1
00:00:02,657 --> 00:00:04,547
Welcome to Cyber Traps podcast.

2
00:00:04,547 --> 00:00:15,057
I'm Jethro Jones, your host, and I am here again at the Inch 360 conference in beautiful Spokane, Washington on the Gonzaga campus, and I've got Adam Gigstad here with me.

3
00:00:15,057 --> 00:00:16,137
Adam, welcome to the show.

4
00:00:16,137 --> 00:00:17,037
Appreciate you being here.

5
00:00:17,187 --> 00:00:17,877
Yeah, thanks.

6
00:00:17,877 --> 00:00:18,507
Glad to be here.

7
00:00:18,537 --> 00:00:18,837
Yeah.

8
00:00:18,837 --> 00:00:20,727
Why don't you tell us a little bit about what you do.

9
00:00:21,507 --> 00:00:30,537
Yeah, so I have a consultancy that I work with a number of partners and agencies and companies to basically use ai.

10
00:00:30,537 --> 00:00:46,017
We try to use the most cutting edge AI we can find to solve problems that are causing issues in both day-to-day flows, and pushing the envelope on how we can do things better and innovate in ways more quickly.

11
00:00:46,407 --> 00:00:50,907
And, cover more area than we would without those tools.

12
00:00:50,967 --> 00:01:00,517
And so I'm always checking out the latest tools that are being published and iterated on and trying to integrate those into the work of both existing projects.

13
00:01:00,577 --> 00:01:10,147
And I consult with startups to help them kind of get established more quickly and have couple that I am on a leadership team myself.

14
00:01:10,282 --> 00:01:11,392
Yeah, very cool.

15
00:01:11,842 --> 00:01:17,142
The AI space, there's a lot going on there and a lot of fast developments and things that.

16
00:01:17,652 --> 00:01:21,762
You know, it's been around for decades and yet everybody learned about it.

17
00:01:21,762 --> 00:01:27,162
Essentially, it became accessible is how I talk about it in 2022, when chat GPT was released.

18
00:01:27,532 --> 00:01:32,362
But people have known about it and done things with it for a long time before that.

19
00:01:32,362 --> 00:01:40,132
So one of the things that I'm really interested in is everybody's talking about using AI to speed things up and do things faster.

20
00:01:40,642 --> 00:01:43,462
When I talk about this, one of the things I ask is.

21
00:01:44,047 --> 00:01:46,417
Do you even need to do that thing in the first place?

22
00:01:46,567 --> 00:01:50,287
Because sometimes it, you don't need AI to help you do it faster.

23
00:01:50,287 --> 00:01:53,497
You just need to not do it because it's not actually essential.

24
00:01:53,767 --> 00:01:56,107
What first, what's your take on that approach?

25
00:01:57,637 --> 00:02:05,367
Yeah, I think there's a lot of inertia in the processes we have, and so part of it's a people problem of asking anyone do you actually need to do this?

26
00:02:05,367 --> 00:02:08,847
You know, sometimes it's kind of, they get defensive like, well, this is the way we've always done

27
00:02:08,847 --> 00:02:08,937
it.

28
00:02:08,937 --> 00:02:09,207
Yeah,

29
00:02:09,297 --> 00:02:09,567
yeah.

30
00:02:10,077 --> 00:02:12,640
So sometimes I think the first step is.

31
00:02:13,017 --> 00:02:25,187
You know, if you can use AI to get early wins and bring the cost of doing that down that can be helpful for a while you figure out is this whatever it is, something we should actually be doing or doing in this way.

32
00:02:25,237 --> 00:02:25,567
Yeah.

33
00:02:26,017 --> 00:02:33,807
So the other piece that I talk about AI a lot is using it for innovation or to do things that you couldn't do beforehand.

34
00:02:34,177 --> 00:02:40,505
And so what are your thoughts on expanding our capabilities using AI instead of just speeding things up?

35
00:02:40,942 --> 00:02:41,392
yes.

36
00:02:41,402 --> 00:02:42,752
I have a lot of thoughts on this.

37
00:02:42,752 --> 00:02:57,392
As a software engineer that uses a lot of AI tools, the capabilities that I have as someone who's been in software for a while to do things with languages and tools that I have never even heard about sometimes is just extraordinary.

38
00:02:57,642 --> 00:03:03,142
For example, I'm working in Postgres and, oh, there's not a Postgres extension for this thing I'm gonna do.

39
00:03:03,242 --> 00:03:05,612
I'll just use the AI tool to go write one.

40
00:03:05,822 --> 00:03:14,822
And it might be 10 lines and I can review it another AI to make sure it's not like having glaring security issues and deploy it.

41
00:03:14,882 --> 00:03:20,882
And that's a couple hour process versus like, oh, I need to go hire a Postgres extension developer.

42
00:03:21,122 --> 00:03:21,512
Right.

43
00:03:21,812 --> 00:03:22,862
And you're seeing that with.

44
00:03:23,187 --> 00:03:32,657
A lot of people in their development processes who are embracing that kind of workflow to really get those those multipliers and their output by doing things in that way.

45
00:03:33,617 --> 00:03:36,937
/
 
Yeah I think those kinds of examples are incredibly powerful.

46
00:03:37,237 --> 00:03:43,147
And if I may just share one that, that makes a lot of sense for me and is a little more close to home.

47
00:03:43,297 --> 00:03:52,997
My oldest daughter has Down Syndrome and she needed to give a talk and she can write and text and communicate fairly well.

48
00:03:53,057 --> 00:03:57,497
But doing something like a five minute talk is a pretty big ask for her.

49
00:03:58,037 --> 00:04:02,367
And so what we did is I had her write everything out in her.

50
00:04:02,422 --> 00:04:13,702
In her long form and she texted it to me and it was just gibberish long run on sentence it was hard to understand and, but I put it into the AI and I said, here's what she's talking about.

51
00:04:13,702 --> 00:04:19,222
Here's the background, here's her perspective and here's what she wrote so far.

52
00:04:19,612 --> 00:04:20,422
Write a talk.

53
00:04:20,902 --> 00:04:24,677
And so then she, it wrote this talk that actually turned out to be.

54
00:04:25,477 --> 00:04:32,377
Very coherent, really got to the heart of what she was trying to say so that the meaning was there and the value was there.

55
00:04:32,767 --> 00:04:35,937
And then when she went to go deliver it, she can read.

56
00:04:35,937 --> 00:04:41,727
And so she was able to read that talk and do a much better job than she would have otherwise.

57
00:04:42,567 --> 00:04:51,957
Couple months later, she was actually here at Gonzaga talking to a special education class, and I didn't do any of that, and I put her in front of those kids and.

58
00:04:53,127 --> 00:05:00,177
And it was just, it was really terrible because she didn't have a tool that could help her articulate what she was really feeling.

59
00:05:00,567 --> 00:05:04,457
And in the moment she got too flustered and couldn't communicate effectively.

60
00:05:04,757 --> 00:05:17,467
Whereas had we gotten some questions beforehand and written those out and given her responses, then used AI to help her write those in a way that other people could understand, then it would've been a much more beneficial thing.

61
00:05:17,467 --> 00:05:20,062
And when I did that with her, I was like, okay, this is where.

62
00:05:20,787 --> 00:05:25,137
This is really powerful because it's not just like, let's write an email.

63
00:05:25,137 --> 00:05:28,977
We didn't need to write, or let's write an email that you could just write yourself.

64
00:05:29,007 --> 00:05:47,367
This was taking her capability and this paper came out a few months ago that was about cognitive debt, and I call this cognitive equity that
you are putting money in the bank, essentially getting equity on this skillset that you don't ha have access to and how powerful that is.

65
00:05:47,697 --> 00:05:48,837
Any thoughts about that?

66
00:05:49,212 --> 00:05:50,322
I do actually.

67
00:05:50,322 --> 00:06:11,052
I've been thinking about this a lot of, I mean, I think that's adjacent to the idea of using AI for accessibility improvements and so myself, I've historically had some
issues with attention exec and executive function, and so when I've been using software and I've been fortunate to have those skills to build software that helps me offload.

68
00:06:11,652 --> 00:06:15,852
Mental tasks that might have been difficult to juggle into systems that I can create.

69
00:06:15,852 --> 00:06:18,232
It takes a long time for me to create them, but I can do it.

70
00:06:18,292 --> 00:06:25,852
'cause I've been trying at it for a long time, but then when these AI tools came out, it's like, it was immediately like, oh wow, okay, this is even better.

71
00:06:25,852 --> 00:06:27,202
I can do more of this.

72
00:06:27,282 --> 00:06:36,732
Just in terms of managing information both just in, individual sessions, but I'm very interested in like memory systems for organizing large.

73
00:06:37,287 --> 00:06:52,157
Pools of information and I suspect that maybe if you don't have some of those issues, you actually might miss the potential of these tools to help with that for everyone regardless of issues they have.

74
00:06:52,527 --> 00:07:00,207
And so since I've thought about it so much, I feel like I recognize that potential in a lot of these tools right away for situations just like that,

75
00:07:00,697 --> 00:07:00,757
Yeah.

76
00:07:00,802 --> 00:07:01,222
Yeah.

77
00:07:01,287 --> 00:07:09,057
I think that's one of the things that, that my life experience has given me, this perspective that other people may not have.

78
00:07:09,552 --> 00:07:23,352
And I see ways that it can really be empowering and supportive and very helpful and like make things possible for my daughter specifically, that just were not possible before.

79
00:07:23,742 --> 00:07:32,242
And so in, in the field of disability and assistive technology devices, there are a lot of things that open those doors for kids.

80
00:07:32,242 --> 00:07:56,272
Kids who are nonverbal, who can use, like things like Prolo q to be able to say something the AI makes it possible for them to put those things together and then not just say, me, go to the
store, but they can say, I'd like to go to the store so that it's something that is not this robotic, awkward thing, but it actually can bring in some of their personality, some of their voice.

81
00:07:57,577 --> 00:07:58,682
Those things are possible.

82
00:07:58,752 --> 00:08:04,587
They're not perfect yet for sure, but they are certainly possible and can really help people who have disabilities.

83
00:08:04,947 --> 00:08:11,547
And like you said, people who don't have those disabilities or don't have exposure to 'em may not see what a boon that could be.

84
00:08:11,877 --> 00:08:12,177
Yeah.

85
00:08:12,237 --> 00:08:21,097
and actually there was a hackathon style event a couple years ago that I was the lead on a project that was presenting there, and I still think it would be an interesting thing to build.

86
00:08:21,097 --> 00:08:36,977
I didn't end up building it, but even then a few, maybe six months after GPT came out this idea of building a while you're in a
meeting or a conversation even like this, having a visualization that gives you a tree view of the concepts that are being discussed.

87
00:08:37,457 --> 00:08:38,807
As they're happening.

88
00:08:38,867 --> 00:08:43,637
So you can look at that and anchor yourself in the context of the con conversation.

89
00:08:43,637 --> 00:08:48,677
Say you have working memory issues or some other disability, like just that whole body of software.

90
00:08:49,007 --> 00:08:50,567
I think there's so much we can do there.

91
00:08:50,567 --> 00:08:53,687
And even now with the bigger, faster models.

92
00:08:54,027 --> 00:09:00,212
I mean in general I'm optimist about AI and I think we should be using it to help people make their lives better and maybe

93
00:09:01,087 --> 00:09:04,777
Work less and have higher quality interactions with people and their families.

94
00:09:04,777 --> 00:09:08,587
And I think those types of things really fit in that space.

95
00:09:08,857 --> 00:09:09,937
Yeah, for sure.

96
00:09:10,417 --> 00:09:10,717
Yeah.

97
00:09:10,717 --> 00:09:13,087
And we're at the very beginning of this, right?

98
00:09:13,587 --> 00:09:17,967
Like it's, we got a long way to go before we see what these things can really do.

99
00:09:17,967 --> 00:09:25,867
And I imagine the empowerment of my daughter being able to lead a more normal life because of some of these interventions that can help her.

100
00:09:26,017 --> 00:09:27,667
Yeah, it's it's interesting.

101
00:09:27,697 --> 00:09:35,357
One of my colleagues, Graham Moorhead told me that when he got into ai around 2000, so a little while ago

102
00:09:35,437 --> 00:09:36,547
bit before chat, GPT

103
00:09:36,757 --> 00:09:39,967
he like he was getting into it late and that he might've missed it.

104
00:09:40,147 --> 00:09:40,567
Yeah.

105
00:09:40,637 --> 00:09:44,397
And now looking back, it's like, yeah, that was maybe a miscalculation.

106
00:09:44,647 --> 00:09:48,267
But he got a great headstart and con contribute a lot now.

107
00:09:48,317 --> 00:09:54,717
But I think and a few years ago when I was having that same thought of like, oh, I'm doing these other things, but maybe AI seems interesting.

108
00:09:54,767 --> 00:10:01,557
I was actually working at Google and testing out their internal LLMs that hadn't been released yet.

109
00:10:01,617 --> 00:10:05,437
They hadn't started really releasing them for mainstream use, and I was using it.

110
00:10:06,222 --> 00:10:09,942
And just thought, wow, okay, so this is where we're at now.

111
00:10:10,212 --> 00:10:15,492
This is weird and awesome and maybe I should go build in this space.

112
00:10:15,552 --> 00:10:17,142
And so now that's what I'm doing.

113
00:10:17,572 --> 00:10:22,132
But even then I was like, okay, well I don't have a master's in AI and ml.

114
00:10:22,222 --> 00:10:24,172
Am I gonna be able to break into this?

115
00:10:24,482 --> 00:10:26,892
And it turns out that's not a strict requirement

116
00:10:26,997 --> 00:10:27,297
Yeah.

117
00:10:27,717 --> 00:10:41,137
That might be a strict requirement to have a specific job at a specific company, but that's not necessarily a requirement to do things with it, especially with all these commercially available consumer facing tools that do exist.

118
00:10:41,192 --> 00:10:43,502
And you can do a lot of cool stuff with them already.

119
00:10:43,602 --> 00:10:44,958
Yeah, definitely.

120
00:10:44,972 --> 00:10:45,182
yeah.

121
00:10:45,182 --> 00:10:47,102
So Adam, how do people connect with you?

122
00:10:47,102 --> 00:10:48,493
Learn more about what you're doing?

123
00:10:48,977 --> 00:10:57,192
Yeah, so, they can find me on LinkedIn or message me over email adammakingstadtech.com working on a lot of cool projects.

124
00:10:57,192 --> 00:11:01,092
So always glad to connect with other people working on things or who need help with things.

125
00:11:01,312 --> 00:11:03,172
Very passionate about this area.

126
00:11:03,202 --> 00:11:05,412
And so, love to connect with anyone who's interested.

127
00:11:05,742 --> 00:11:06,042
Cool.

128
00:11:06,102 --> 00:11:07,049
Thank you very much.

129
00:11:07,283 --> 00:11:07,657
Thank you.