00:00:00
now some people will tell you to take
00:00:02
each paper you read and write a summary
00:00:04
paragraph and then edit all those
00:00:06
paragraphs together but this is a
00:00:08
terrible idea and it's a terrible idea
00:00:10
because when you first start reading you
00:00:12
don't know how an individual paper fits
00:00:14
into the broader context and there might
00:00:16
be Concepts that you don't yet
00:00:18
understand you also might not be able to
00:00:21
tell if the paper is any good because
00:00:24
not all published papers are and this is
00:00:27
one of the many basic basic things that
00:00:30
the AI Advocates fail to acknowledge a
00:00:33
lot of published papers are just bad so
00:00:36
even if you trust the AI platform which
00:00:39
you shouldn't you can't always trust the
00:00:42
literature that it's
00:00:47
summarizing there are countless videos
00:00:49
online showing you how to use AI to
00:00:51
write your literature review and all of
00:00:54
them make it look easy but all of the
00:00:56
ones that I've seen get fundamental
00:00:58
things wrong and the problem as with a
00:01:01
lot of bad advice is that it will seem
00:01:04
to work at first but cause major or even
00:01:07
irreparable damage later so the danger
00:01:11
is that AI can help you produce
00:01:13
something that looks like a literature
00:01:15
review but on a topic that you don't
00:01:18
understand and to the uninformed reader
00:01:21
it might look impressive but there won't
00:01:23
be any real insight into the topic it
00:01:26
will probably also contain serious
00:01:28
mistakes that you might might not notice
00:01:30
if you don't have the
00:01:32
expertise so as I've said before
00:01:35
anything you submit as your own work you
00:01:37
have to be able to defend yeah if you
00:01:40
have a lazy examiner who doesn't read
00:01:42
your thesis you might be okay but a good
00:01:44
examiner will quite quickly figure out
00:01:47
that you've used AI they will find the
00:01:49
parts where you don't understand what
00:01:51
you've submitted as your work and they
00:01:53
will quite rightly give you a very hard
00:01:56
time and if they discover that you
00:01:58
didn't write major sections of your
00:02:01
thesis you will
00:02:04
fail so can AI be used ethically and
00:02:08
effectively well yes probably but if you
00:02:12
use it you shouldn't be dependent on it
00:02:16
you should treat it like an enthusiastic
00:02:18
but slightly incompetent intern in that
00:02:21
you can get it to do some work but you
00:02:23
need to check everything that it's done
00:02:27
and while it might make suggestions you
00:02:29
should never let it make decisions for
00:02:32
you and if you don't know enough about a
00:02:34
topic to judge what AI is doing you've
00:02:38
got to fix
00:02:39
that so I started my PhD back in
00:02:44
2003 which means I'm closer to the
00:02:47
generation that searched the literature
00:02:50
by going to the library and reading
00:02:52
physical printed copies of journals then
00:02:55
I am to the current generation with all
00:02:57
of these AI tools
00:03:00
but despite all of the advances in
00:03:03
technology over the last 21 years the
00:03:05
fundamentals have not changed you still
00:03:08
need a good knowledge and understanding
00:03:11
of the literature to be able to write
00:03:13
about it and isn't that the point surely
00:03:17
you want to develop your own expertise
00:03:20
you want to know the literature and have
00:03:22
some insight of your own rather than
00:03:24
following some chat gbt generated
00:03:27
template and surely you want to develop
00:03:31
your own skills so that you can write
00:03:33
with confidence without being dependent
00:03:35
on an AI
00:03:37
platform so here's how to write a
00:03:39
literature review without using AI it
00:03:42
might be slower especially in the
00:03:44
beginning but you will be so much
00:03:46
stronger for having done the heavy
00:03:48
lifting
00:03:49
yourself but before we get started if we
00:03:52
haven't met before my name is James
00:03:53
Hatton I'm a former physicist and since
00:03:57
2010 I've worked full-time coaching PhD
00:04:00
students from all kinds of academic
00:04:02
disciplines in academic writing and
00:04:05
other General project management skills
00:04:08
with the overall aim of making your PhD
00:04:11
a more positive experience so if you
00:04:14
want to know more about what I do check
00:04:16
out my website at phd. Academy and sign
00:04:19
up for email notifications so I can let
00:04:21
you know directly when I publish new
00:04:24
videos for this video I'm going to
00:04:26
assume that you already have some idea
00:04:28
of what you want to study study but if
00:04:30
you want to know how to find or rather
00:04:33
develop a research topic then leave a
00:04:35
comment below and I'll cover it in a
00:04:37
separate
00:04:38
video so I'm going to start with a
00:04:42
fundamental Rule and that is to never
00:04:46
write anything that you don't understand
00:04:49
yourself now some people will tell you
00:04:51
to take each paper you read and write a
00:04:53
summary paragraph and then edit all
00:04:55
those paragraphs together but this is a
00:04:57
terrible idea and it's a terrible idea
00:05:00
because when you first start reading you
00:05:02
don't know how an individual paper fits
00:05:04
into the broader context and there might
00:05:06
be Concepts that you don't yet
00:05:08
understand you also might not be able to
00:05:11
tell if the paper is any good because
00:05:14
not all published papers are and this is
00:05:17
one of the many basic basic things that
00:05:20
the AI Advocates fail to acknowledge a
00:05:23
lot of published papers are just bad so
00:05:26
even if you trust the AI platform which
00:05:29
you should
00:05:30
you can't always trust the literature
00:05:32
that it's
00:05:33
summarizing and if you follow the
00:05:36
standard advice of writing these summary
00:05:38
paragraphs or taking key points and then
00:05:41
paraphrasing them without understanding
00:05:43
them you will fill pages with content
00:05:46
that looks like a literature review but
00:05:49
you will know deep down that you don't
00:05:51
understand what you're writing about and
00:05:53
if you have any kind of impostor
00:05:55
syndrome this approach is just going to
00:05:58
make it worse
00:06:00
so we have to build some knowledge and
00:06:03
understanding of the literature before
00:06:04
we can write about it with any
00:06:06
confidence this means forgetting about
00:06:09
writing at least initially and taking
00:06:12
some time to just
00:06:14
read now I can't give you an exact
00:06:17
step-by-step process here because it
00:06:19
will vary depending on your situation
00:06:22
your current level of knowledge the
00:06:24
state of your field and what you need to
00:06:26
find out so if you want a process that
00:06:29
you can just follow from start to finish
00:06:31
without thinking or using your own
00:06:33
judgment it doesn't exist whether you
00:06:36
use AI or not but there are general
00:06:40
principles that you can follow and adapt
00:06:43
as your needs change over
00:06:45
time so generally speaking there are
00:06:48
different levels of knowledge that you
00:06:50
need to develop so let's start with the
00:06:53
broad picture so this is an
00:06:55
understanding of the trends and debates
00:06:58
and the kinds of problems people are
00:07:00
working on so we're not too worried
00:07:03
about the details of individual papers
00:07:06
here because initially we're just
00:07:09
getting a low resolution picture of the
00:07:12
field and you can usually do this fairly
00:07:15
quickly so if you take a handful of
00:07:18
recent papers on a specific topic and
00:07:21
just read the introductions you will
00:07:23
probably notice that they all say
00:07:26
similar things about the current state
00:07:28
of the field
00:07:30
so they will mention the same problems
00:07:32
or ongoing debates they will mention
00:07:35
many of the same Concepts and they will
00:07:37
refer to the same key papers for example
00:07:42
one of my projects involved looking at
00:07:45
silicon nanoparticle luminescence
00:07:48
basically getting silicon to give off
00:07:51
light every single Paper said the same
00:07:54
thing in the
00:07:55
introduction that there had been a
00:07:57
surprise result that others had
00:08:00
reproduced it but nobody knew the
00:08:02
underlying mechanism and if we could
00:08:04
figure this out then there was huge
00:08:06
technological
00:08:07
potential now there was a lot of heavy
00:08:09
technical detail behind it and there
00:08:11
were many different experiments and
00:08:13
proposed theoretical explanations but
00:08:15
the basic story is easy to
00:08:19
understand and the key paper that
00:08:22
everybody referenced was by Canam a name
00:08:25
that I still remember two decades later
00:08:27
even though I haven't looked at the
00:08:29
topics since I don't remember all the
00:08:31
others but I remember that one and this
00:08:34
brings us to the next key Point not all
00:08:37
the literature is of equal value most
00:08:40
papers will have very little impact on
00:08:43
the field but there will be a few that
00:08:46
have disproportionate impact perhaps
00:08:48
because they made some major Discovery
00:08:50
or invented a new technique or developed
00:08:53
a new Theory and these get cited far
00:08:56
more than anything
00:08:58
else and if everybody else is referring
00:09:01
to these people you need to know who
00:09:03
they are and what they did that was so
00:09:06
important and this brings us to the next
00:09:09
rule you should always go back to the
00:09:13
original source and check what they
00:09:15
actually did and said because when
00:09:18
somebody else summarizes a paper unless
00:09:21
they quote directly there is always a
00:09:24
subtle change now in some cases the
00:09:28
secondary source expl explains the
00:09:29
concepts better but I've seen so many
00:09:32
instances where they get it wrong and
00:09:35
then other people have cited the
00:09:37
secondary source so it's always a good
00:09:40
idea to go back to those key original
00:09:43
sources and try to understand what
00:09:47
problem they
00:09:48
addressed what they did that was new and
00:09:51
what influence this had on the
00:09:53
field and although it's important to
00:09:57
take some time to try to understand
00:09:58
these sources is you maybe don't need to
00:10:01
get too lost in the detail because you
00:10:03
can always come back to these sources
00:10:05
multiple times throughout your
00:10:07
PhD and if you come back to a paper a
00:10:10
few months or years later after reading
00:10:13
a lot more after conducting some of your
00:10:15
own research you might have a very
00:10:18
different understanding of it so a quick
00:10:21
side note here years ago I saw a
00:10:25
professor on Twitter saying that they
00:10:28
only ever read anything once and I wish
00:10:31
I had saved it because it was such a
00:10:33
perfect example of the kind of
00:10:35
misleading that you see online
00:10:38
from people who really should know
00:10:39
better any Professor who says they only
00:10:43
ever read anything once is either a
00:10:46
genius is lying or is really bad at
00:10:50
their job so you can and should come
00:10:53
back to the most important sources and
00:10:54
read them multiple times because your
00:10:57
perspective on them will change as your
00:10:59
expertise grows so as you build this
00:11:02
initial picture of the field the general
00:11:05
Trends and debates and the key
00:11:06
influential figures you will inevitably
00:11:09
come across Concepts theories and
00:11:11
techniques that you don't understand and
00:11:13
it's worth taking some time to try to
00:11:16
understand some of these now you will
00:11:18
need to be selective and prioritize here
00:11:21
to avoid getting overwhelmed and part of
00:11:24
this is accepting that you can't know
00:11:26
everything and there will always be gaps
00:11:29
in your
00:11:30
knowledge this can be a little scary as
00:11:32
you've probably come through an
00:11:33
education system where you're graded
00:11:36
based on how much of the syllabus you
00:11:38
know but at PhD level we have to let go
00:11:42
of that way of thinking because the
00:11:44
amount of literature and knowledge is
00:11:47
endless you will never know everything
00:11:50
but you can identify some key Concepts
00:11:53
that you need to understand for your
00:11:55
project and one of the most important
00:11:57
aspects here that is so often overlook
00:11:59
looked is developing an understanding of
00:12:02
common research and analytical
00:12:05
techniques and this is crucial not only
00:12:08
for your own research but to understand
00:12:11
and assess the literature that you
00:12:13
read so many videos I've seen about
00:12:16
literature reviews just take what the
00:12:18
paper says without any critical thought
00:12:21
but you need to be able to look at how
00:12:22
they conducted their research and reach
00:12:24
their
00:12:26
conclusions now it takes time to build
00:12:28
this knowledge so again it's important
00:12:30
to be selective either based on the kind
00:12:33
of research you want to do or the most
00:12:35
common techniques used in the field and
00:12:38
it might be worth finding good
00:12:40
authoritative sources on specific
00:12:43
techniques so if you're doing IPA for
00:12:46
example maybe you should get a copy of
00:12:49
the book by Smith flowers and
00:12:51
Lin again you don't need to get lost in
00:12:54
all of the detail but you can get an
00:12:57
initial understanding of the basic
00:12:59
principles and then keep going back to
00:13:01
that Source whenever you need
00:13:02
to and investing time in that kind of
00:13:06
basic understanding will then make it
00:13:08
possible to understand the papers that
00:13:11
use those
00:13:13
techniques and then finally we will have
00:13:17
specific papers that are highly relevant
00:13:19
to your own particular study now these
00:13:22
might be papers that look at the same or
00:13:24
similar problems but from a different
00:13:27
perspective or perhaps they serve as a
00:13:30
foundation for part of your research if
00:13:32
you base your methods on something
00:13:34
someone else has done and there will
00:13:37
probably be a relatively small number of
00:13:40
these but you need to know them really
00:13:42
well so to recap as a foundation you
00:13:46
need a basic knowledge of the general
00:13:49
Trends problems and debates in your
00:13:51
field and you can get this from the
00:13:53
introductions of a few papers or from a
00:13:56
good published literature review you
00:13:58
need to know the big influential papers
00:14:01
and what they did that were so important
00:14:03
and how that influenced the field and
00:14:06
you need to understand the most
00:14:07
important Concepts and the most commonly
00:14:09
used techniques and you need to know the
00:14:12
most relevant literature for your
00:14:14
specific
00:14:15
project and for all of these you should
00:14:18
build up a collection of key high
00:14:21
quality sources that you can go back to
00:14:24
whenever you need and the key word here
00:14:26
is quality because it's not about the
00:14:29
number of papers in your bibliography
00:14:32
but the standard of the sources you rely
00:14:35
upon and if you understand even a
00:14:38
handful of important highquality papers
00:14:41
it is far better than having extensive
00:14:43
notes or summaries of hundreds of papers
00:14:46
that you don't really
00:14:48
understand and then with that strong
00:14:51
Foundation you can start to look in more
00:14:53
detail at the literature around specific
00:14:57
issues of Interest depending on what you
00:14:59
need at any given
00:15:01
time and the way you should approach
00:15:03
this depends on what you're trying to do
00:15:05
so if you need to do a systematic review
00:15:09
then maybe you need to follow Prisma
00:15:11
guidelines but maybe you're looking for
00:15:14
a solution to a specific problem or
00:15:17
different methods that have been used to
00:15:19
measure something or for the latest
00:15:22
Cutting Edge results and to find these
00:15:25
you might need to follow a more organic
00:15:28
process trying out different keyword
00:15:30
searches or chasing up references in the
00:15:33
bibliographies of other papers or just
00:15:36
by talking to people because
00:15:38
universities are full of people with
00:15:40
useful
00:15:41
expertise but however you approach it
00:15:44
there will be dead end and you will have
00:15:47
to sort through a lot of papers that are
00:15:49
not useful to you or maybe they're just
00:15:51
not useful to you right now but that's
00:15:54
the nature of academic
00:15:57
research now you might be able to
00:16:00
supplement this process using AI but you
00:16:03
cannot delegate the process to AI you
00:16:07
need the ability and the patience to
00:16:10
search and filter the literature
00:16:12
yourself it is a fundamental skill that
00:16:15
you must have it's slow and sometimes
00:16:19
frustrating especially at the start but
00:16:22
you will get better and better at
00:16:23
filtering through the literature to find
00:16:25
the best and most relevant sources as
00:16:28
your knowledge
00:16:30
grows in part two I will talk about how
00:16:33
to turn that knowledge of the literature
00:16:35
into writing but if you have any
00:16:37
questions on this video then please
00:16:38
leave a comment below because I have
00:16:41
skipped over some points quite quickly
00:16:44
and if you'd like to know when I publish
00:16:45
part two please head to my website at
00:16:48
phd. Academy and leave your email so I
00:16:50
can let you know directly when it's
00:16:52
available and If you found this video
00:16:54
useful please share it with somebody who
00:16:57
needs it so that's all for me thank you
00:16:59
so much for listening and I'll see you
00:17:01
in part two