March 20, 2023 • 30 mins
In this episode, we chat with Kenji Lamb from the team here at CDN on how AI (Artificial Intelligence) is playing its part in college education and how it might play out in the future. This is tied to an event 'AI in College Education' that we're running this week which if you can't make it, will be recorded and available on the CDN website.
Links to content mentioned in the episode:
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:07):
Hello and welcome to the latest episode
of the Pedagogy Matters Podcast.
The purpose of this podcast is to bring to
the fore some key topics of conversation
in relation to learning and teaching,
to discuss and breakdown aspects
of practice and provide snippets,
advice and guidance is to help integrate
these into our daily practice.
Today I'm delighted to be joined by
Kenji Lam, Digital lead here at CDN,
where we're discussing AI Kenji.
(00:28):
Good afternoon, how are you?
I am good. And I haven't come alone.
I've brought a I with me.
That's one way to whistle.
Already from the outset,
you've kind of brought a I with us.
Well, in that case,
I think what we'll do is hang.
Fair enough. For now.
What I what I think I'm Keenan doing.
I'm sure this is all is to a complete novice.
(00:50):
What is a I? What is it?
What does it means?
There's lots of talk about the minute,
but what does it actually mean?
Stand for yeah.
Over to you.
Artificial intelligence AI is when a
machine has been designed to mimic
human ability in some kind of way,
and this has been going on since the 50s.
(01:13):
I remember when I started studying
computing and people were talking
about AI and I had this amazing
idea as to what that actually was.
At that time it was probably visions
of Hal in the Space Odyssey 2001,
or in my case de Rymon,
which was a robotic cat.
(01:35):
But when I started learning about it,
it turned out that AI was really
a set of fairly complicated rules,
almost like a series of if then statements,
and I was slightly disappointed
by the whole thing,
but they worked really well.
There's a famous example,
I think back in the 60s called ELISA,
(01:58):
which played the role of a
computer psychotherapist,
which was good because basically you
ask a lot of questions and you get
the person to talk about themselves.
So it was based on this kind of
pattern matching framework where
you would type something in.
It would pick out the keyword and then
(02:19):
spit back a predefined response or a
question to tease more information out.
So if you said, ah,
I'm feeling great today,
it would come back with oh why
are you feeling great?
And it would simulate this
kind of conversation.
(02:41):
From there, we've moved on quite a bit.
Computers use machine
learning and deep learning,
which allow them to really learn
from their own input and experience,
and improve the models as they go.
Michael Webb from Jsk,
who's speaking at our AI and
college education event.
(03:02):
He really explains how the current systems.
Work and operate in a simple way that
you know using words that someone like
me can understand that's interesting
and that and you can you've alluded
to a couple of things that I've kind
of made it more prevalent recently.
But yeah,
what I'm keen on pick is,
is how has it evolved over the
(03:24):
recent years kind of so you've
mentioned there's loads of examples.
Could you give us kind of one or
two examples of as to where it's
part of our common place and kind of
where it came from and kind of and
how we've got to where we are now?
So the AI is everywhere,
working behind the scenes where
we probably don't know it's there.
If we're searching for stuff on Google,
(03:45):
there's definitely AI that is bringing
back or helping to bring back a list
of results that are more relevant
to what we're probably looking for.
And those results have got
better and better over time,
although it's incremental,
so we might not always notice
the effect that it has.
(04:05):
It's in everyday life.
I'm diabetic and I go for annual
screenings for my eyes just to see if
there's any problems being developed.
And typically they take a
couple of pictures of your eyes.
Then a consultant or a specialist
will look to see if there's the
potential of damage or problems that
(04:26):
are coming up that they can address.
And that's, well,
I kind of assumed that it was always just a.
People doing that.
But I was at my last visit,
I was talking to the specialist
and she was saying, ah,
we've been using AI for years.
So we use this program called
(04:47):
Autograder and it basically looks at
the scans and it picks out or flags
the scans that they think have problems.
And then a consultant will come
double check it and sign it off.
And she was telling me when
it was first used,
people are really skeptical.
It won't be as good as a human threat to
(05:10):
my job and all the kind of usual stuff.
But they decided,
apparently at the time only to
use the auto grader.
During peak demand,
when there was a rush and you
just needed to get things done,
it was an extra pay device.
And she said the strange thing is
over the years it's got better
with more data and more examples,
(05:31):
better training, better software,
it detects a lot more incidences and
it's not exactly comparable to a consultant,
but getting closer and the thing
that's changed the most is attitude,
really.
Just the way it's become
accepted as another
(05:51):
tool for me. So I have a background in ESOL.
I started out as an ESOL teacher.
That's my educational root in
before I moved over to technology,
so I've always appreciated
just the humble spell check.
To be fair, I'm an English teacher,
(06:11):
but spelling. There are times
you know when you're typing fast.
I'm telling you, and I used to work,
I used to manage a group of teachers,
and I swear they used to stare at
all of the emails I used to send
out and just wait for all the
inevitable slip up that I made.
So spell check has certainly saved my life,
and that's only got better.
(06:32):
You can see it when you're typing
emails in Google or Microsoft.
You can see the way it's suggesting
to complete your sentence.
And more often than not,
I'm just hitting enter and
accepting the suggested phrasing.
Those things just increasingly
I've come to accept,
(06:54):
and it works well in things like
speech to text or voice recognition.
I remember as an English teacher,
I used to teach lessons on pronunciation.
I'd write out a sentence.
And I would get my students to read
it out to the computer through some
(07:16):
dictation exercise so that they could
replicate what was written on the screen.
And in the early days,
this is going back a couple of decades,
you would find that the speech
recognition wasn't that great.
So unless you enunciated each
word in a particular accent,
(07:37):
North American at that time,
because I was in Canada.
It wouldn't really pick up
on what you were saying,
and students had to struggle to
sort of copy the accent that
the computer was expecting.
And too much frustration.
It was a fun exercise,
(07:57):
to be honest,
and it really got them to
concentrate on the particular
pronunciation of particular sounds.
So it was good for me,
but the technology beat me.
I mean, over time.
Those algorithms processing
that natural language AI has
meant that now when you speak,
(08:19):
even if you mess up on the pronunciation
or if you've got a strong accent,
increasingly more accents don't
really trip up the computer so much.
And those exercises are
quite difficult to do now.
Not always,
but they can be still quite difficult,
but it just shows how the
AI has progressed over time.
(08:42):
It appears in things like advertising.
This is not very teachery,
but it's something that causes
me a bit of frustration based
on what you're looking for.
Searching for, buying,
you're always getting presented with.
Oh, why don't you look at this?
Wouldn't you be interested
in buying one of these?
(09:03):
And Amazon is definitely one
of the major culprits here.
It does it far too well for my leg.
And you might want this,
you might want that, yeah.
So it's funny because especially and again,
I've always been a big tech lover,
but I've not been immersed in tech
like you have across your career.
So I guess in my simple world,
this has obviously evolved significantly
since going to these of tech become
(09:23):
a lot more mainstreaming ideas as
well as with our teaching lives.
Is that fair to say?
And fair to assume?
Yeah. And and to be fair,
the best technology is probably
the tech that's invisible.
That you don't see this just happening
and you don't question it that's that's
where it makes a real difference.
I think we just accept it or we've
(09:44):
grown accustomed to it over time.
But yes AI is everywhere.
So you could have touched upon sorry
ChatGPT and obviously which I'm
going to say came to prominence to
the to the non tech experts kind of
really around December last year as.
The answer to AI and the solution to
(10:04):
AI and look, this came out of nowhere,
surprised that it's going to fix
your world change or whatever.
But it's not quite as simple as that, is it?
Because obviously it's been around
for years and it's been lots of
other AI platforms there as well.
So I know when we were talking
about this initially,
you said actually version 3.5.
Is that what you were saying?
Is that correct? Technically, yes.
Yeah.
So tell us a bit around that in particular
(10:26):
and kind of I guess where we are now.
Obviously we've got to chat
some context in terms of where
we are now in terms of AI.
So yes, ChatGPT arrived November last
year and it's made a lot of headlines,
just people fascinated by what it can do.
(10:47):
And the GTP model at the time was
released earlier, people had played
around with it for a year or two.
People knew that it could create.
A song, a poem, write some text on a subject.
But when the general public got access
to it in November and started to
(11:09):
play around with the latest version,
well, at the time, 3.5.
Now it's moved on to four and people were
just so impressed by what was being produced.
It was to another level, and the speed
of the technology moves very quickly.
So the first examples were
(11:30):
really around those kind of wow.
It can write a short essay.
It can have a dialogue,
something that people overlook.
Usually they just ask one question,
get one answer and stop there.
The real thing about ChatGPT is the chat bit.
It's about the fact that you
can have a dialogue and continue
talking about the same subject,
(11:51):
referring back to things
that you've said before,
and the conversation
continues and builds that in.
That whole experience was something
close to the sci-fi characters
that we'd read about in the past.
Now, it's not the only model
(12:12):
that's out there. Facebook, Google,
others have these and other
platforms have been developed,
but this is the one that
we got direct access to,
that we could play around with and.
The GPT its model.
It powers lots of things,
everything from language
(12:33):
translation to content creation.
These tools have been available
through other paid services,
where you could get it to write short
blog posts on a variety of topics.
And I have friends who are writers
for the web, and they're continuously.
Churning out stuff that makes
it sound worse than it is,
(12:53):
but they have a lot of pressure
to produce a lot of material
for the insatiable appetite that
is the web and tools like this,
which allow you to produce
something that's almost copy ready,
just needs a bit of tweaking and
they can put it straight up there.
It's there, it's being used in industry.
(13:15):
The headlines really came about
when people started.
Asking ChatGPT to answer
that test. Basically they took questions
from whatever test paper was around,
and they got ChatGPT to come back
with the response and the response,
although it wasn't scoring A's
(13:38):
or whatever at the high end,
it was good enough and it was much
better than people had been expecting.
And I think that started
the whole discussion around.
Worries around what it would mean
for assessment, education, schools,
colleges, universities and that
(13:58):
has driven a lot of the attention.
But the technology now it can
be applied in so many ways.
Another popular area is coding.
This is available through various services,
but again, open Ai's.
(14:18):
Copilot, which is available through
GitHub and free to students,
allows you to either have code explained
or generated through the tool,
and it's impressive.
I mean, you can do the same things
through ChatGPT as well,
but in copilot it's what they call a pair
(14:41):
programmer now acts as a pair programmer.
And if you're on your own and starting out.
It's pretty impressive that you
can just describe what it is
you're looking for something,
a piece of code to do,
and it will generate a pretty
good example of that code.
It's not perfect.
There are elements where if you
(15:03):
think about the cybersecurity or the
efficiency of the particular code,
or it doesn't quite understand
what your needs are,
but you can refine it as you go
through it offers a brilliant
starting point alongside that, the.
There are areas such as there's a
great example of film production,
which is not something you might think
(15:25):
that a text generator might be able to do.
But there is an example of a studio that
produces a short film through ChatGPT.
So basically they get GPT to write a script,
but then also write director notes to
direct the production of the short film.
(15:47):
And I'll add a link to show
you what they came up with.
It's not brilliant,
but an interesting adaptation
of the technology.
And of course,
when we're talking about teaching,
you can generate useful teaching
resources through this.
For example,
Richard Scott from Ayershire,
(16:09):
who's again speaking at our event.
He'll go through the process
of creating lesson plans.
Which is a really popular
topic at the moment,
and he's been building lesson
plans in his own curricular area,
but also working with
staff across his college,
looking and exploring about different lesson
(16:29):
plans for different curricular areas.
And what he's generating
is really interesting.
There is creativity there,
like coming up with approaches and ideas
that you might not first thought of yourself,
but giving you that sort of
starting point to build on it.
Yeah,
the possibilities seem to be endless.
(16:52):
I think that's fascinating.
I know kind of from our conversations you've
shared look it's not just chat I always got,
but it's not just ChatGPT.
It's it's it's immersed everywhere.
You've talked about imagery and it's
we kind of integrate into Bing and
then all sorts of different kind of
avenues now would become part of our
daily work and our daily practice.
(17:14):
Back and AI as well as ChatGPT or
something else and plus there are
many other solutions out there right
now that are currently active that
can be used as part of our practice.
There has been a lot with images and AI
and they've been around for a while now.
You'll see a lot of applications
(17:34):
that can take something that's in
your existing image and rub it
out as if it wasn't there so.
You're taking a picture of a crowd,
but one person you might
not want to be there.
You can almost magically rub your cursor
(17:55):
over that person and they disappear and the
background that was behind them suddenly,
magically appears.
It's always impressive when that happens,
but that's AI making an
educated guess as to what.
Should be behind that
person and replacing them.
(18:16):
The tools that have come onto the scene,
like mid Journey,
which is a popular one,
Dally II version two now and stable
diffusion are image creators that with
a simple text prompt can produce an
(18:37):
entirely new original piece of art.
Really.
Although when I say simple,
it is possible just to type in text
describing some scene in your head,
but what comes back often needs a
lot of tweaking to get the actual
image that you can use and oh,
(18:59):
last year Dean from.
Fife College had helped us produce well.
He produced a new image,
a new logo for our Scottish
Moodle user group.
Smug.
Literally the best acronym
that we've ever come up with.
(19:20):
But we'd been using this logo for years.
This backdrop every time we use presentations
and record content that we put up.
And Dean had come up with this
new image that looked great.
And he produced it entirely
within mid journey.
That kind of image creation
is going to be useful.
(19:41):
I mean just the ability to produce
something quickly and original
based on any kind of text input.
I mean, you could put anything,
poetry,
anything that just flies into your head.
The computer will produce something
interpreting the language.
To create some really original stuff.
(20:04):
Now there is some controversy
in that these models have been
trained on millions and millions
of images in some cases.
And the images
there's a copyright issue because
the authors haven't been consulted
when the training took place.
(20:25):
And there's a question as to what.
Well, what's the legality of that?
And that's playing out
in the courts just now,
although the genie is out at the bottle,
so these tools are not going away.
And as well as using these tools by students
and staff to come up with new images,
maybe to spark imagination around a story
(20:47):
you're creating or as a kickoff point
to get students to write an essay based
on some image or idea they have practical.
Uses too.
You can for example you can get,
you can get an image creator to
develop a website design type in the
(21:07):
kind of website that you're looking
for the kind of pages and it will
spit back like an amazing after a bit
tweaking a design that just looks ready
made to go up onto the web I mean.
You'll have to code behind it,
but other tools will produce
the code for your website.
(21:28):
But that imagery is.
It's such a useful starting point.
But if you want to see a good examples
of how AI is used more broadly,
JSC have their National Center for AI and on
there there's an explorer part of the site.
(21:50):
That takes you through how AI has
been used for kind of image capture,
image generation.
There is ChatGPT and other tools.
How you can use it creatively
in educational sort of.
It is like when you're writing an essay,
start out with an outline and
then build from there.
(22:11):
There's lots of really interesting
sort of examples to play through,
and some of it requires an institutional.
Login.
But if you're one of the colleges
or universities that, well,
all of them are on the just network,
they'll give you access to a
really good set of examples.
(22:31):
And that's another aspect of these AI tools,
accessibility.
It helps with things like Apple Voiceover
when you've got all of that text to speech,
the tools now.
Will recognize words that
are in our daily view.
(22:53):
When we look at signs,
the AI will determine when part
of a picture includes words,
and then read those words out to
somebody who has visual difficulty,
for example.
That kind of optical character recognition,
or CR,
is being part of the suite of tools
that are available to us even now.
(23:14):
You'll see the latest ones.
When the PC looks at any scene,
they can start to pick out and
identify the objects in front of it.
Recognizing people.
You see that when you use
your camera on your phone,
or any reasonably modern camera now
will pick out the face and zoom in and
(23:35):
make sure that that's focused upon.
When you take your pictures,
it's it's all over the place.
I keep saying that,
but it's it's so true.
One thing I should say is that it's not
always reliable and that was the next
point that comes.
(23:55):
This was kind of a little player not
knowing yet the level that you have.
You're absolutely right there.
You know, for me it allows you
to ask some questions and kind of
curate some information and kind
of gather some information, but.
I'm still the belief then you have to
kind of interpret that in your own way,
put your own stamp on that and kind of
reframe that and for how you want to use it.
And that's just kind of one example as
to how it can be used there as well.
(24:16):
But as always mindful of time I think it's
we could talk about for two 3-4 hours.
And the good news is we can do kind of
in the coming weeks which we'll talk
about shortly and you've kind of alluded
to a little bit about kind of what
the future look likes looks like rather.
But I guess we'll be fair to say we
don't actually know what the future
will look like in the space because.
(24:36):
You know,
the phrase used a lot is kind of
genie out the bottle kind of first
with technology and now with AI and
kind of how quickly it's evolved,
how it's adapted, but also how quickly.
I'm going to say mainstream audience has
picked it up as well and kind of being
keen to explore and kind of look at
how it can be adopted in the real world.
But I guess at this point in time,
(24:57):
now, March 2023,
what's the future look in this space?
So there's going to be more choice in AI.
Microsoft are already expanding
their AI offer into word outlook.
All those things are going to come soon.
Google are doing the same.
They are putting their own AI
(25:17):
models into their own services.
And there's a lot of commercial companies who
are just rushing to market with solutions.
And even though AI isn't perfect,
may never be,
it's getting better all the time.
So all these features will just improve.
(25:40):
There's a quote going around by a
guy called Santiago who says AI is
not going to replace you,
but someone using AI will.
Okay.
So that's it makes it sound like a threat,
(26:00):
but it's not.
AI is an opportunity,
and it's going to be.
Everywhere.
It's something I suppose,
that we have to embrace,
and we have to show our students
how to get the most from it.
And the best way of doing that is
(26:20):
really just playing around with
it just now to see what it can do,
to see what the possibilities are.
Because people are coming up with
new ideas all the time, I suppose.
It makes sense.
I did say I was bringing AI with me.
So this is not ChatGPT,
but another model called well,
(26:43):
which has a voice interface called Alice,
and it's produced by a company called DID.
Always getting that wrong anyway, right?
So let me ask Alice a question and let
me just turn that on here.
Okay. Hey, Alice.
Can you tell me how AI is going
(27:06):
to change college education?
Sure, Kenji, the future of AI in
college education is very exciting.
Many experts believe that AI will
revolutionize the way students
learn and how colleges operate.
AI will help personalize education and
provide students with a more engaging
and interactive learning experience.
(27:28):
AI powered tools can help educators to
identify areas where students are struggling
and provide targeted interventions.
Moreover, AI will help colleges to
streamline administrative tasks
and optimize their operations.
Colleges will be able to use
AI for tasks such as grading,
scheduling,
and course planning.
This will free up time for
(27:48):
educators to focus on teaching and
research and ultimately improve
the quality of education.
Yeah, you know, I did think about
just getting AI to answer everything
you asked me and that's excellent but
frightening at the same time, you know.
And yeah, and having worked,
(28:09):
you know, across education for
obviously not for too many years,
that's kind of a negative phrase.
But kind of 161718,
I forget how many years it's the
workforce is currently diverse
in terms of digital capability,
digital awareness.
So, yeah, it's incredible,
frightening at the same time.
And I think that's gonna help us
(28:30):
bring to close this conversation,
not that I want to,
but just mindful of time is you've
alluded to a couple of points
in terms of the final point
for me is gonna where now how,
where do we start, what do we do?
You touch the point.
Yeah, actually go play with it.
Go miss yourself in this space.
Talk to people.
Understand what I can and can't do.
Speak to these students to
establish what I can and can't
do and how they're using it.
Speak to industry again to understand.
(28:53):
And that will be ever changing in
terms of kind of how they currently
is and how they will be using for
the future because they'll be we're
preparing our students for the next step,
which typically is industry.
So yeah,
there's a lot to start there and
Ken Jones will go on events on 22nd,
is that correct, 22nd of March.
22nd of March and we're drawing
speakers from the college community
who are already exploring and
(29:15):
using this with their students,
and so we'll provide some guidance
around how you can put together
the best prompts if you want
to get something from AI.
We'll talk at just what the
implications are for policy.
How colleges might respond
to this and encouraging,
I would say hopefully encouraging your
students and staff to get the best
(29:35):
from it and talk around just examples
about what the student perspective is,
how it's being used,
and how it could be used in educational ways.
And we'll record everything
and make it available.
Although if you're listening
to this in April or May,
I swear that everything
will have moved on by then.
Just.
Go to YouTube and look it up or or
in fact just ask your smart speaker,
(29:56):
it probably knows no, fantastic.
Thank you Kenji.
And again for information
on an event the 22nd,
please look at CDN website and
kind of get that information.
And really this is just a start of
the conversation because there's a lot
more than there's a lot of potential
benefits around the educational
workload and so on and so forth.
But yeah,
we kind of follow Kenji and follow
ourselves kind of look kind of the
(30:17):
work we're doing in that space.
But Kenji as always fantastic.
Thank you for your time.
Front of all knowledge and yeah,
look forward to seeing all this evolves.
Always a pleasure.
Hello and welcome to the latest episode
of the Pedagogy Matters Podcast.
The purpose of this podcast is to bring to
the fore some key topics of conversation
in relation to learning and teaching,
to discuss and breakdown aspects
of practice and provide snippets,
advice and guidance is to help integrate
these into our daily practice.
Today I'm delighted to be joined by
Kenji Lam, Digital lead here at CDN,
where we're discussing AI Kenji.
(00:28):
Good afternoon, how are you?
I am good. And I haven't come alone.
I've brought a I with me.
That's one way to whistle.
Already from the outset,
you've kind of brought a I with us.
Well, in that case,
I think what we'll do is hang.
Fair enough. For now.
What I what I think I'm Keenan doing.
I'm sure this is all is to a complete novice.
(00:50):
What is a I? What is it?
What does it means?
There's lots of talk about the minute,
but what does it actually mean?
Stand for yeah.
Over to you.
Artificial intelligence AI is when a
machine has been designed to mimic
human ability in some kind of way,
and this has been going on since the 50s.
(01:13):
I remember when I started studying
computing and people were talking
about AI and I had this amazing
idea as to what that actually was.
At that time it was probably visions
of Hal in the Space Odyssey 2001,
or in my case de Rymon,
which was a robotic cat.
(01:35):
But when I started learning about it,
it turned out that AI was really
a set of fairly complicated rules,
almost like a series of if then statements,
and I was slightly disappointed
by the whole thing,
but they worked really well.
There's a famous example,
I think back in the 60s called ELISA,
(01:58):
which played the role of a
computer psychotherapist,
which was good because basically you
ask a lot of questions and you get
the person to talk about themselves.
So it was based on this kind of
pattern matching framework where
you would type something in.
It would pick out the keyword and then
(02:19):
spit back a predefined response or a
question to tease more information out.
So if you said, ah,
I'm feeling great today,
it would come back with oh why
are you feeling great?
And it would simulate this
kind of conversation.
(02:41):
From there, we've moved on quite a bit.
Computers use machine
learning and deep learning,
which allow them to really learn
from their own input and experience,
and improve the models as they go.
Michael Webb from Jsk,
who's speaking at our AI and
college education event.
(03:02):
He really explains how the current systems.
Work and operate in a simple way that
you know using words that someone like
me can understand that's interesting
and that and you can you've alluded
to a couple of things that I've kind
of made it more prevalent recently.
But yeah,
what I'm keen on pick is,
is how has it evolved over the
(03:24):
recent years kind of so you've
mentioned there's loads of examples.
Could you give us kind of one or
two examples of as to where it's
part of our common place and kind of
where it came from and kind of and
how we've got to where we are now?
So the AI is everywhere,
working behind the scenes where
we probably don't know it's there.
If we're searching for stuff on Google,
(03:45):
there's definitely AI that is bringing
back or helping to bring back a list
of results that are more relevant
to what we're probably looking for.
And those results have got
better and better over time,
although it's incremental,
so we might not always notice
the effect that it has.
(04:05):
It's in everyday life.
I'm diabetic and I go for annual
screenings for my eyes just to see if
there's any problems being developed.
And typically they take a
couple of pictures of your eyes.
Then a consultant or a specialist
will look to see if there's the
potential of damage or problems that
(04:26):
are coming up that they can address.
And that's, well,
I kind of assumed that it was always just a.
People doing that.
But I was at my last visit,
I was talking to the specialist
and she was saying, ah,
we've been using AI for years.
So we use this program called
(04:47):
Autograder and it basically looks at
the scans and it picks out or flags
the scans that they think have problems.
And then a consultant will come
double check it and sign it off.
And she was telling me when
it was first used,
people are really skeptical.
It won't be as good as a human threat to
(05:10):
my job and all the kind of usual stuff.
But they decided,
apparently at the time only to
use the auto grader.
During peak demand,
when there was a rush and you
just needed to get things done,
it was an extra pay device.
And she said the strange thing is
over the years it's got better
with more data and more examples,
(05:31):
better training, better software,
it detects a lot more incidences and
it's not exactly comparable to a consultant,
but getting closer and the thing
that's changed the most is attitude,
really.
Just the way it's become
accepted as another
(05:51):
tool for me. So I have a background in ESOL.
I started out as an ESOL teacher.
That's my educational root in
before I moved over to technology,
so I've always appreciated
just the humble spell check.
To be fair, I'm an English teacher,
(06:11):
but spelling. There are times
you know when you're typing fast.
I'm telling you, and I used to work,
I used to manage a group of teachers,
and I swear they used to stare at
all of the emails I used to send
out and just wait for all the
inevitable slip up that I made.
So spell check has certainly saved my life,
and that's only got better.
(06:32):
You can see it when you're typing
emails in Google or Microsoft.
You can see the way it's suggesting
to complete your sentence.
And more often than not,
I'm just hitting enter and
accepting the suggested phrasing.
Those things just increasingly
I've come to accept,
(06:54):
and it works well in things like
speech to text or voice recognition.
I remember as an English teacher,
I used to teach lessons on pronunciation.
I'd write out a sentence.
And I would get my students to read
it out to the computer through some
(07:16):
dictation exercise so that they could
replicate what was written on the screen.
And in the early days,
this is going back a couple of decades,
you would find that the speech
recognition wasn't that great.
So unless you enunciated each
word in a particular accent,
(07:37):
North American at that time,
because I was in Canada.
It wouldn't really pick up
on what you were saying,
and students had to struggle to
sort of copy the accent that
the computer was expecting.
And too much frustration.
It was a fun exercise,
(07:57):
to be honest,
and it really got them to
concentrate on the particular
pronunciation of particular sounds.
So it was good for me,
but the technology beat me.
I mean, over time.
Those algorithms processing
that natural language AI has
meant that now when you speak,
(08:19):
even if you mess up on the pronunciation
or if you've got a strong accent,
increasingly more accents don't
really trip up the computer so much.
And those exercises are
quite difficult to do now.
Not always,
but they can be still quite difficult,
but it just shows how the
AI has progressed over time.
(08:42):
It appears in things like advertising.
This is not very teachery,
but it's something that causes
me a bit of frustration based
on what you're looking for.
Searching for, buying,
you're always getting presented with.
Oh, why don't you look at this?
Wouldn't you be interested
in buying one of these?
(09:03):
And Amazon is definitely one
of the major culprits here.
It does it far too well for my leg.
And you might want this,
you might want that, yeah.
So it's funny because especially and again,
I've always been a big tech lover,
but I've not been immersed in tech
like you have across your career.
So I guess in my simple world,
this has obviously evolved significantly
since going to these of tech become
(09:23):
a lot more mainstreaming ideas as
well as with our teaching lives.
Is that fair to say?
And fair to assume?
Yeah. And and to be fair,
the best technology is probably
the tech that's invisible.
That you don't see this just happening
and you don't question it that's that's
where it makes a real difference.
I think we just accept it or we've
(09:44):
grown accustomed to it over time.
But yes AI is everywhere.
So you could have touched upon sorry
ChatGPT and obviously which I'm
going to say came to prominence to
the to the non tech experts kind of
really around December last year as.
The answer to AI and the solution to
(10:04):
AI and look, this came out of nowhere,
surprised that it's going to fix
your world change or whatever.
But it's not quite as simple as that, is it?
Because obviously it's been around
for years and it's been lots of
other AI platforms there as well.
So I know when we were talking
about this initially,
you said actually version 3.5.
Is that what you were saying?
Is that correct? Technically, yes.
Yeah.
So tell us a bit around that in particular
(10:26):
and kind of I guess where we are now.
Obviously we've got to chat
some context in terms of where
we are now in terms of AI.
So yes, ChatGPT arrived November last
year and it's made a lot of headlines,
just people fascinated by what it can do.
(10:47):
And the GTP model at the time was
released earlier, people had played
around with it for a year or two.
People knew that it could create.
A song, a poem, write some text on a subject.
But when the general public got access
to it in November and started to
(11:09):
play around with the latest version,
well, at the time, 3.5.
Now it's moved on to four and people were
just so impressed by what was being produced.
It was to another level, and the speed
of the technology moves very quickly.
So the first examples were
(11:30):
really around those kind of wow.
It can write a short essay.
It can have a dialogue,
something that people overlook.
Usually they just ask one question,
get one answer and stop there.
The real thing about ChatGPT is the chat bit.
It's about the fact that you
can have a dialogue and continue
talking about the same subject,
(11:51):
referring back to things
that you've said before,
and the conversation
continues and builds that in.
That whole experience was something
close to the sci-fi characters
that we'd read about in the past.
Now, it's not the only model
(12:12):
that's out there. Facebook, Google,
others have these and other
platforms have been developed,
but this is the one that
we got direct access to,
that we could play around with and.
The GPT its model.
It powers lots of things,
everything from language
(12:33):
translation to content creation.
These tools have been available
through other paid services,
where you could get it to write short
blog posts on a variety of topics.
And I have friends who are writers
for the web, and they're continuously.
Churning out stuff that makes
it sound worse than it is,
(12:53):
but they have a lot of pressure
to produce a lot of material
for the insatiable appetite that
is the web and tools like this,
which allow you to produce
something that's almost copy ready,
just needs a bit of tweaking and
they can put it straight up there.
It's there, it's being used in industry.
(13:15):
The headlines really came about
when people started.
Asking ChatGPT to answer
that test. Basically they took questions
from whatever test paper was around,
and they got ChatGPT to come back
with the response and the response,
although it wasn't scoring A's
(13:38):
or whatever at the high end,
it was good enough and it was much
better than people had been expecting.
And I think that started
the whole discussion around.
Worries around what it would mean
for assessment, education, schools,
colleges, universities and that
(13:58):
has driven a lot of the attention.
But the technology now it can
be applied in so many ways.
Another popular area is coding.
This is available through various services,
but again, open Ai's.
(14:18):
Copilot, which is available through
GitHub and free to students,
allows you to either have code explained
or generated through the tool,
and it's impressive.
I mean, you can do the same things
through ChatGPT as well,
but in copilot it's what they call a pair
(14:41):
programmer now acts as a pair programmer.
And if you're on your own and starting out.
It's pretty impressive that you
can just describe what it is
you're looking for something,
a piece of code to do,
and it will generate a pretty
good example of that code.
It's not perfect.
There are elements where if you
(15:03):
think about the cybersecurity or the
efficiency of the particular code,
or it doesn't quite understand
what your needs are,
but you can refine it as you go
through it offers a brilliant
starting point alongside that, the.
There are areas such as there's a
great example of film production,
which is not something you might think
(15:25):
that a text generator might be able to do.
But there is an example of a studio that
produces a short film through ChatGPT.
So basically they get GPT to write a script,
but then also write director notes to
direct the production of the short film.
(15:47):
And I'll add a link to show
you what they came up with.
It's not brilliant,
but an interesting adaptation
of the technology.
And of course,
when we're talking about teaching,
you can generate useful teaching
resources through this.
For example,
Richard Scott from Ayershire,
(16:09):
who's again speaking at our event.
He'll go through the process
of creating lesson plans.
Which is a really popular
topic at the moment,
and he's been building lesson
plans in his own curricular area,
but also working with
staff across his college,
looking and exploring about different lesson
(16:29):
plans for different curricular areas.
And what he's generating
is really interesting.
There is creativity there,
like coming up with approaches and ideas
that you might not first thought of yourself,
but giving you that sort of
starting point to build on it.
Yeah,
the possibilities seem to be endless.
(16:52):
I think that's fascinating.
I know kind of from our conversations you've
shared look it's not just chat I always got,
but it's not just ChatGPT.
It's it's it's immersed everywhere.
You've talked about imagery and it's
we kind of integrate into Bing and
then all sorts of different kind of
avenues now would become part of our
daily work and our daily practice.
(17:14):
Back and AI as well as ChatGPT or
something else and plus there are
many other solutions out there right
now that are currently active that
can be used as part of our practice.
There has been a lot with images and AI
and they've been around for a while now.
You'll see a lot of applications
(17:34):
that can take something that's in
your existing image and rub it
out as if it wasn't there so.
You're taking a picture of a crowd,
but one person you might
not want to be there.
You can almost magically rub your cursor
(17:55):
over that person and they disappear and the
background that was behind them suddenly,
magically appears.
It's always impressive when that happens,
but that's AI making an
educated guess as to what.
Should be behind that
person and replacing them.
(18:16):
The tools that have come onto the scene,
like mid Journey,
which is a popular one,
Dally II version two now and stable
diffusion are image creators that with
a simple text prompt can produce an
(18:37):
entirely new original piece of art.
Really.
Although when I say simple,
it is possible just to type in text
describing some scene in your head,
but what comes back often needs a
lot of tweaking to get the actual
image that you can use and oh,
(18:59):
last year Dean from.
Fife College had helped us produce well.
He produced a new image,
a new logo for our Scottish
Moodle user group.
Smug.
Literally the best acronym
that we've ever come up with.
(19:20):
But we'd been using this logo for years.
This backdrop every time we use presentations
and record content that we put up.
And Dean had come up with this
new image that looked great.
And he produced it entirely
within mid journey.
That kind of image creation
is going to be useful.
(19:41):
I mean just the ability to produce
something quickly and original
based on any kind of text input.
I mean, you could put anything,
poetry,
anything that just flies into your head.
The computer will produce something
interpreting the language.
To create some really original stuff.
(20:04):
Now there is some controversy
in that these models have been
trained on millions and millions
of images in some cases.
And the images
there's a copyright issue because
the authors haven't been consulted
when the training took place.
(20:25):
And there's a question as to what.
Well, what's the legality of that?
And that's playing out
in the courts just now,
although the genie is out at the bottle,
so these tools are not going away.
And as well as using these tools by students
and staff to come up with new images,
maybe to spark imagination around a story
(20:47):
you're creating or as a kickoff point
to get students to write an essay based
on some image or idea they have practical.
Uses too.
You can for example you can get,
you can get an image creator to
develop a website design type in the
(21:07):
kind of website that you're looking
for the kind of pages and it will
spit back like an amazing after a bit
tweaking a design that just looks ready
made to go up onto the web I mean.
You'll have to code behind it,
but other tools will produce
the code for your website.
(21:28):
But that imagery is.
It's such a useful starting point.
But if you want to see a good examples
of how AI is used more broadly,
JSC have their National Center for AI and on
there there's an explorer part of the site.
(21:50):
That takes you through how AI has
been used for kind of image capture,
image generation.
There is ChatGPT and other tools.
How you can use it creatively
in educational sort of.
It is like when you're writing an essay,
start out with an outline and
then build from there.
(22:11):
There's lots of really interesting
sort of examples to play through,
and some of it requires an institutional.
Login.
But if you're one of the colleges
or universities that, well,
all of them are on the just network,
they'll give you access to a
really good set of examples.
(22:31):
And that's another aspect of these AI tools,
accessibility.
It helps with things like Apple Voiceover
when you've got all of that text to speech,
the tools now.
Will recognize words that
are in our daily view.
(22:53):
When we look at signs,
the AI will determine when part
of a picture includes words,
and then read those words out to
somebody who has visual difficulty,
for example.
That kind of optical character recognition,
or CR,
is being part of the suite of tools
that are available to us even now.
(23:14):
You'll see the latest ones.
When the PC looks at any scene,
they can start to pick out and
identify the objects in front of it.
Recognizing people.
You see that when you use
your camera on your phone,
or any reasonably modern camera now
will pick out the face and zoom in and
(23:35):
make sure that that's focused upon.
When you take your pictures,
it's it's all over the place.
I keep saying that,
but it's it's so true.
One thing I should say is that it's not
always reliable and that was the next
point that comes.
(23:55):
This was kind of a little player not
knowing yet the level that you have.
You're absolutely right there.
You know, for me it allows you
to ask some questions and kind of
curate some information and kind
of gather some information, but.
I'm still the belief then you have to
kind of interpret that in your own way,
put your own stamp on that and kind of
reframe that and for how you want to use it.
And that's just kind of one example as
to how it can be used there as well.
(24:16):
But as always mindful of time I think it's
we could talk about for two 3-4 hours.
And the good news is we can do kind of
in the coming weeks which we'll talk
about shortly and you've kind of alluded
to a little bit about kind of what
the future look likes looks like rather.
But I guess we'll be fair to say we
don't actually know what the future
will look like in the space because.
(24:36):
You know,
the phrase used a lot is kind of
genie out the bottle kind of first
with technology and now with AI and
kind of how quickly it's evolved,
how it's adapted, but also how quickly.
I'm going to say mainstream audience has
picked it up as well and kind of being
keen to explore and kind of look at
how it can be adopted in the real world.
But I guess at this point in time,
(24:57):
now, March 2023,
what's the future look in this space?
So there's going to be more choice in AI.
Microsoft are already expanding
their AI offer into word outlook.
All those things are going to come soon.
Google are doing the same.
They are putting their own AI
(25:17):
models into their own services.
And there's a lot of commercial companies who
are just rushing to market with solutions.
And even though AI isn't perfect,
may never be,
it's getting better all the time.
So all these features will just improve.
(25:40):
There's a quote going around by a
guy called Santiago who says AI is
not going to replace you,
but someone using AI will.
Okay.
So that's it makes it sound like a threat,
(26:00):
but it's not.
AI is an opportunity,
and it's going to be.
Everywhere.
It's something I suppose,
that we have to embrace,
and we have to show our students
how to get the most from it.
And the best way of doing that is
(26:20):
really just playing around with
it just now to see what it can do,
to see what the possibilities are.
Because people are coming up with
new ideas all the time, I suppose.
It makes sense.
I did say I was bringing AI with me.
So this is not ChatGPT,
but another model called well,
(26:43):
which has a voice interface called Alice,
and it's produced by a company called DID.
Always getting that wrong anyway, right?
So let me ask Alice a question and let
me just turn that on here.
Okay. Hey, Alice.
Can you tell me how AI is going
(27:06):
to change college education?
Sure, Kenji, the future of AI in
college education is very exciting.
Many experts believe that AI will
revolutionize the way students
learn and how colleges operate.
AI will help personalize education and
provide students with a more engaging
and interactive learning experience.
(27:28):
AI powered tools can help educators to
identify areas where students are struggling
and provide targeted interventions.
Moreover, AI will help colleges to
streamline administrative tasks
and optimize their operations.
Colleges will be able to use
AI for tasks such as grading,
scheduling,
and course planning.
This will free up time for
(27:48):
educators to focus on teaching and
research and ultimately improve
the quality of education.
Yeah, you know, I did think about
just getting AI to answer everything
you asked me and that's excellent but
frightening at the same time, you know.
And yeah, and having worked,
(28:09):
you know, across education for
obviously not for too many years,
that's kind of a negative phrase.
But kind of 161718,
I forget how many years it's the
workforce is currently diverse
in terms of digital capability,
digital awareness.
So, yeah, it's incredible,
frightening at the same time.
And I think that's gonna help us
(28:30):
bring to close this conversation,
not that I want to,
but just mindful of time is you've
alluded to a couple of points
in terms of the final point
for me is gonna where now how,
where do we start, what do we do?
You touch the point.
Yeah, actually go play with it.
Go miss yourself in this space.
Talk to people.
Understand what I can and can't do.
Speak to these students to
establish what I can and can't
do and how they're using it.
Speak to industry again to understand.
(28:53):
And that will be ever changing in
terms of kind of how they currently
is and how they will be using for
the future because they'll be we're
preparing our students for the next step,
which typically is industry.
So yeah,
there's a lot to start there and
Ken Jones will go on events on 22nd,
is that correct, 22nd of March.
22nd of March and we're drawing
speakers from the college community
who are already exploring and
(29:15):
using this with their students,
and so we'll provide some guidance
around how you can put together
the best prompts if you want
to get something from AI.
We'll talk at just what the
implications are for policy.
How colleges might respond
to this and encouraging,
I would say hopefully encouraging your
students and staff to get the best
(29:35):
from it and talk around just examples
about what the student perspective is,
how it's being used,
and how it could be used in educational ways.
And we'll record everything
and make it available.
Although if you're listening
to this in April or May,
I swear that everything
will have moved on by then.
Just.
Go to YouTube and look it up or or
in fact just ask your smart speaker,
(29:56):
it probably knows no, fantastic.
Thank you Kenji.
And again for information
on an event the 22nd,
please look at CDN website and
kind of get that information.
And really this is just a start of
the conversation because there's a lot
more than there's a lot of potential
benefits around the educational
workload and so on and so forth.
But yeah,
we kind of follow Kenji and follow
ourselves kind of look kind of the
(30:17):
work we're doing in that space.
But Kenji as always fantastic.
Thank you for your time.
Front of all knowledge and yeah,
look forward to seeing all this evolves.
Always a pleasure.
Popular Podcasts
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
Intentionally Disturbing
Join me on this podcast as I navigate the murky waters of human behavior, current events, and personal anecdotes through in-depth interviews with incredible people—all served with a generous helping of sarcasm and satire. After years as a forensic and clinical psychologist, I offer a unique interview style and a low tolerance for bullshit, quickly steering conversations toward depth and darkness. I honor the seriousness while also appreciating wit. I’m your guide through the twisted labyrinth of the human psyche, armed with dark humor and biting wit.