1
00:00:00,000 --> 00:00:07,980
well this week uh well we're catching up a little bit i i think the amazon reinvent was was uh was

2
00:00:07,980 --> 00:00:14,420
that last week ed or the week yes they reinvented the world again yeah that's right that's right

3
00:00:14,420 --> 00:00:18,760
uh you know i think there's a couple of things that would be fun to highlight there there's all

4
00:00:18,760 --> 00:00:24,460
sorts of stuff that came out and uh there's uh you know i wasn't on my own my other podcast last

5
00:00:24,460 --> 00:00:28,420
week on software to find talk but they they covered things it's it's back there somewhere

6
00:00:28,420 --> 00:00:33,660
that's why i'm pointing uh and so and there's some other good roundups i don't know i don't know if

7
00:00:33,660 --> 00:00:38,360
y'all listen to the uh the cloudcast brian graceley over there he did some roundups but i think there's

8
00:00:38,360 --> 00:00:43,640
two things you've picked out in particular ed that are that are fun uh to look over and uh we'll just

9
00:00:43,640 --> 00:00:48,140
talk about things in general and then hopefully we'll get to uh the exciting adventures with the

10
00:00:48,140 --> 00:00:55,200
ai chatbot thing uh terribly called open wait chat gpt i don't i don't know why they didn't come up

11
00:00:55,200 --> 00:01:01,440
with a better name but that that'll be fun and we'll see what what else but first now now ben you've been

12
00:01:01,440 --> 00:01:07,920
diving into backstage quite a bit i watched one of your uh your your recent videos it's basically a

13
00:01:07,920 --> 00:01:14,920
tour of how the tanzu application platform uses backstage uh as as i recall and i think i think

14
00:01:14,920 --> 00:01:20,620
you gave a a great demo of what we call accelerators but i'm i'm starting to think the rest of the world

15
00:01:20,620 --> 00:01:26,800
calls templates uh which is basically just like here's here's the default setup for your your your project

16
00:01:26,800 --> 00:01:32,520
directories uh and here's kind of like a default configuration that you have things like that and i

17
00:01:32,520 --> 00:01:36,960
really liked the uh the demo of here here's the part i want you to talk a little bit about

18
00:01:36,960 --> 00:01:42,440
we've got this thing called a supply chain in there which which you could also just call uh your build

19
00:01:42,440 --> 00:01:47,140
pipeline but that's cool but i think what makes it a supply chain is that it has a lot of the stuff

20
00:01:47,140 --> 00:01:51,860
that's kind of interesting it has a lot of the uh tracking and governance and audit stuff in there

21
00:01:51,860 --> 00:01:57,360
and that's the part that i think is like so what like if i were to imagine myself not as a programmer

22
00:01:57,360 --> 00:02:03,240
but maybe as a i know maybe we don't use this phrase around tanzu labs people but a project manager

23
00:02:03,240 --> 00:02:08,600
or you know someone who's kind of like checking in on stuff or a governance person like like what

24
00:02:08,600 --> 00:02:12,900
what are the things i have available for me in the supply chain so i can kind of come in and see

25
00:02:12,900 --> 00:02:19,620
what's going on to check in on stuff uh well i think uh probably a headline headline act uh i reckon

26
00:02:19,620 --> 00:02:27,500
for the supply chain is this uh this idea of an s bomb or a bill of materials a software bill of

27
00:02:27,500 --> 00:02:35,480
materials uh which is a really uh cool thing it allows you to take a piece of source code or or an

28
00:02:35,480 --> 00:02:41,440
image you know it depends on how you want to generate your s bombs um but you can you can basically

29
00:02:41,440 --> 00:02:47,020
once you've checked those out against some sort of vulnerability scanner of some kind and there are

30
00:02:47,020 --> 00:02:54,040
many different uh ones of those available on the market of course then it will basically create a

31
00:02:54,040 --> 00:03:00,720
file that that contains all the dependencies inside that particular piece of code and all the um

32
00:03:00,720 --> 00:03:06,740
vulnerabilities that would have been checked for as well so you can use that in auditing and in in sort

33
00:03:06,740 --> 00:03:12,520
of establishing your security posture at any point in time you know you want to do an audit one day

34
00:03:12,520 --> 00:03:17,960
just go do your audit get all your s bombs together go figure out what you've got but in the uh tens of

35
00:03:17,960 --> 00:03:25,040
application platform and in the um in the supply chain we we create those s bombs and we store them away

36
00:03:25,040 --> 00:03:32,320
in a metadata store and then we expose those through a security analysis dashboard so across all the

37
00:03:32,320 --> 00:03:38,360
workloads that you're running on your kubernetes platform you can see okay for this workload has these

38
00:03:38,360 --> 00:03:43,500
vulnerabilities we know about them the safe has these ones that are switched off the false positives has

39
00:03:43,500 --> 00:03:48,960
these ones that are on and that we should be worried about so you can see instantly you know what's going

40
00:03:48,960 --> 00:03:56,200
on across all of the workloads that you've got on that particular platform so it's great and and i get i get

41
00:03:56,200 --> 00:04:03,340
does it will it like update like what's in production or is it just what was put into production does that

42
00:04:03,340 --> 00:04:08,440
distinction make sense like can you have it like auto generate like what's currently running yeah so

43
00:04:08,440 --> 00:04:14,620
because it's based on dependencies if you run a piece of code through the supply chain and it spots a

44
00:04:14,620 --> 00:04:19,420
dependency let's say it's log4j right because there was a big one of those couple of years ago and

45
00:04:19,420 --> 00:04:24,580
this should be log4j so let's everyone's favorite let's just say it was log4j and it notices that

46
00:04:24,580 --> 00:04:30,320
that's a dependency in that particular piece of code and that there's a new vulnerability against that

47
00:04:30,320 --> 00:04:39,180
dependency because the sbomb tracks every piece of code that also shares the same dependency it can then

48
00:04:39,180 --> 00:04:45,980
mark all the other items that also contain the same dependency as being potentially vulnerable so that

49
00:04:45,980 --> 00:04:52,500
you can do something about it so you don't just get information about the one workload as things change

50
00:04:52,500 --> 00:04:57,220
and get updated you get more information about the rest of them as well so it's it's really good

51
00:04:57,220 --> 00:05:03,520
and there's there's still tons that we'll be doing with that as well you know it's um it's something

52
00:05:03,520 --> 00:05:07,860
that we're adding to all the time so i'm really looking forward to the future as well i think there's

53
00:05:07,860 --> 00:05:12,760
a ton of really powerful things we'll be able to do in the future with that and i'm really excited to

54
00:05:12,760 --> 00:05:18,020
see what happens there yeah it's it seems like uh there's a lot of effort going on to do

55
00:05:18,020 --> 00:05:24,320
that the the the first part of it like like uh templating if you will like here's the defaults

56
00:05:24,320 --> 00:05:30,800
of things but then also just i i don't know putting putting a pipeline into your web browser

57
00:05:30,800 --> 00:05:35,880
that's that's a silly way of putting it but like giving giving you like a gui that makes it a lot

58
00:05:35,880 --> 00:05:40,280
easier to kind of like and of course the command line stuff is always there but just kind of like

59
00:05:40,280 --> 00:05:44,320
taking care of a lot of that build stuff like in in the demo you were doing like you know it was

60
00:05:44,320 --> 00:05:49,700
nice it it looks simple but you could just log in and be like all right you know go to the next stage

61
00:05:49,700 --> 00:05:55,320
of this build and then it would do that which which was uh which was kind of fun like i think i was i

62
00:05:55,320 --> 00:06:00,940
was talking with a friend of mine in town last week and oh yeah i remember how he got into this like

63
00:06:00,940 --> 00:06:06,580
somehow he got into being a pci compliance person which like whenever someone tells you that being a

64
00:06:06,580 --> 00:06:12,340
like a u.s retail regulation if you want to accept credit card payments that you have to you have to

65
00:06:12,340 --> 00:06:18,400
use and uh you know as one would ask i was like how did that happen and of course it was kind of

66
00:06:18,400 --> 00:06:23,160
accidental but you know i think maybe in lots of regulations and including like pci stuff there's

67
00:06:23,160 --> 00:06:29,020
something like if you wrote the code you can't deploy the code which i guess makes sense but like

68
00:06:29,020 --> 00:06:33,040
you know you got all sorts of fun options like that if you just uh if you click on things but

69
00:06:33,040 --> 00:06:39,040
it was it was an opportune demo because going to uh a lot of the announcements that amazon reinvent

70
00:06:39,040 --> 00:06:45,160
i think they they also have you know you can see this idea of templates scurrying about in in aws stuff

71
00:06:45,160 --> 00:06:51,020
they have they have like uh they announced this code catalyst thing which is kind of like uh

72
00:06:51,020 --> 00:06:56,800
you know like my i of course i haven't used it because i only make slides but it seemed like another

73
00:06:56,800 --> 00:07:01,300
go at like if you want to if you want to get like started easily on some applications running in a

74
00:07:01,300 --> 00:07:06,520
cloud thing here you go uh it'll do things for you now i don't think that one supports kubernetes

75
00:07:06,520 --> 00:07:12,620
which is fine they have many other options and uh you know there's even there's other op other

76
00:07:12,620 --> 00:07:16,560
things out there like you were just pointing out there's a something called scaffolding or scaffold

77
00:07:16,560 --> 00:07:21,460
i think that's part of uh backstage and other stuff and man there's lots of that templating

78
00:07:21,460 --> 00:07:27,120
going on and of course our own favorite spring boot has had that for a long time but but since you

79
00:07:27,120 --> 00:07:32,680
actually read this stuff ed or at least you type about it i assume you've read it yeah exactly

80
00:07:32,680 --> 00:07:37,580
what's what's what's your take on like and the other thing that was interesting in this area

81
00:07:37,580 --> 00:07:43,760
you know this kind of templating area is the the app composer which is like it's like i i think we

82
00:07:43,760 --> 00:07:49,660
need this new category it's not high code and not low code it's like mid code uh for for like helping

83
00:07:49,660 --> 00:07:54,280
doing uh serverless stuff or something but but what what caught your eye about those two things

84
00:07:54,280 --> 00:07:59,720
well i said they were interesting services they're gap fillers aren't they for amazon right they've

85
00:07:59,720 --> 00:08:04,300
already got a million services they're you know renowned for having whatever it is 17 18 ways of

86
00:08:04,300 --> 00:08:10,260
running containers add another one in probably now with code catalyst um and they all seem i mean you

87
00:08:10,260 --> 00:08:14,520
know my takeaway after i watched everything around and read a bit about it i do a bit of reading but not

88
00:08:14,520 --> 00:08:19,120
a lot um it was just like yours actually templates on steroids i was like well i get this is a

89
00:08:19,120 --> 00:08:23,080
different area of templating right you've always been able to spin up whether it's an eks cluster

90
00:08:23,080 --> 00:08:28,640
choose whether you're using things like fargate how much the the technical bits do you care about

91
00:08:28,640 --> 00:08:33,860
um do you want to go for ec2 and do it yourself do you want eks do you want fargate do you want

92
00:08:33,860 --> 00:08:39,700
serverless completely just forget whatever bits you want to forget and then code catalyst is another

93
00:08:39,700 --> 00:08:45,460
pick your level of abstraction almost you know you're a developer you actually don't really want to

94
00:08:45,460 --> 00:08:49,880
even bother about separating out your project management to your code that right just spin it

95
00:08:49,880 --> 00:08:54,800
all up in code catalyst and it will instantiate all your aws services it'll spin up an environment

96
00:08:54,800 --> 00:08:59,920
it will handle your sort of issue tracking all in one place i mean you could probably do that with

97
00:08:59,920 --> 00:09:05,260
a bunch of disparate aws services already um but this is another way of saying you know what if you

98
00:09:05,260 --> 00:09:09,600
don't care about pieces a b and c i'll do those for you and then you worry about some other pieces

99
00:09:09,600 --> 00:09:16,800
um the old the old fully integrated stack yeah pretty much and you know aws is a building kit and

100
00:09:16,800 --> 00:09:21,340
you just pick how much of it you want to build and how much you don't i mean you know again i've said

101
00:09:21,340 --> 00:09:24,780
there's been a lot of focus from a lot of people on this getting started but i think everyone's

102
00:09:24,780 --> 00:09:29,940
realized in aws probably the worst they've almost provided too many tools and it's like we can do all

103
00:09:29,940 --> 00:09:34,120
this stuff there you go pick from one of these 50 and everyone's like i i don't know which one to

104
00:09:34,120 --> 00:09:39,120
use and which one how to use part three with part four okay pick part one that'll do one two and

105
00:09:39,120 --> 00:09:43,280
three for you which others do you want to add to it and it's just another one of those right we've

106
00:09:43,280 --> 00:09:48,560
got so many services it's a bit overwhelming let's wrap a bunch of them up for you and present it a

107
00:09:48,560 --> 00:09:53,300
little more simply um and that for me was code catalyst so i think all those individual pieces are

108
00:09:53,300 --> 00:09:58,300
there already but not as nicely bundled not as well integrated not as therefore appealing someone who

109
00:09:58,300 --> 00:10:04,600
says just give me an easy button for a load of this stuff and you know it it's similar to some of

110
00:10:04,600 --> 00:10:08,860
what tap does it comes from a different angle doesn't it i wouldn't say it's trying to solve a

111
00:10:08,860 --> 00:10:13,860
different problem it's still making it easier for developers to get code into production but it's so

112
00:10:13,860 --> 00:10:20,160
as as you're as you're looking at these templates and accelerators bin like it seems this tell me

113
00:10:20,160 --> 00:10:24,600
that like like now that we're thinking about it it feels to me like eventually there's just going

114
00:10:24,600 --> 00:10:30,340
to be like some sort of open standard right or like you know just a few of them that are because

115
00:10:30,340 --> 00:10:35,100
and then by the time we have one someone will come up with something new and then the open standard

116
00:10:35,100 --> 00:10:38,980
will be the older thing that everyone's using and then you're trying to move to the new thing but

117
00:10:38,980 --> 00:10:45,360
nonetheless like yeah like how how many different ways could could there possibly be to like set up

118
00:10:45,360 --> 00:10:51,620
your project structure and generate your to use an older word your stubs uh that that you would use

119
00:10:51,620 --> 00:10:58,880
so i i i started to scratch beneath the surface a little bit on the code catalyst stuff and as i as i

120
00:10:58,880 --> 00:11:03,440
went down i did find an open standard as it happens there's uh there's something underneath the dev

121
00:11:03,440 --> 00:11:09,920
environment setup called um uh dev tools i think they've called it which is uh it looks kind of like

122
00:11:09,920 --> 00:11:15,960
a manifest for a developer environment you know so i'm expecting to use this language i'm expecting to

123
00:11:15,960 --> 00:11:22,120
use this idea i'm expecting to use these services i'm expecting to have these containers that sort of

124
00:11:22,120 --> 00:11:28,800
stuff where you can kind of define what goes into a developer environment and that's that on the face

125
00:11:28,800 --> 00:11:34,020
of it sounds like a a good idea you know and and there's been quite a few tools this year that have

126
00:11:34,020 --> 00:11:40,000
sort of tried to edge into this oh we'll create your ide for you but it will be in the cloud i know

127
00:11:40,000 --> 00:11:44,220
google did a similar thing at their google next didn't they they had a similar service to this

128
00:11:44,220 --> 00:11:50,340
where you could go and specify yeah how how you know what sort of developer environment you wanted

129
00:11:50,340 --> 00:11:54,920
what sort of compute should be behind that and then um you know that would appear for you

130
00:11:54,920 --> 00:12:00,260
i'm really interested to find out what happens with traction on these and how far they

131
00:12:00,260 --> 00:12:04,700
you know how how much traction they get with them because that's always been a bit of a sticking point

132
00:12:04,700 --> 00:12:11,900
i think developers kind of like the the local control that they often have over their own

133
00:12:11,900 --> 00:12:19,040
environment and being able to curate um what that environment behaves like and feels like is is

134
00:12:19,040 --> 00:12:24,040
something that they're quite passionate about you know they're like personalization essentially

135
00:12:24,040 --> 00:12:31,560
so i i yeah i don't know but but then we were talking weren't we about accelerators and

136
00:12:31,560 --> 00:12:36,880
templates and that sort of stuff i think about that as different i think about that as project source

137
00:12:36,880 --> 00:12:43,800
code you know where do i get the source code from because in an enterprise there are always

138
00:12:43,800 --> 00:12:51,140
private libraries you know that are unique to that enterprise that you need to use in order to get

139
00:12:51,140 --> 00:12:55,840
your stuff done and the larger the enterprise is the more of those they are the more difficult they are

140
00:12:55,840 --> 00:13:02,900
to bring together into a project you quite often have to go searching for different items and different

141
00:13:02,900 --> 00:13:09,320
bits and bring them all together so the the templating in in things like the accelerators is about

142
00:13:09,320 --> 00:13:18,160
making those patterns and conventions and practices part of the code generation step at the beginning of a

143
00:13:18,160 --> 00:13:25,500
project so um you know would you then take to your ide which traditionally is running on your laptop or on

144
00:13:25,500 --> 00:13:30,400
your on your desktop so um they're sort of slightly different to me i think about them maybe is

145
00:13:30,400 --> 00:13:37,700
slightly separate concerns yeah you know you bring up something uh uh i i have i have in my my uh ever

146
00:13:37,700 --> 00:13:42,300
growing list of homework assignments for myself i think there was there was there was remember you remember

147
00:13:42,300 --> 00:13:48,540
inner sourcing as a concept there was like an inner sourcing conference a little while ago uh that uh

148
00:13:48,540 --> 00:13:54,340
has some new sessions and i've been thinking like all this like developer portal and and kind of talk as

149
00:13:54,340 --> 00:14:01,900
as you were just going over like this idea of like inside an organization having a whole like world of

150
00:14:01,900 --> 00:14:08,700
code or whatever i've lost track of a sentence there but like that's that's what that whole inner

151
00:14:08,700 --> 00:14:15,760
sourcing thing was about back in the late 2000s so maybe it's worth revisiting that as a uh as an idea

152
00:14:15,760 --> 00:14:21,760
but i interrupted you ed what were you saying well hearing you talk about that there's about a lot of

153
00:14:21,760 --> 00:14:27,840
this for me is it's getting that right level of um opinionation should we say so i i tell you

154
00:14:27,840 --> 00:14:32,360
great developers love their laptops and i used to i used to like installing software getting my hands on

155
00:14:32,360 --> 00:14:36,320
the bits because i didn't have to worry about someone else at that point that was just me with my

156
00:14:36,320 --> 00:14:41,420
laptop i can largely do what i want to do that's always going to be right for me and then there's

157
00:14:41,420 --> 00:14:46,900
that mix of saying but whether it's your company or in this case whether it's cloud provider someone

158
00:14:46,900 --> 00:14:51,140
prescribes what's right for you even if it's correct right company says here are our guardrails

159
00:14:51,140 --> 00:14:55,840
here's the things we need for compliance therefore you do these there's a bit of you think can i skip

160
00:14:55,840 --> 00:14:59,820
some of that and just do it through a laptop or whatever so i think that's why people always love

161
00:14:59,820 --> 00:15:04,100
laptop because you you pick and choose exactly what you want without someone saying yeah it'd be

162
00:15:04,100 --> 00:15:08,840
lovely if but you can't because because so whether it's tap whether it's going to be code catalyst

163
00:15:08,840 --> 00:15:14,260
some of the how well they work and how well they're adopted are how much do you constrain people

164
00:15:14,260 --> 00:15:19,020
do you actually if they make it quicker and easier for dev they'll do it right i think the same is

165
00:15:19,020 --> 00:15:23,440
true the cloud idees if they give you exactly what you've got in your laptop but it's quick and easy

166
00:15:23,440 --> 00:15:29,180
reproducible but people will go with it it just hasn't yeah and and if you work for twitter and you

167
00:15:29,180 --> 00:15:33,920
have to print out your code you might not have access to a printer in the cloud so you know

168
00:15:33,920 --> 00:15:40,940
you've got to think about these things right exactly it needs a good export to pdf or docx file

169
00:15:40,940 --> 00:15:46,240
uh option that that would be great you know that's the kind of that's the kind of feature that if you

170
00:15:46,240 --> 00:15:51,720
focus on it too much you'd end up with like like setting the fonts that you use and the uh the you

171
00:15:51,720 --> 00:15:57,500
know justification like alignment you'd really go down a rabbit hole well you don't need to do

172
00:15:57,500 --> 00:16:03,360
what was there anything else from the uh the giant reinvent that uh popped out at you two that seemed

173
00:16:03,360 --> 00:16:08,020
interesting i mean there was there was tons of things as i mentioned earlier but like uh you know

174
00:16:08,020 --> 00:16:15,920
it's always fun just to choose a couple of them yeah genuinely go ahead dad i'd say tons of things but

175
00:16:15,920 --> 00:16:21,220
it went over my head more than in previous years i think they're again moving up the stack right so as

176
00:16:21,220 --> 00:16:26,120
an infrastructure sort of grounded person there was always new instances this at the other and this

177
00:16:26,120 --> 00:16:30,800
year it was a lot more industry related things you know even the keynotes were organized more into

178
00:16:30,800 --> 00:16:36,340
topics and sessions um and therefore there was more of it i listened and glazed over because i know it

179
00:16:36,340 --> 00:16:40,460
runs forever and there's always that oh what's going to be next it's going to be some big headline thing

180
00:16:40,460 --> 00:16:45,600
yeah not so much um which is just maturity i suppose they've been around for so long we kind of know

181
00:16:45,600 --> 00:16:54,380
what to expect so no massive takeaways yeah i interrupted yeah no i missed i missed most of it to be

182
00:16:54,380 --> 00:16:59,560
honest i was preparing for other talks and and things so i didn't i did want to see a little bit more

183
00:16:59,560 --> 00:17:05,180
of it but i i didn't get to but but ed was there anything you know the present climate uh you know

184
00:17:05,180 --> 00:17:11,300
worldwide that is uh you know difficult challenging for a lot of a lot of businesses was there was there

185
00:17:11,300 --> 00:17:18,120
any visible signs of any sort of slowdown in the clothes cloud space or any consolidations or any

186
00:17:18,120 --> 00:17:22,300
anything that would give you an indication that there's you know economic headwinds

187
00:17:22,980 --> 00:17:28,320
not well they talked about that a little bit um and actually bernard golden wrote a good piece

188
00:17:28,320 --> 00:17:33,700
which i did include in my newsletter for those who get that one um which was essentially saying no for

189
00:17:33,700 --> 00:17:39,360
the cloud providers they will knuckle down a bit right and they will no doubt see customers saying we

190
00:17:39,360 --> 00:17:44,500
need to contain costs to set the other but they are all pointing out that they're continuing to build

191
00:17:44,500 --> 00:17:50,240
and they will stop building when the demand goes um and actually all the cloud builders are still

192
00:17:50,240 --> 00:17:56,340
investing heavily so you know this article by bern golden it's not the m1 it's out on the interwebs

193
00:17:56,340 --> 00:17:59,540
can't remember what the title is i'm sure if you search for it you'll find it but it was still

194
00:17:59,540 --> 00:18:03,380
essentially saying growth for the cloud providers is still going to be good they're still building out

195
00:18:03,380 --> 00:18:08,500
cloud regions of billions of dollars in revenue and investing in various countries i think business

196
00:18:08,500 --> 00:18:13,040
will carry on their numbers will dip a little right just from seasonal should we say but um

197
00:18:13,040 --> 00:18:18,140
i think it's business as usual for them well i i don't know if you two have encountered this but i

198
00:18:18,140 --> 00:18:24,180
think in the past month uh i i've encountered uh almost completely contradictory like stories

199
00:18:24,180 --> 00:18:29,680
about about all of your macroeconomic secular headwinds and tailwinds and everything uh you

200
00:18:29,680 --> 00:18:34,060
know one there will be a story about all these tech people being laid off and then there's another

201
00:18:34,060 --> 00:18:39,220
story about how uh you know survey shows uh large companies still can't hire enough tech people

202
00:18:39,220 --> 00:18:42,940
and then and then like what's happening with spending and then i saw another thing that was

203
00:18:42,940 --> 00:18:48,400
like you know cio's plan to keep spending lots of money on their it stuff so so we'll have to see

204
00:18:48,400 --> 00:18:55,760
but i think maybe you know uh those articles we should write a third round of them using this this

205
00:18:55,760 --> 00:19:00,960
brand new chat gpt thing because i think i'm obsessed with this thing on you you've been you've been

206
00:19:00,960 --> 00:19:07,100
hitting this all week i i think i think i just i well i only started yesterday in in my defense was it

207
00:19:07,100 --> 00:19:12,840
only yesterday maybe at the tail end of monday so but it's only wednesday as we record but you know

208
00:19:12,840 --> 00:19:18,580
i think like all of us i generate a lot of text and i also consume a lot of text about the secular

209
00:19:18,580 --> 00:19:23,800
uh economic headwinds that are pushing your tailwinds uh and things like that in the macro

210
00:19:23,800 --> 00:19:30,900
environment uh with currency conversions and uh like i i think a lot of that text could be written

211
00:19:30,900 --> 00:19:36,020
pretty effectively by this this thing now to give you an idea of it you know it's the same thing that

212
00:19:36,020 --> 00:19:41,220
one that was making all those funny pictures over the summer but you log into and it's a chat interface

213
00:19:41,220 --> 00:19:46,720
and as i here here was i'm not going to read it i was thinking we could do a whole episode just

214
00:19:46,720 --> 00:19:52,220
generated by this thing but that would be a little too much but for example uh one of the things i did

215
00:19:52,220 --> 00:19:59,520
yesterday is i asked it to write like an email uh to send to my boss about why okrs are bad and we

216
00:19:59,520 --> 00:20:03,960
shouldn't do them and and i gave it a few little hints here and there like some of the points i wanted

217
00:20:03,960 --> 00:20:09,260
to make and then i said it should be about six paragraphs and it wrote a pretty good like i gave

218
00:20:09,260 --> 00:20:14,720
it the only two you can since it's a chat you can ask it to like uh do things differently so the only

219
00:20:14,720 --> 00:20:21,500
tuning that i did was to ask it to be more casual and also add in a part you know just to ps at the

220
00:20:21,500 --> 00:20:26,460
end that we still needed to approve budget for the hot dog party that's coming up soon and we're cutting

221
00:20:26,460 --> 00:20:33,460
costs by having people bring their own condiments uh but i did that and it was a pretty persuasive uh memo i feel

222
00:20:33,460 --> 00:20:39,860
like i could have maybe sent that to to someone to argue against okrs now the fun thing to do when

223
00:20:39,860 --> 00:20:46,020
you do that is to say like that was great now argue the other side and it'll perfectly put out another

224
00:20:46,020 --> 00:20:52,020
argument about why okrs are good and um i don't know have you all had a chance to uh it's totally free

225
00:20:52,020 --> 00:21:02,960
to use have you all had a chance to uh generate some text ask some questions well i haven't yet i will try it but i

226
00:21:02,960 --> 00:21:09,360
haven't yet and that's because of dali and very you know those sort of uh picture generating tools

227
00:21:09,360 --> 00:21:16,260
i see the pictures that other people generate based on like you know a sentence and then i give it a

228
00:21:16,260 --> 00:21:25,900
sentence and the stuff that comes out is absolute garbage so you're like what what how am i doing

229
00:21:25,900 --> 00:21:32,560
this wrong i don't get it you know you get these beautiful i don't know uh cloud punk style images

230
00:21:32,560 --> 00:21:38,400
or whatever with super fine detail at 4k resolution and i put it in there and it looks like

231
00:21:38,960 --> 00:21:47,680
a toddler's painting with water paints you know yeah complete rubbish so yeah draw me a banana in a

232
00:21:47,680 --> 00:21:53,040
toddler style or whatever you know i i tried using that over the summer and i wanted a picture of

233
00:21:53,040 --> 00:21:59,440
of waylon jennings and nietzsche reading a news article and nietzsche was the the barista and i

234
00:21:59,440 --> 00:22:02,960
didn't quite understand what i was going for but the picture was interesting like you know i wanted

235
00:22:02,960 --> 00:22:08,240
i wanted him serving up some uh some coffee or something but at least waylon jennings had a

236
00:22:08,240 --> 00:22:13,520
cowboy hat and like a cigar a cigarette so you know it was pretty good but you know i think the tech

237
00:22:13,520 --> 00:22:18,880
stuff is a little better i mean i don't i'm sure there's some sort of like how humans work stuff behind

238
00:22:18,880 --> 00:22:24,800
this but like for some reason like a picture has to be better than text maybe like because when

239
00:22:24,800 --> 00:22:32,080
you're reading text you can kind of like fill in the details but a picture like has to be super

240
00:22:32,080 --> 00:22:38,160
abstract like i was just thinking what is the deal with all the kids cartoons nowadays like all the

241
00:22:38,160 --> 00:22:43,680
characters look the same with like giant heads and their midsections are also giant and bulbous like

242
00:22:43,680 --> 00:22:50,400
there's this distinct style that's just like very displeasing uh but with text you know it doesn't

243
00:22:50,400 --> 00:22:56,320
actually have to be that great or even accurate as the stuff generated by this this shows like let

244
00:22:56,320 --> 00:23:02,080
now now on the other hand let me let me read you all another thing i asked it this this is very short

245
00:23:02,080 --> 00:23:08,000
so i'll read you i ask it write a haiku that speculates that ranch dressing existed before the big bang

246
00:23:08,800 --> 00:23:18,640
and it said ranch dressing divine before the big bang comma it was eternal condiment and uh i think that's

247
00:23:18,640 --> 00:23:24,880
pretty good it is surprisingly effective i mean i had to play around and it the stuff i asked it it

248
00:23:24,880 --> 00:23:30,480
generally sort of answered there's a mixture either the stock answer i am an ai blah blah blah and i don't

249
00:23:30,480 --> 00:23:37,040
know about specific frameworks or companies or whatever it might be yet sometimes it does you know whoever it

250
00:23:37,040 --> 00:23:41,680
or spectro cloud did a blog post where they asked it a load of kubernetes questions well for me that's a

251
00:23:41,680 --> 00:23:46,800
specific technology or friend but it came up with very good answers for that exactly that was a good

252
00:23:46,800 --> 00:23:52,640
find it just said no don't know anything about that whatsoever um so it's still quite hit and miss but

253
00:23:52,640 --> 00:23:57,200
it is persuasive when you get to a decent conversation a bit of back and forth which is obviously what it's

254
00:23:57,200 --> 00:24:02,880
designed for um yeah and that's the thing that's what i that's what i haven't experimented with the most is

255
00:24:02,880 --> 00:24:08,320
like actually i think now that you pointed out that interview i think that's something i want to try

256
00:24:08,320 --> 00:24:13,200
next right because there's there's three things i tried with you can tell i'm excited about it because

257
00:24:13,200 --> 00:24:18,000
i just keep talking about it but like one i asked it to tell me what my wife wants for christmas

258
00:24:18,000 --> 00:24:24,160
and it truthfully said oh i don't have access to uh you know private information but you could try

259
00:24:24,160 --> 00:24:31,600
asking her which is just like whoa hold on common sense person it's a very good consultant it depends

260
00:24:31,600 --> 00:24:37,360
is very often it's answer totally totally well there's various things you might want to consider

261
00:24:37,360 --> 00:24:41,520
yeah yeah and and that that showed up in that interview about kubernetes where the guy even

262
00:24:41,520 --> 00:24:46,000
asked it like you always say that and and there was a very good defense of it it was like well

263
00:24:46,000 --> 00:24:51,680
there are many different uh situations and options and then and then the the the next thing i i think

264
00:24:51,680 --> 00:24:58,880
i want to experiment with it is like i don't i i think this would be fine i wrote about this in

265
00:24:58,880 --> 00:25:05,680
my little newsletter this week is like you know my my son is 12 and my daughter is nine and and they

266
00:25:05,680 --> 00:25:12,000
are having to write more and more in school and i feel like you could ask this they could talk with

267
00:25:12,000 --> 00:25:18,080
this thing to like help them learn about writing now the danger is that they just plagiarize it which

268
00:25:18,080 --> 00:25:22,080
that's unethical but it's more a danger of like if you just copy and paste stuff you don't actually

269
00:25:22,080 --> 00:25:27,840
learn how to do it but there's got to be some like way you could just have it kind of teach you the

270
00:25:27,840 --> 00:25:34,000
basics of it right like i don't i don't know what y'all had in um i always get what a private school

271
00:25:34,000 --> 00:25:40,080
versus a public school is in the uk mixed up but like during your your schooling like but you know

272
00:25:40,080 --> 00:25:45,920
in in in u.s schools you learn the five paragraph essay you spend like three years doing this and

273
00:25:45,920 --> 00:25:52,240
like it just like i can't imagine the teachers are like oh goody time to teach the five paragraph

274
00:25:52,240 --> 00:25:58,160
essay right whereas like you you could probably go to go to this thing and say like write me a

275
00:25:58,160 --> 00:26:04,480
five paragraph essay about dinosaurs and like you could have you could have each student generate

276
00:26:04,480 --> 00:26:08,560
that and then in the same way that you might diagram a sentence you could kind of like diagram

277
00:26:08,560 --> 00:26:13,440
the essay and you're kind of like reverse engineering what a five paragraph essay is and that seems like

278
00:26:13,440 --> 00:26:19,680
it would be a fun pedagogical thing uh to do so we'll see if that works well i like the fact that

279
00:26:19,680 --> 00:26:26,960
you can tell chat gpt to write in the style of someone else can't you so i could say write me

280
00:26:28,160 --> 00:26:34,800
a letter in the style of michael Coté and i would get Coté isms uh throughout right that would that

281
00:26:34,800 --> 00:26:40,720
would be you know it says it's corpus uh only goes back to 2021 but thankfully that's still 20 years

282
00:26:40,720 --> 00:26:46,640
of me writing stuff online but this is the thing right i think we should all be very worried because

283
00:26:46,640 --> 00:26:56,560
if you think about it we've seen already deep fakes dali um james earl jones voice in star wars is now

284
00:26:56,560 --> 00:27:06,560
going to be an ai right plus something that'll write scripts so between those things like there's not much

285
00:27:06,560 --> 00:27:11,760
left for us to do is there you could you could you could either be the victim of a deep fake yourself

286
00:27:11,760 --> 00:27:19,760
where they say you know here's michael's face and then here's how he talks and then here's how he you

287
00:27:19,760 --> 00:27:27,200
know uh writes and then make it look like michael said something it's exactly right now now that's

288
00:27:27,200 --> 00:27:33,040
the third thing that i'm i'm thinking about and experimenting with using it is like how does this

289
00:27:34,320 --> 00:27:41,840
how far does this go into the writing process for for writing stuff now now you know i i have i've been

290
00:27:41,840 --> 00:27:47,360
in many roles in my career where you basically just need to like write some stuff and and like you know

291
00:27:47,360 --> 00:27:54,480
for better or worse and so like uh like this time of year uh people always wanted me to write prediction

292
00:27:54,480 --> 00:28:02,320
pieces right and i actually went in and asked it to write me five predictions for 2023 i i i forgot

293
00:28:02,320 --> 00:28:05,920
about this i need to go back and look at it and you know i suggested that use some gartner

294
00:28:05,920 --> 00:28:14,240
and forrester numbers and talk about blockchain and and a few other things and so like and then i also

295
00:28:14,240 --> 00:28:18,640
asked it there's two other things that are there there's there's three other things that are that

296
00:28:18,640 --> 00:28:24,640
are instructive here one one of our colleagues asked it to write a conference abstract about secure

297
00:28:24,640 --> 00:28:29,680
software supply chains and kubernetes and i have to say it was a pretty good start and then i and then

298
00:28:29,680 --> 00:28:35,360
having submitted to a lot of talks what i suggested to him is he asked it to this part see might be

299
00:28:35,360 --> 00:28:42,160
really useful is like i asked him to shorten it down to like 300 uh words because i think if you submit

300
00:28:42,160 --> 00:28:49,280
to like sessionize you're restricted to 300 words in your abstract right and so like it chopped it down

301
00:28:49,280 --> 00:28:55,200
to 300 words and it was still pretty effective right so next time i have like an 800 word session abstract

302
00:28:55,200 --> 00:29:00,480
and the things like you only get you know 350 words i'll maybe i'll go over here and just be like can you

303
00:29:00,480 --> 00:29:06,720
shorten this to 150 words and and we'll see what it says that is good you get that i get that all the

304
00:29:06,720 --> 00:29:12,400
time because every time i submit a piece of content you know internally you got to fill in forms and

305
00:29:12,400 --> 00:29:19,360
stuff and they all exactly size boxes for that yeah and oh and now you're making me think that uh

306
00:29:19,360 --> 00:29:23,680
grammarly on it does it fight with itself you know one artificial intelligence that would be awesome

307
00:29:23,680 --> 00:29:27,360
the sentence structure is incorrect and it automatically rejigs it or something that'd be great

308
00:29:28,240 --> 00:29:35,360
a face off yeah and and so so there's and then two other things i tried one so if you remember in

309
00:29:35,360 --> 00:29:41,360
the the devops platform engineering thing last month there was a big a big uh a big stink up about like

310
00:29:41,360 --> 00:29:46,560
uh some some headlines about devops is dead and it's going to be replaced so i gave it a little bit

311
00:29:46,560 --> 00:29:52,400
of coaching and you know i i posted these on my blog but i asked it to write a piece uh a thought

312
00:29:52,400 --> 00:29:58,080
piece that about devops is dead right and i kind of gave it some ideas and it wrote a a pretty good

313
00:29:58,080 --> 00:30:04,560
thought piece again it will not again but you know it gets it's like a 40 to 70 percent start like you

314
00:30:04,560 --> 00:30:08,400
would have to take it and as you were kind of indicating earlier ben you'd have to go in and like

315
00:30:09,280 --> 00:30:15,920
make it accurate well i saw right i saw that and your your prompts were very good already

316
00:30:16,560 --> 00:30:21,280
yeah yeah right it was already enough in the prompts for you to kind of go totally and where

317
00:30:21,280 --> 00:30:26,000
michael's going with this it's it's it's really well totally and it's kind of like chopping down

318
00:30:26,000 --> 00:30:30,800
it's the at the opposite of chopping down an abstract to 300 words is what you can tell it is like

319
00:30:31,440 --> 00:30:37,680
here's like some ideas can you write 1200 words on that and it'll just go through and like

320
00:30:37,680 --> 00:30:41,120
you know you'd have to polish it a little bit because you can start to detect that this is

321
00:30:41,120 --> 00:30:46,640
basically like uh you know a middle school person writing an article it doesn't quite what's the

322
00:30:46,640 --> 00:30:52,560
thing in the symptoms the simpsons that where it's like in short libya is a land of contrasts

323
00:30:52,560 --> 00:30:58,640
like you know it almost gets to the land of contrast uh problem there but so but then what you did what i

324
00:30:58,640 --> 00:31:04,800
did is you can ask it to write the opposite of that so basically like well let's say if if uh i

325
00:31:04,800 --> 00:31:10,720
thought the devops is not dead and again i i fed it a few ideas and it and writes that and then uh also

326
00:31:10,720 --> 00:31:17,920
along those lines uh i asked it to write uh a piece that was basically like you should uh not build your

327
00:31:17,920 --> 00:31:22,640
own kubernetes versus you should uh you know you should not build your own kubernetes and you should

328
00:31:22,640 --> 00:31:29,840
buy it versus you should totally build your own kubernetes install and so like you know the i i to

329
00:31:29,840 --> 00:31:33,680
this is the first time i think maybe this phrase has been used literally these things just write

330
00:31:33,680 --> 00:31:43,680
themselves like and so like i i feel like like you know if if there's a lot of thought pieces out

331
00:31:43,680 --> 00:31:48,640
there across and i i know especially you probably see these ed when you're scanning the headlines just

332
00:31:49,200 --> 00:31:56,640
across all sorts of random tech news websites out there and man this is like perfect for those like

333
00:31:56,640 --> 00:32:01,920
if if if you were like there's a certain scenario that i'm sure none of us have done where someone

334
00:32:01,920 --> 00:32:07,280
comes to you and had like wants to write on some topic and they might even have some ideas and they

335
00:32:07,280 --> 00:32:12,560
might even have like ghost written something for you and then you know you could go in there and

336
00:32:12,560 --> 00:32:17,520
just you got to go in there and kind of like fix it up but like that whole step of someone having a topic

337
00:32:17,520 --> 00:32:23,120
and like getting the starter for you you could totally do it in like 30 seconds which with just

338
00:32:23,120 --> 00:32:28,240
a little bit of ideas of what you want to say it's uh it's pretty bonkers so to your point ben

339
00:32:28,960 --> 00:32:33,760
uh it's good that you're getting back into programming because i don't know if marketing

340
00:32:33,760 --> 00:32:39,600
is going to exist anymore just we're going to what i'm curious with is this seems to do some coding

341
00:32:39,600 --> 00:32:44,000
doesn't it i saw which i've yes i've got people talking about it saying i've given a code sample

342
00:32:44,000 --> 00:32:48,800
and it fixed it and i was like now how does that work because i understand like the ai and the pattern

343
00:32:48,800 --> 00:32:55,040
matching and just giving it a big database for conversational stuff where does it get coding

344
00:32:55,040 --> 00:33:00,000
knowledge from how much other stuff is in there you know can you give it maths equations chemical

345
00:33:00,000 --> 00:33:06,640
formulas how much sort of knowledge does it have in that respect probably yeah there's no safety in

346
00:33:06,640 --> 00:33:15,360
coding Coté well i i think there you know i i am no i am no ai ml whatever person right but it does seem

347
00:33:15,360 --> 00:33:21,280
like the limitations unless you applied some sort of like william burroughs like cut up technique

348
00:33:21,280 --> 00:33:26,400
the limitation is like you won't be able to come up with something new like because the problem is like

349
00:33:26,400 --> 00:33:30,720
like like i was saying you need the the corpus the body of stuff that it does its predictive writing

350
00:33:30,720 --> 00:33:37,200
on and and so like if you were to for example if we were to make up a new word well this would be a

351
00:33:37,200 --> 00:33:45,200
fun experience like uh let's call it like buternetes which is is just like a new technology you should be

352
00:33:45,200 --> 00:33:51,200
looking into and we ask it to write that essay it probably wouldn't be that useful right like it might have

353
00:33:51,200 --> 00:33:57,280
all the shape of someone promoting a buternetes but like it wouldn't actually say anything right like

354
00:33:57,280 --> 00:34:03,280
there wouldn't be anything useful whereas like if for some reason the spring community and the

355
00:34:03,280 --> 00:34:08,800
kubernetes community decide to merge and become one thing and name the output buternetes right like they

356
00:34:08,800 --> 00:34:14,080
would and they wrote for four years about it it could probably generate some overview of it and talk

357
00:34:14,080 --> 00:34:19,680
about it like this reminds me of another thing for example uh i was thinking like you know ben and i make

358
00:34:19,680 --> 00:34:25,840
these little videos and log videos and i thought like what if i have it generate the script for these

359
00:34:25,840 --> 00:34:32,560
little videos so i asked i did two of them i asked one of them how does a product manager for software

360
00:34:32,560 --> 00:34:39,680
manage a backlog i think it did a pretty good job like it was kind of like a uh how long did it last

361
00:34:39,680 --> 00:34:46,480
maybe like 30 60 seconds and then i asked another one which it it it didn't do good at the prompt i asked it to

362
00:34:46,480 --> 00:34:52,640
define to talk about what a spotify golden path is and it got a little tripped up and said it it talked

363
00:34:52,640 --> 00:34:57,440
about what continuous integration and continuous deployment is but it did it pretty well so like

364
00:34:57,440 --> 00:35:03,120
you could just go in and remove the spotify stuff but then i took those and i wasn't pretending like i

365
00:35:03,120 --> 00:35:06,960
wasn't doing this i was very upfront about it so i could be ethical but then i just went and made two of

366
00:35:06,960 --> 00:35:13,040
those little youtube shorts and there you go so you could imagine again this is why our marketing jobs

367
00:35:13,040 --> 00:35:18,880
we got to start worrying ben like i don't know or or maybe this is maybe i'll get to the last point

368
00:35:18,880 --> 00:35:25,920
the jevons paradox thing but like you know if if you could pump out like 50 of these videos a month

369
00:35:25,920 --> 00:35:30,960
you could probably start to tune and like think about the right kind of content and the reason you could do it is like

370
00:35:30,960 --> 00:35:37,680
like it only took me i don't know maybe 30 minutes to do that video like the one about each of those

371
00:35:37,680 --> 00:35:43,920
videos which as you know ben is very quick for for a video like it's you know of course you've got to

372
00:35:43,920 --> 00:35:49,840
do like the one the one take thing and then use like descript or something to chop it up or whatever

373
00:35:49,840 --> 00:35:55,360
you even have to put your pants on and everything don't you you have to well not not with the current

374
00:35:55,360 --> 00:36:00,240
view that i have here i could just i could just have yeah but you do yours outside i'm pretty sure

375
00:36:00,240 --> 00:36:06,640
you're not going out yeah yeah and and so so like so that that's all as you can tell i think it's all

376
00:36:06,640 --> 00:36:10,640
pretty thrilling but it was also making me think and this is like what i wrote up in the newsletter is

377
00:36:10,640 --> 00:36:16,560
like you know you know the idea of insight reliability engineering of developer of toil right like

378
00:36:16,560 --> 00:36:23,440
it's repetitive manual tasks that don't add any value and an sre person should spend i think it's

379
00:36:23,440 --> 00:36:27,280
like 50 of their time i would imagine if you're doing your platform stuff initially it's probably

380
00:36:27,280 --> 00:36:34,320
like 80 of your time but you should always spend a lot of time automating toil and as you go through

381
00:36:34,320 --> 00:36:39,040
like what a lot of these things what this thing is good at is making me realize it's really good at just like

382
00:36:39,680 --> 00:36:45,440
office worker toil and even marketing toil right like it's kind of a toil thing to come up with

383
00:36:45,440 --> 00:36:50,880
like your first pass at some copy and it's a little bit of toil if you want to make a video defining a

384
00:36:50,880 --> 00:36:57,360
backlog to like think about how you would phrase it and then it's even more toil to like ask you

385
00:36:57,360 --> 00:37:02,960
know to do budget justification for a hot dog party right like and and like all of and i was thinking

386
00:37:02,960 --> 00:37:07,280
like back when i was in corporate strategy like i would have these drive-by strategy assignments

387
00:37:07,280 --> 00:37:13,120
where like someone would come by and be like oh one of the uh one of the board members thinks that it

388
00:37:13,120 --> 00:37:19,120
would be a good idea for us to get into uh the the software development tools business so i'm gonna

389
00:37:19,120 --> 00:37:24,080
need can you write me a memo to send them totally with my name on it instead of me the person writing

390
00:37:24,080 --> 00:37:31,120
it uh like a memo talking about what we should do here and it's pretty good at doing that too you can just

391
00:37:31,120 --> 00:37:36,160
like i mean it just to to i won't use the word since this is family friendly but the um

392
00:37:37,040 --> 00:37:43,040
the uh uh you know the the the cow excrement word like if it's that kind of that kind of work

393
00:37:43,040 --> 00:37:48,880
like you could totally eliminate a ton of that work which then to use a james water notion you can

394
00:37:48,880 --> 00:37:55,040
go up the value line you don't really have to worry about like all that low level uh cow waste stuff

395
00:37:55,680 --> 00:38:01,040
so i think that's a good way for us us uh knowledge workers to start experimenting with it to to focus

396
00:38:01,040 --> 00:38:06,720
on how can we eliminate knowledge worker toil uh that that we need to do kind of like your grammarly

397
00:38:06,720 --> 00:38:11,760
thing maybe what we should do is the next time we have to file expenses we could just say we can get

398
00:38:11,760 --> 00:38:18,160
the concur ai to talk with this ai and just kind of like tell us the result instead of having to do

399
00:38:18,160 --> 00:38:29,840
anything you you owe 15 000 pounds yeah maybe we wouldn't like that result oh that that's well

400
00:38:29,840 --> 00:38:35,920
because what you would have to say is and don't take no for an answer and then and then maybe it

401
00:38:35,920 --> 00:38:44,560
would just burn out that's when you ask it to argue the other way maybe you could get it to onboard

402
00:38:44,560 --> 00:38:51,440
suppliers maybe though yeah no totally that i think that's you know and and i was trying to think of

403
00:38:51,440 --> 00:38:55,920
the commercialization things right and that's the part that like would be valuable is you can't really

404
00:38:55,920 --> 00:39:04,960
like you couldn't for example like let's say you had to do gdpr verification right like for a i i don't

405
00:39:04,960 --> 00:39:11,600
know theoretically at some company you might have to do uh gdpr verification for a new third-party vendor

406
00:39:11,600 --> 00:39:17,520
and so you could kind of like feed it your company's policy and then feed it a bunch of stuff from the

407
00:39:17,520 --> 00:39:23,040
third-party vendor and basically just be like can you fill this out for me and do a good job in the

408
00:39:23,040 --> 00:39:29,440
style of me and then like like that would be amazing and then but see like you could think that's terrible

409
00:39:29,440 --> 00:39:34,720
but then the whole point of the the adopting the mentality of the toil thing is like well instead of me

410
00:39:34,720 --> 00:39:40,320
even doing it the person who's doing this audit should just do that on their own like why am i

411
00:39:40,320 --> 00:39:45,520
even involved in this right like just you know get that person doing it can do it no need for like a

412
00:39:45,520 --> 00:39:51,760
help desk and so anyhow maybe there's hope for a lot of that stuff we'll see next time i have to argue

413
00:39:51,760 --> 00:39:56,960
something in a bureaucracy i i might i might try to use it there's not much hope if you work on the help desk

414
00:39:56,960 --> 00:40:03,760
well but that but see this is this gets to the jevons paradox thing is like what you see when

415
00:40:03,760 --> 00:40:08,720
you automate things is that like it's not that people and and this is this is the there's two

416
00:40:08,720 --> 00:40:13,680
there's a there's a uh capitalist friendly and then a worker not friendly aspect of this is like

417
00:40:14,720 --> 00:40:19,040
and i think jevons paradox it comes comes from one of your uh one of your countrymen way back in

418
00:40:19,040 --> 00:40:25,520
the industrial revolution is is they realized that the more efficient you make a factory uh it causes

419
00:40:25,520 --> 00:40:32,480
more energy use so which is because like you free up time so it means like you create more waistcoats

420
00:40:32,480 --> 00:40:38,480
or whatever right like you don't there's a reason we have the phrase do more with less we don't have

421
00:40:38,480 --> 00:40:43,920
the phrase do less with less right like you always want you want to do more so like that means

422
00:40:43,920 --> 00:40:47,680
that like if you automate this stuff then again maybe we can focus on things that are more

423
00:40:47,680 --> 00:40:52,640
interesting but then the downside of that is jevons paradox doesn't say that when we automate things

424
00:40:52,640 --> 00:40:57,600
now we get to have like a 20 hour work week that's really what we would want that maybe that's Coté's

425
00:40:57,600 --> 00:41:04,320
rejoiner to jevons paradox is like i i would like this to mean we have we get to work less as we're more

426
00:41:04,320 --> 00:41:14,320
efficient i i love how excited by this you are i really do i find that really uh really good but to me

427
00:41:14,320 --> 00:41:19,680
it feels like a jurassic park moment like oh yes oh look we can make real dinosaurs

428
00:41:21,440 --> 00:41:28,960
that's gonna end well let's do that i i can't help but think about the other uses and if you follow

429
00:41:28,960 --> 00:41:33,760
podcasts like i don't know dark net diaries or something like that you know where where technology

430
00:41:33,760 --> 00:41:41,280
has been subverted in so many ways so many times you can't help but be concerned about that there are

431
00:41:41,280 --> 00:41:47,280
new laws coming into the uk around the use of tools like um deepfake for example you know there's talk

432
00:41:47,280 --> 00:41:52,720
about new new laws around that because they are being abused for things that they shouldn't be

433
00:41:52,720 --> 00:42:00,640
used for or that we wouldn't consider them a valid use for so um yeah i i can't help but can also

434
00:42:00,640 --> 00:42:06,720
have a slight maybe it's because i'm british maybe that's what it is i i think i think i think i mean it's

435
00:42:06,720 --> 00:42:11,040
it's it's like any other like big technology this isn't a dismissive thing but yeah i agree there's

436
00:42:11,040 --> 00:42:16,800
all sorts of like crazy bad stuff you could do but maybe we should finish out yeah unless what are

437
00:42:16,800 --> 00:42:22,080
your what are your last thoughts ed are you going to uh are you going to start using this for uh for

438
00:42:22,080 --> 00:42:31,200
something gpt what it thinks i should say i don't know i'll i'll give i'll give you one one more

439
00:42:31,200 --> 00:42:38,080
example in favor of this right an example of toil is i posted this as well over on my my my weblog

440
00:42:38,720 --> 00:42:43,760
at Coté.io but we we asked recently to get our kids out of school over here in the netherlands for

441
00:42:43,760 --> 00:42:49,920
thanksgiving uh you know because we're americans and and i this was in the past but i thought like

442
00:42:49,920 --> 00:42:55,120
what if we need to do this again so i asked it to write a letter asking the school and it was like uh

443
00:42:56,000 --> 00:43:01,840
it was pretty perfect like it had it had the right tone that you would be asking an authority figure

444
00:43:01,840 --> 00:43:06,240
and it like kind of asked in the right way and it even had the thing of like i understand this will

445
00:43:06,240 --> 00:43:10,960
take some consideration like that that kind of stuff in there and it was just like wow i would have

446
00:43:10,960 --> 00:43:16,480
never thought to write it like that like and this is like pretty good and then i bet also

447
00:43:17,760 --> 00:43:23,040
if they replied to your email you could put it back into the chat and be like can you reply to this

448
00:43:23,040 --> 00:43:28,080
email right like unfortunately like the the tier that it has isn't persistent enough that it would

449
00:43:28,080 --> 00:43:32,720
it lasts that long anyways but the last thing that i did you know you can also do funny stuff like the

450
00:43:32,720 --> 00:43:38,400
ranch dressing is is i asked it to come up with 12 ideas for blog posts on monster truck models

451
00:43:39,200 --> 00:43:44,000
uh and and and it's it's a blog it's a blog that that covers monster truck models but occasionally

452
00:43:44,000 --> 00:43:49,360
writes about microsoft clippy and i'll just get i won't read all of them but it got it got pretty

453
00:43:49,360 --> 00:43:54,960
good it was it was building the perfect monster truck model diorama that's that's a good idea and then

454
00:43:54,960 --> 00:44:00,400
you got uh the top monster truck models competition to check out also good i like that one and then

455
00:44:00,400 --> 00:44:06,160
finally microsoft clippy meets the world of monster truck models interesting

456
00:44:08,560 --> 00:44:14,400
well on that note i i'm gonna go ask uh maybe go ask this thing to write some show notes for us

457
00:44:14,400 --> 00:44:19,200
uh probably about 10 000 words worth we'll see how many words it can go up to i'm gonna go and start a

458
00:44:19,200 --> 00:44:28,000
text blog may or may not be me and uh with that i can't i can't wait until they add voice onto this

459
00:44:28,000 --> 00:44:36,000
if you added voice recognition and then speech you know be like a really useful google voice thingy

460
00:44:36,000 --> 00:44:42,160
right alexa write my reinvent keynote yeah it's like alexa's useless right but if it could do this

461
00:44:42,160 --> 00:44:49,280
this would be amazing i mean once once you can feed your own stuff into it i mean i mean that's how

462
00:44:49,280 --> 00:44:53,040
that's got to be one of the ways they would commercialize it right it also it also like just

463
00:44:53,040 --> 00:44:58,640
to trail off it makes you wonder like wow maybe all that watson stuff was just before its time

464
00:44:58,640 --> 00:45:04,960
like this this could actually uh i don't know we'll see next week if uh if we're still here or not or if uh

465
00:45:04,960 --> 00:45:10,080
if if uh microsoft clippy has taken over max headroom yeah be max headroom yeah i was gonna say we'll

466
00:45:10,080 --> 00:45:15,760
come max headroom yeah i love that tv show i found i found where you can uh they have all those in

467
00:45:15,760 --> 00:45:22,560
youtube you can uh go peruse them if you want it's kind of it's impossible to buy them gpt max headroom

468
00:45:22,560 --> 00:45:28,160
will live again oh yeah i gotta get i need to try doing everything in the style of max headroom

469
00:45:29,120 --> 00:45:33,280
yeah and folks if you uh watch us on youtube next week it may be three max headrooms

470
00:45:37,920 --> 00:45:43,520
all right well if you if you want to see the show notes to to all of this i'll try to put in some

471
00:45:43,520 --> 00:45:49,600
links to to uh well obviously the the aws stuff and maybe some of those examples of uh output that

472
00:45:49,600 --> 00:45:56,720
i talked about you can go to tonsutalk.com and uh just find this episode uh and and as always we uh

473
00:45:57,280 --> 00:46:01,680
we're not here every week but we usually also if you're listening to the podcast we stream it live

474
00:46:01,680 --> 00:46:07,920
there was some fun commentary uh in in the the uh the chat there and uh we'll see everyone next time

475
00:46:07,920 --> 00:46:14,400
bye bye all right well i haven't ended the broadcast yet but that that was good

476
00:46:15,680 --> 00:46:22,560
and uh just in time yep like i was saying in the style of max headroom we'll see what we'll see what

477
00:46:22,560 --> 00:46:28,720
it's like i think i think oh you know another thing i tried that was fun is i i asked it you were kind

478
00:46:28,720 --> 00:46:34,080
of not alluding but you're getting close to this bin i asked it to uh write a favorable review of the

479
00:46:34,080 --> 00:46:41,200
columns i wrote at the register between 2012 and 2018 and it was pretty good although it exp it also

480
00:46:41,200 --> 00:46:46,960
exposed it also exposed one of the things you know you you were talking about is i then asked it to

481
00:46:46,960 --> 00:46:53,280
specific i mean to reference specific columns and so it used it used two column you know titles for

482
00:46:53,280 --> 00:46:57,920
column names and i was like is that a column name that i used and so i went and searched and it just

483
00:46:57,920 --> 00:47:04,320
said totally made up names for columns like not not a title i used at all so you gotta you gotta check

484
00:47:04,320 --> 00:47:15,120
into that i think i think ben is still speechless he's like this uh you're you're unleashing a torrent

485
00:47:15,120 --> 00:47:23,680
of problems that's yeah that's what's going on here honestly yeah i well yeah yeah thanks to

486
00:47:23,680 --> 00:47:29,360
everyone who uh who was on the chat as well as a few tindall was on the chat today uh tindall uh

487
00:47:29,360 --> 00:47:35,680
is uh everyone's talking about chat gpt and i'm here building crud and being totally out of the loop

488
00:47:35,680 --> 00:47:42,240
absolutely you haven't got up to cqrs yet uh maybe maybe you have a look at that wow what a joke

489
00:47:45,120 --> 00:47:52,080
one for the deaths and uh ed someone also said are you also uh house sharing with all those other

490
00:47:52,080 --> 00:47:55,520
people are you who also i do i have a very generous house sharing policy

491
00:47:59,040 --> 00:48:03,840
i will find i would so that's what i need next week is i need to get a dali background obviously

492
00:48:03,840 --> 00:48:08,480
something that is genuinely unique yes that would be awesome yeah maybe you could ask

493
00:48:08,480 --> 00:48:14,400
unique podcast background and see you you could you could ask the chat thing to give you a prompt to

494
00:48:14,400 --> 00:48:20,160
give dali that would come up with a unique zoom background and then and then see if you can chain

495
00:48:20,160 --> 00:48:26,080
that prompt together and visualize and it should be it should include the dark corners of your mind as

496
00:48:26,080 --> 00:48:30,640
well i think you know not just the light ones but the dark ones as well should be a little bit of

497
00:48:30,640 --> 00:48:34,000
i don't know freddy krueger in there you know some sort of nightmarish there you go

498
00:48:34,000 --> 00:48:40,800
like like do it something something like and and make subtle allusions to the dark corners of my mind

499
00:48:40,800 --> 00:48:52,080
palace brilliant yeah all right well let's put it that way i'm gonna wrap up the broadcast here

500
00:48:52,080 --> 00:48:56,000
we'll see the streamers next time bye bye bye again