problem is why it's an
00:02:45
important problem creating this course
00:02:47
to onboard P participants at isip why
00:02:50
that's important um the approach you
00:02:53
know using AI to build the course and
00:02:56
make sure that it's you know evaluates
00:02:59
course three typ YP of courses the
00:03:00
self-paced the uh AI tutor the leader Le
00:03:04
and make sure you know there's 30
00:03:06
learning items or something around that
00:03:09
number of some concept some isip some
00:03:12
engagement ones and that they're
00:03:14
effective you actually learn something a
00:03:16
preest needed for that that they're
00:03:18
efficient that you know it doesn't take
00:03:20
more than an hour and that it's engaging
00:03:22
it's relevant to the students or career
00:03:25
professionals who are taking it and and
00:03:27
they understand why it's relevant and
00:03:28
then the progress well that's what
00:03:30
you're doing right now is making
00:03:31
progress on that course and then there
00:03:34
is um future directions and concluding
00:03:37
remarks there should be an
00:03:39
acknowledgement section which should
00:03:41
also have an ethical acknowledgement
00:03:43
which includes the uh authors uh
00:03:46
repeatedly used these AI tools uh then
00:03:50
there's references you'll notice in the
00:03:52
isip slack I put in you should read this
00:03:54
paper you should read this paper this is
00:03:56
relevant and so I have used um
00:04:00
you know Google Scholar as I've shown
00:04:02
you in the past you can go to Google
00:04:04
Scholar and
00:04:07
um look up papers um so if I go like to
00:04:11
Google
00:04:12
Scholar and you know you can look up the
00:04:15
name of the paper that I gave you like
00:04:18
um you know there was one by you know
00:04:20
Matt Bean AI Dum or something like that
00:04:24
and uh you know and you can look up
00:04:29
those and find um uh click on how to
00:04:32
site it and you know just copy that into
00:04:35
the reference section of the paper after
00:04:37
you read the paper of course um so here
00:04:40
I'm building out the references this is
00:04:42
always important in a scientific paper
00:04:44
know how to use Google Scholar uh look
00:04:46
at the papers I put into the eyes of
00:04:48
slack reference them somewhere in your
00:04:50
white paper then very important and this
00:04:52
is where we get start getting into the
00:04:54
AI tutor stuff as you're using your AI
00:04:57
tools what I do is I add appendices to
00:05:00
the end of the white paper and again
00:05:02
you're trying to build a big white paper
00:05:04
so that the AI these AIS are getting
00:05:06
better and better and they have more and
00:05:07
more context you want to be able to feed
00:05:11
your AI lots of data so this first
00:05:14
appendix was one that I use chat GPD
00:05:17
generating the C uh the course project
00:05:20
foundations with the 30 learning items
00:05:22
in the weekly schedule and this is you
00:05:24
know the input that I gave and the um
00:05:28
output for that
00:05:30
and this is use those appendices because
00:05:33
what you're doing is you're designing
00:05:35
three things you're designing prompts
00:05:38
and you're designing documents like this
00:05:41
and then you iterate with the AI tools
00:05:44
you put in a document you put in a new
00:05:46
prompt and you get out output that you
00:05:48
can use to make your document better and
00:05:51
then guess what you go back to maybe a
00:05:53
different AI tool and you put in the
00:05:54
improved document and improved prompts
00:05:57
and that journey is a process design
00:06:00
Journey so you're designing prompts
00:06:03
you're curating the output into design
00:06:05
documents that keep getting better and
00:06:08
then you're designing the process from
00:06:10
start to finish and usually at the end
00:06:12
of a project you can go back and think
00:06:14
oh my gosh I could have done this
00:06:15
project so much faster if I just started
00:06:17
with these prompts and these documents
00:06:19
and done it in this staged way I would
00:06:22
have finished in a much shorter amount
00:06:24
of time and it would have been much
00:06:25
better so learn this technique of
00:06:28
Designing prompts designing documents
00:06:31
and designing the process and repeat so
00:06:35
this is an example this is one of your
00:06:37
deliverables that you're going to need
00:06:38
is the white paper and I'm showing you
00:06:41
how it's designed with the sections that
00:06:43
you can add to and the references the
00:06:47
ethical citation and acknowledgement and
00:06:49
these appendices which are the use of
00:06:51
the tools now let's um go to the um this
00:06:58
um uh abstract I generated by going out
00:07:02
to um uh I think it was chat GPT let me
00:07:07
bring that up here probably this one
00:07:11
right here so yeah I went out to chat
00:07:14
GPT and I um uh asked for an abstract
00:07:19
for the paper I FedEd the paper and I
00:07:22
said please create an abstract and boom
00:07:25
it created that abstract and then I cut
00:07:27
and paste that abstract into the
00:07:29
abstract section obviously you got to
00:07:31
read it and check it and make sure it's
00:07:33
correct but but that's how you improve
00:07:35
the documents by um creating prompts to
00:07:39
create sections you you create kind of a
00:07:40
draft of a section then you ask the AI
00:07:43
to improve it and then you cut and paste
00:07:45
after you check it make sure you got
00:07:46
everything you asked for and that it's
00:07:48
correct then you cut and paste that and
00:07:51
replace this kind of prompt part with
00:07:54
the actual uh stuff that the AI
00:07:57
generates and you can continue to
00:07:58
improve it and add references and stuff
00:08:00
like that um now let's go to um the uh
00:08:06
ai2 depart now for the AI tutor as I
00:08:11
recall I um used sorry I got too many
00:08:16
windows open here
00:08:19
um let's see if I can find the right
00:08:22
window yeah here we go I used Google
00:08:25
Gemini and what I did is
00:08:29
I uploaded the uh draft course with the
00:08:33
30 learning items and I said please
00:08:34
summarize the 30 learning items in the
00:08:37
course in a table with the column one
00:08:40
being a short description of the
00:08:41
learning item and column two being the
00:08:44
relevance to a particular learner like a
00:08:46
student learner so boom you know
00:08:49
instantaneously I get here's the
00:08:50
learning item description here's the
00:08:52
relevance to a student in the isip
00:08:55
community I looked through this stuff it
00:08:57
looked pretty good and then I said okay
00:09:01
now my prompt says you are a super tutor
00:09:06
who will help me learn these 30 AI
00:09:09
learning items and I should mention that
00:09:11
in the white paper there is a reference
00:09:14
to a paper that just came out I put this
00:09:16
into your slack and said go read this
00:09:20
it's a paper by Harvard uh this guys at
00:09:23
Harvard it's called the cybernetic
00:09:25
teammate a field experiment on
00:09:27
generative AI reshaping teamwork
00:09:29
expertise and around page 50 of this
00:09:32
paper you can get to it by this link
00:09:35
they um have a really really good um set
00:09:40
of prompts to use and so I modeled this
00:09:43
prompt after some of their good prompts
00:09:45
and they say you are you are now a super
00:09:48
tutor who will help me learn these 30 AI
00:09:50
learning items and remember and IAC
00:09:52
participant they'll upload a document
00:09:55
and then they're going to cut and paste
00:09:56
a prompt and that prompts got to be
00:09:58
redesigned better and better but here's
00:10:01
a starting prompt I want you to teach me
00:10:04
them one after the other start by
00:10:06
summarizing the first learning item and
00:10:09
what I will learn and what I need to
00:10:10
know then ask me if I need an example to
00:10:13
make it relevant to my current role or
00:10:15
future career also ask if I want a
00:10:17
multiple choice question to see if I've
00:10:19
learned the item and then give me
00:10:21
feedback if I get it right or wrong uh
00:10:24
then ask me if I'm ready to go to the
00:10:25
next learning item however first tell me
00:10:28
what we are going to do next so boom um
00:10:31
it tells me okay here's what we're going
00:10:33
to do and it it was good it summarized
00:10:35
what I had just said and so let's begin
00:10:37
so the first one you can see they did it
00:10:39
out of order they're going to teach me
00:10:41
about service education and here's a
00:10:43
nice little description of what service
00:10:45
education is that it's
00:10:46
multi-disciplinary and it helps you
00:10:48
apply knowledge to become more expert in
00:10:51
your professional role and then I said
00:10:53
please give an example notice at the
00:10:56
very end it said would you like an
00:10:57
example on how service education relates
00:10:59
to a specific role or career path you're
00:11:02
interested in I said yes please give me
00:11:04
a specific example of how this is
00:11:06
relevant and please make the example fun
00:11:08
and engaging so it came up with you're a
00:11:10
student you're interested in the future
00:11:13
of urban farming and here's how service
00:11:15
education can help you and then I said
00:11:18
okay that's great uh I'm all charged up
00:11:21
please give me a multiple choice example
00:11:23
to see if I've learned this it gave me a
00:11:25
multiple choice uh it was a really good
00:11:28
multiple choice C was right answer so I
00:11:30
said see the answer is that right yes
00:11:32
absolutely correct and then said it's
00:11:33
correct for these reasons and then I
00:11:36
said it said are you ready to move to
00:11:38
the next learning item and I say no not
00:11:40
yet ready I'm almost answered D can you
00:11:43
help me understand why that would have
00:11:45
been the wrong answer of course and it
00:11:47
says it's great that you're thinking
00:11:49
critically about why other options are
00:11:51
incorrect they are a key part of
00:11:53
learning and here's what it says here's
00:11:56
why D is wrong and it gives me a nice
00:11:58
answer and then I say okay I'm ready to
00:11:59
go to the next one and then it gives me
00:12:01
a learning example on service science
00:12:03
and you can see that process of okay
00:12:05
give me a concrete example is relevant
00:12:06
to me give me a multiple choice question
00:12:08
and I can decide yes I want that or no I
00:12:10
don't want that or move on to the next
00:12:12
question depending on okay I've got this
00:12:14
one that's easy so that's how you turn
00:12:16
your uh large language model and I could
00:12:19
have done that with any of these um but
00:12:21
that's how you turn your large language
00:12:23
model into a tutor by designing a
00:12:25
document that gets uploaded with what
00:12:28
has to be learned and designing a prompt
00:12:31
that makes the uh large language model
00:12:34
go into an expert tutor mode with
00:12:36
specific things you want that tutor to
00:12:37
do and you can refine that and improve
00:12:40
that over time so now um let's uh go
00:12:45
back to that word document and now if we
00:12:48
scroll down all the way
00:12:50
to um
00:12:53
appendex I think it's appendix
00:12:56
3 uh let's see here
00:12:59
see there's appendix 2 and I'll give you
00:13:02
a copy of this uh word okay appendix 3
00:13:05
AI tutor so you can see I just copied
00:13:09
the
00:13:10
prompts and copied the output
00:13:15
into uh the appendix of my white paper
00:13:19
and that's what you should do you know
00:13:21
you don't have to have the full thing of
00:13:23
every time you use the AI but when you
00:13:25
do something significant that will help
00:13:28
your
00:13:30
AI produce a better white paper stick
00:13:33
that work into an appendix and then uh
00:13:37
the AI can summarize some of that work
00:13:39
that you did up into different sections
00:13:42
of the body of the paper so that's a
00:13:44
trick you're you're designing promp
00:13:46
you're using the AI tools you're
00:13:48
designing prompts you're designing these
00:13:52
documents that get better and better
00:13:54
over time and then you can every once in
00:13:56
a while pause and reflect and say okay
00:13:58
what's the process I went through of
00:14:00
prompts and papers and could I've done
00:14:02
that better and faster and then you're
00:14:05
learning how to drive your AI car you
00:14:07
are learning how to safely get the
00:14:11
skills you need to use these tools to go
00:14:14
to interesting places just like somebody
00:14:16
learning to drive a car so I will um
00:14:20
upload this uh video
00:14:24
to uh your slack along with the white
00:14:27
paper oh one other thing I want wanted
00:14:29
to show you is it's very important that
00:14:31
you create I suggested that you create
00:14:34
uh
00:14:59
and you can see what I've done here is
00:15:02
um I've created a repo that's what they
00:15:07
call it a GitHub repository called the
00:15:09
isip course it has a read me and
00:15:13
describes what we're doing here and some
00:15:14
of the previous student projects and
00:15:16
then you notice it's got the input data
00:15:18
part and this is where you can upload
00:15:20
things to have them on the web um so
00:15:23
like this uh draft white paper that I
00:15:25
just showed you was already up here and
00:15:28
there it is um that's the white paper
00:15:30
that I just showed you and again it's
00:15:33
important to have your GitHub account to
00:15:35
learn how to design these prompts learn
00:15:37
how to design these documents that you
00:15:41
feed in when you do a prompt and these
00:15:43
workflows and it's very important I
00:15:45
think to have a GitHub uh where you can
00:15:50
uh start working on your workflows uh
00:15:52
with these documents and props and
00:15:54
recording that all right I'll get this
00:15:57
uploaded bye for now
00:16:00
so I'm going to stop the
00:16:02
share I'm going to stop the recording
