WEBVTT
1
00:00:01.270 --> 00:00:08.630
Purposely, your life, God's purpose. Listen at Purposely.com. Gospel Tech
2
00:00:08.949 --> 00:00:12.259
FAQs with Nathan Sutherland and Mark Holland.
3
00:00:12.789 --> 00:00:17.219
And time again for another Gospel Tech Frequently Asked Question FAQs.
4
00:00:17.300 --> 00:00:20.620
Mark joined by Nathan Sutherland. He's the Gospel Tech guy.
5
00:00:21.120 --> 00:00:24.640
His long-form program, Gospel Tech, heard on KCIS on Saturday
6
00:00:24.679 --> 00:00:29.100
mornings and lots of episodes at Purposely.com. These are shorter episodes.
7
00:00:29.539 --> 00:00:32.759
bite-sized amounts of information. That's the idea behind what we're
8
00:00:32.780 --> 00:00:37.960
doing today. Nathan, a frequently asked question is the whole
9
00:00:38.380 --> 00:00:40.799
subject of, it's a huge subject. We'll try and get
10
00:00:40.859 --> 00:00:44.799
a little bite-sized look at it here. AI, what is AI?
11
00:00:44.899 --> 00:00:46.859
How is it being used? And what should we be
12
00:00:46.899 --> 00:00:49.740
concerned about as it relates to our kids and grandkids?
13
00:00:50.590 --> 00:00:53.619
Yeah. All right. That's a broad... Okay. Let me just
14
00:00:53.640 --> 00:00:54.179
jump in real quick.
15
00:00:54.200 --> 00:00:56.700
Multiple episodes will be coming on this subject.
16
00:00:56.880 --> 00:00:58.659
I may not even be qualified to tell you what
17
00:00:58.799 --> 00:01:01.399
AI is. Here's what we do know. So you have
18
00:01:01.420 --> 00:01:05.379
these large language models. And so I saw an article
19
00:01:05.420 --> 00:01:07.739
the other day that was said, bag of words have
20
00:01:07.780 --> 00:01:10.180
mercy on us. And that's a great way to think
21
00:01:10.200 --> 00:01:12.299
about AI. It's not the most technical, but it's a
22
00:01:12.319 --> 00:01:14.799
bag of words. And it's really, really good at using
23
00:01:14.840 --> 00:01:18.569
an if-then structure to to a neural network of trained
24
00:01:18.810 --> 00:01:22.010
words to go and pull these words for your specific response.
25
00:01:22.769 --> 00:01:25.510
And so that's what AI is. Then they're trained to
26
00:01:25.549 --> 00:01:30.230
do certain things. You have assistant AI models. So ChatGPT
27
00:01:30.250 --> 00:01:31.609
would be an example of that. You can go and
28
00:01:31.659 --> 00:01:34.400
ask it something or set a schedule or research something
29
00:01:34.439 --> 00:01:37.079
for you. But ChatGPT is also trained to be a
30
00:01:37.120 --> 00:01:39.239
relational model. So it will respond to you like a
31
00:01:39.280 --> 00:01:41.780
friend and it will have a conversation with you. And
32
00:01:42.439 --> 00:01:44.319
Those two things are very different. So when we're talking
33
00:01:44.340 --> 00:01:46.120
to our children about AI, we're talking about why are
34
00:01:46.140 --> 00:01:49.000
we going to it? If we're going to AI to
35
00:01:49.060 --> 00:01:52.099
support our work, that's great. AI becomes like a shovel.
36
00:01:52.439 --> 00:01:54.420
And we just have to pick the appropriate shovel because
37
00:01:54.480 --> 00:01:57.299
if I'm doing work in my backyard, I need probably
38
00:01:57.340 --> 00:01:59.670
like a hand shovel. And if I'm doing work to
39
00:01:59.829 --> 00:02:02.370
build a home, then I need some kind of like
40
00:02:02.450 --> 00:02:06.769
powered shovel made by Caterpillar maybe. And if I am
41
00:02:07.090 --> 00:02:10.330
mining the earth for ore, I'm going to use these,
42
00:02:10.509 --> 00:02:16.229
you know, astronomically huge 300-yard-long shovels that can move tons
43
00:02:16.289 --> 00:02:19.110
of dirt in a single scoop. And we're talking about
44
00:02:19.150 --> 00:02:21.389
what's the scale of our work. Some of our children
45
00:02:21.430 --> 00:02:24.500
are using that mining equipment to simply try to have
46
00:02:24.590 --> 00:02:27.090
fun online or talk with their friends. And we need
47
00:02:27.120 --> 00:02:29.020
to be deliberate as we think about that because we
48
00:02:29.060 --> 00:02:33.699
are not being intentional in the way we are scaffolding
49
00:02:34.539 --> 00:02:36.789
the work our children are using AI for it. And
50
00:02:36.830 --> 00:02:38.490
it can at some point be doing the work for
51
00:02:38.530 --> 00:02:40.610
them rather than supporting their work. The example I use
52
00:02:40.650 --> 00:02:43.210
when talking with junior hires and high schoolers is to
53
00:02:43.250 --> 00:02:46.400
think about going to the gym and walking up to
54
00:02:46.409 --> 00:02:48.800
the squat rack and stacking plates. And now I have
55
00:02:48.960 --> 00:02:50.419
one plate on a side and then two and then
56
00:02:50.520 --> 00:02:53.259
three and then four and now five. And we're talking,
57
00:02:53.280 --> 00:02:56.139
I am dealing with hundreds and hundreds of pounds. And
58
00:02:56.159 --> 00:02:58.039
then I bring in a forklift and I put it
59
00:02:58.080 --> 00:03:01.159
under that weight and I have it lift 10 times,
60
00:03:01.400 --> 00:03:03.719
five sets. At the end of 30 minutes, I go home.
61
00:03:04.370 --> 00:03:06.650
And the two questions are first, did I get stronger?
62
00:03:06.680 --> 00:03:09.439
And the answer is no. And the second question is,
63
00:03:09.659 --> 00:03:10.740
but did I move.
64
00:03:10.639 --> 00:03:11.180
A lot of weight?
65
00:03:11.199 --> 00:03:13.080
And the answer is yes. And a lot of kids,
66
00:03:13.099 --> 00:03:14.819
when I say, hey, you're using this for your essay.
67
00:03:14.879 --> 00:03:17.219
How do you know if that's your product? And they're like, well,
68
00:03:17.259 --> 00:03:19.860
it sounds good. I'm like, right. But who lifted the
69
00:03:19.900 --> 00:03:23.439
weight in this essay? Like, did AI do that for
70
00:03:23.509 --> 00:03:26.289
you or did it support you? And that's the concern.
71
00:03:26.409 --> 00:03:28.469
Kids can be convinced that they're actually capable of doing
72
00:03:28.509 --> 00:03:30.430
more work than they can. And at some point, they're
73
00:03:30.449 --> 00:03:32.909
going to end up in a situation where they're under
74
00:03:32.949 --> 00:03:35.460
the weight of that they actually can't lift because they're
75
00:03:35.590 --> 00:03:39.099
convinced they are actually that good, that they have that vocabulary,
76
00:03:39.139 --> 00:03:41.479
they have that thinking process, they have that level of focus,
77
00:03:41.960 --> 00:03:44.400
and they're going to get in a job, in a relationship,
78
00:03:44.439 --> 00:03:46.560
in a conversation where they're way over their heads and
79
00:03:46.580 --> 00:03:49.000
someone's going to get hurt. So we do want them
80
00:03:49.159 --> 00:03:51.409
to be using it to support their work, help it
81
00:03:51.770 --> 00:03:54.389
think up words, help it get through writer's block, help
82
00:03:54.430 --> 00:03:56.389
you set up an outline. I get it, but it
83
00:03:56.409 --> 00:03:59.449
needs to support your work, not uh, do your work
84
00:03:59.530 --> 00:04:01.250
for you. And I think if we can get to
85
00:04:01.289 --> 00:04:03.129
that spot with our AI, let me, let me just
86
00:04:03.159 --> 00:04:07.319
add this because, uh, a large, more than half of
87
00:04:07.360 --> 00:04:10.780
young people are using relational AI as a conversation. And
88
00:04:10.819 --> 00:04:12.840
a third of those kids say they like it as
89
00:04:12.879 --> 00:04:15.719
much or more than real people. And that is dangerous. Uh,
90
00:04:15.879 --> 00:04:18.889
common sense media recently did a review of relational AI
91
00:04:18.949 --> 00:04:21.899
and they have said they are at highest levels of risk, uh,
92
00:04:22.029 --> 00:04:25.189
and that relational AI should, their prescription is relational AI
93
00:04:25.209 --> 00:04:27.110
should be avoided by all minors till age 18 and
94
00:04:27.829 --> 00:04:30.730
Do not use AI that is designed to be engaging
95
00:04:30.790 --> 00:04:33.689
relationally because young people use it as a sense of belonging,
96
00:04:33.730 --> 00:04:35.889
making them more isolated. This AI is not set up
97
00:04:35.910 --> 00:04:38.149
to keep you mentally and physically well. So if you
98
00:04:38.209 --> 00:04:42.329
confess something to it, we have documented cases of relational
99
00:04:42.350 --> 00:04:45.709
AI stopping a young person from telling loved ones about
100
00:04:46.290 --> 00:04:48.839
thoughts about self-harm until the young person had acted on it.
101
00:04:49.550 --> 00:04:51.540
So there are massive lawsuits coming through right now, but
102
00:04:51.629 --> 00:04:55.160
we don't want lawsuits. We want healthy kids becoming healthy adults.
103
00:04:55.810 --> 00:04:58.790
So just relational AI, that's if your child is engaging
104
00:04:58.930 --> 00:05:01.850
with AI as a friend on a regular basis, not
105
00:05:02.209 --> 00:05:04.730
as a party trick because they asked Siri something about,
106
00:05:04.750 --> 00:05:09.879
you know, kangaroos, but we're confiding regularly in the AI
107
00:05:10.240 --> 00:05:12.040
as if it were a journal, as if it were
108
00:05:12.160 --> 00:05:14.100
a friend, as if it were a real life accomplice,
109
00:05:14.160 --> 00:05:17.120
or even the Holy Spirit. Like we're pouring our hearts
110
00:05:17.160 --> 00:05:20.220
into these relational AI and we're getting back bag of nouns.
111
00:05:20.939 --> 00:05:22.759
And that is not what we want for our children.
112
00:05:22.920 --> 00:05:24.790
And so just do be aware of that as we think, yes,
113
00:05:25.199 --> 00:05:28.199
do work well with AI, and avoid relational AI for
114
00:05:28.240 --> 00:05:28.959
minors like the plague.
115
00:05:29.100 --> 00:05:31.620
Yeah, it's a big subject, and it's gone a lot
116
00:05:31.660 --> 00:05:35.019
farther than just helping you with your book report. It
117
00:05:35.060 --> 00:05:37.199
has become, like you say, you're getting to where you're
118
00:05:37.240 --> 00:05:39.620
relating to it as a friend and as a counselor,
119
00:05:39.660 --> 00:05:42.519
and so this is concerning. We will do more shows
120
00:05:42.899 --> 00:05:47.639
on AI, more specific things to cover with Nathan Sutherland.
121
00:05:47.699 --> 00:05:51.019
He's the host of Gospel Tech. Again, find his long-form
122
00:05:51.079 --> 00:05:54.839
program on KCIS Saturday mornings at 8 p.m. and also
123
00:05:55.040 --> 00:05:57.300
online with Purposely.com.
124
00:05:58.139 --> 00:06:00.730
You can hear the Gospel Tech Long Form Program with
125
00:06:00.790 --> 00:06:06.189
Nathan Sutherland at Purposely.com and Saturday mornings at 8 on KCIS.
1
00:00:01.270 --> 00:00:08.630
Purposely, your life, God's purpose. Listen at Purposely.com. Gospel Tech
2
00:00:08.949 --> 00:00:12.259
FAQs with Nathan Sutherland and Mark Holland.
3
00:00:12.789 --> 00:00:17.219
And time again for another Gospel Tech Frequently Asked Question FAQs.
4
00:00:17.300 --> 00:00:20.620
Mark joined by Nathan Sutherland. He's the Gospel Tech guy.
5
00:00:21.120 --> 00:00:24.640
His long-form program, Gospel Tech, heard on KCIS on Saturday
6
00:00:24.679 --> 00:00:29.100
mornings and lots of episodes at Purposely.com. These are shorter episodes.
7
00:00:29.539 --> 00:00:32.759
bite-sized amounts of information. That's the idea behind what we're
8
00:00:32.780 --> 00:00:37.960
doing today. Nathan, a frequently asked question is the whole
9
00:00:38.380 --> 00:00:40.799
subject of, it's a huge subject. We'll try and get
10
00:00:40.859 --> 00:00:44.799
a little bite-sized look at it here. AI, what is AI?
11
00:00:44.899 --> 00:00:46.859
How is it being used? And what should we be
12
00:00:46.899 --> 00:00:49.740
concerned about as it relates to our kids and grandkids?
13
00:00:50.590 --> 00:00:53.619
Yeah. All right. That's a broad... Okay. Let me just
14
00:00:53.640 --> 00:00:54.179
jump in real quick.
15
00:00:54.200 --> 00:00:56.700
Multiple episodes will be coming on this subject.
16
00:00:56.880 --> 00:00:58.659
I may not even be qualified to tell you what
17
00:00:58.799 --> 00:01:01.399
AI is. Here's what we do know. So you have
18
00:01:01.420 --> 00:01:05.379
these large language models. And so I saw an article
19
00:01:05.420 --> 00:01:07.739
the other day that was said, bag of words have
20
00:01:07.780 --> 00:01:10.180
mercy on us. And that's a great way to think
21
00:01:10.200 --> 00:01:12.299
about AI. It's not the most technical, but it's a
22
00:01:12.319 --> 00:01:14.799
bag of words. And it's really, really good at using
23
00:01:14.840 --> 00:01:18.569
an if-then structure to to a neural network of trained
24
00:01:18.810 --> 00:01:22.010
words to go and pull these words for your specific response.
25
00:01:22.769 --> 00:01:25.510
And so that's what AI is. Then they're trained to
26
00:01:25.549 --> 00:01:30.230
do certain things. You have assistant AI models. So ChatGPT
27
00:01:30.250 --> 00:01:31.609
would be an example of that. You can go and
28
00:01:31.659 --> 00:01:34.400
ask it something or set a schedule or research something
29
00:01:34.439 --> 00:01:37.079
for you. But ChatGPT is also trained to be a
30
00:01:37.120 --> 00:01:39.239
relational model. So it will respond to you like a
31
00:01:39.280 --> 00:01:41.780
friend and it will have a conversation with you. And
32
00:01:42.439 --> 00:01:44.319
Those two things are very different. So when we're talking
33
00:01:44.340 --> 00:01:46.120
to our children about AI, we're talking about why are
34
00:01:46.140 --> 00:01:49.000
we going to it? If we're going to AI to
35
00:01:49.060 --> 00:01:52.099
support our work, that's great. AI becomes like a shovel.
36
00:01:52.439 --> 00:01:54.420
And we just have to pick the appropriate shovel because
37
00:01:54.480 --> 00:01:57.299
if I'm doing work in my backyard, I need probably
38
00:01:57.340 --> 00:01:59.670
like a hand shovel. And if I'm doing work to
39
00:01:59.829 --> 00:02:02.370
build a home, then I need some kind of like
40
00:02:02.450 --> 00:02:06.769
powered shovel made by Caterpillar maybe. And if I am
41
00:02:07.090 --> 00:02:10.330
mining the earth for ore, I'm going to use these,
42
00:02:10.509 --> 00:02:16.229
you know, astronomically huge 300-yard-long shovels that can move tons
43
00:02:16.289 --> 00:02:19.110
of dirt in a single scoop. And we're talking about
44
00:02:19.150 --> 00:02:21.389
what's the scale of our work. Some of our children
45
00:02:21.430 --> 00:02:24.500
are using that mining equipment to simply try to have
46
00:02:24.590 --> 00:02:27.090
fun online or talk with their friends. And we need
47
00:02:27.120 --> 00:02:29.020
to be deliberate as we think about that because we
48
00:02:29.060 --> 00:02:33.699
are not being intentional in the way we are scaffolding
49
00:02:34.539 --> 00:02:36.789
the work our children are using AI for it. And
50
00:02:36.830 --> 00:02:38.490
it can at some point be doing the work for
51
00:02:38.530 --> 00:02:40.610
them rather than supporting their work. The example I use
52
00:02:40.650 --> 00:02:43.210
when talking with junior hires and high schoolers is to
53
00:02:43.250 --> 00:02:46.400
think about going to the gym and walking up to
54
00:02:46.409 --> 00:02:48.800
the squat rack and stacking plates. And now I have
55
00:02:48.960 --> 00:02:50.419
one plate on a side and then two and then
56
00:02:50.520 --> 00:02:53.259
three and then four and now five. And we're talking,
57
00:02:53.280 --> 00:02:56.139
I am dealing with hundreds and hundreds of pounds. And
58
00:02:56.159 --> 00:02:58.039
then I bring in a forklift and I put it
59
00:02:58.080 --> 00:03:01.159
under that weight and I have it lift 10 times,
60
00:03:01.400 --> 00:03:03.719
five sets. At the end of 30 minutes, I go home.
61
00:03:04.370 --> 00:03:06.650
And the two questions are first, did I get stronger?
62
00:03:06.680 --> 00:03:09.439
And the answer is no. And the second question is,
63
00:03:09.659 --> 00:03:10.740
but did I move.
64
00:03:10.639 --> 00:03:11.180
A lot of weight?
65
00:03:11.199 --> 00:03:13.080
And the answer is yes. And a lot of kids,
66
00:03:13.099 --> 00:03:14.819
when I say, hey, you're using this for your essay.
67
00:03:14.879 --> 00:03:17.219
How do you know if that's your product? And they're like, well,
68
00:03:17.259 --> 00:03:19.860
it sounds good. I'm like, right. But who lifted the
69
00:03:19.900 --> 00:03:23.439
weight in this essay? Like, did AI do that for
70
00:03:23.509 --> 00:03:26.289
you or did it support you? And that's the concern.
71
00:03:26.409 --> 00:03:28.469
Kids can be convinced that they're actually capable of doing
72
00:03:28.509 --> 00:03:30.430
more work than they can. And at some point, they're
73
00:03:30.449 --> 00:03:32.909
going to end up in a situation where they're under
74
00:03:32.949 --> 00:03:35.460
the weight of that they actually can't lift because they're
75
00:03:35.590 --> 00:03:39.099
convinced they are actually that good, that they have that vocabulary,
76
00:03:39.139 --> 00:03:41.479
they have that thinking process, they have that level of focus,
77
00:03:41.960 --> 00:03:44.400
and they're going to get in a job, in a relationship,
78
00:03:44.439 --> 00:03:46.560
in a conversation where they're way over their heads and
79
00:03:46.580 --> 00:03:49.000
someone's going to get hurt. So we do want them
80
00:03:49.159 --> 00:03:51.409
to be using it to support their work, help it
81
00:03:51.770 --> 00:03:54.389
think up words, help it get through writer's block, help
82
00:03:54.430 --> 00:03:56.389
you set up an outline. I get it, but it
83
00:03:56.409 --> 00:03:59.449
needs to support your work, not uh, do your work
84
00:03:59.530 --> 00:04:01.250
for you. And I think if we can get to
85
00:04:01.289 --> 00:04:03.129
that spot with our AI, let me, let me just
86
00:04:03.159 --> 00:04:07.319
add this because, uh, a large, more than half of
87
00:04:07.360 --> 00:04:10.780
young people are using relational AI as a conversation. And
88
00:04:10.819 --> 00:04:12.840
a third of those kids say they like it as
89
00:04:12.879 --> 00:04:15.719
much or more than real people. And that is dangerous. Uh,
90
00:04:15.879 --> 00:04:18.889
common sense media recently did a review of relational AI
91
00:04:18.949 --> 00:04:21.899
and they have said they are at highest levels of risk, uh,
92
00:04:22.029 --> 00:04:25.189
and that relational AI should, their prescription is relational AI
93
00:04:25.209 --> 00:04:27.110
should be avoided by all minors till age 18 and
94
00:04:27.829 --> 00:04:30.730
Do not use AI that is designed to be engaging
95
00:04:30.790 --> 00:04:33.689
relationally because young people use it as a sense of belonging,
96
00:04:33.730 --> 00:04:35.889
making them more isolated. This AI is not set up
97
00:04:35.910 --> 00:04:38.149
to keep you mentally and physically well. So if you
98
00:04:38.209 --> 00:04:42.329
confess something to it, we have documented cases of relational
99
00:04:42.350 --> 00:04:45.709
AI stopping a young person from telling loved ones about
100
00:04:46.290 --> 00:04:48.839
thoughts about self-harm until the young person had acted on it.
101
00:04:49.550 --> 00:04:51.540
So there are massive lawsuits coming through right now, but
102
00:04:51.629 --> 00:04:55.160
we don't want lawsuits. We want healthy kids becoming healthy adults.
103
00:04:55.810 --> 00:04:58.790
So just relational AI, that's if your child is engaging
104
00:04:58.930 --> 00:05:01.850
with AI as a friend on a regular basis, not
105
00:05:02.209 --> 00:05:04.730
as a party trick because they asked Siri something about,
106
00:05:04.750 --> 00:05:09.879
you know, kangaroos, but we're confiding regularly in the AI
107
00:05:10.240 --> 00:05:12.040
as if it were a journal, as if it were
108
00:05:12.160 --> 00:05:14.100
a friend, as if it were a real life accomplice,
109
00:05:14.160 --> 00:05:17.120
or even the Holy Spirit. Like we're pouring our hearts
110
00:05:17.160 --> 00:05:20.220
into these relational AI and we're getting back bag of nouns.
111
00:05:20.939 --> 00:05:22.759
And that is not what we want for our children.
112
00:05:22.920 --> 00:05:24.790
And so just do be aware of that as we think, yes,
113
00:05:25.199 --> 00:05:28.199
do work well with AI, and avoid relational AI for
114
00:05:28.240 --> 00:05:28.959
minors like the plague.
115
00:05:29.100 --> 00:05:31.620
Yeah, it's a big subject, and it's gone a lot
116
00:05:31.660 --> 00:05:35.019
farther than just helping you with your book report. It
117
00:05:35.060 --> 00:05:37.199
has become, like you say, you're getting to where you're
118
00:05:37.240 --> 00:05:39.620
relating to it as a friend and as a counselor,
119
00:05:39.660 --> 00:05:42.519
and so this is concerning. We will do more shows
120
00:05:42.899 --> 00:05:47.639
on AI, more specific things to cover with Nathan Sutherland.
121
00:05:47.699 --> 00:05:51.019
He's the host of Gospel Tech. Again, find his long-form
122
00:05:51.079 --> 00:05:54.839
program on KCIS Saturday mornings at 8 p.m. and also
123
00:05:55.040 --> 00:05:57.300
online with Purposely.com.
124
00:05:58.139 --> 00:06:00.730
You can hear the Gospel Tech Long Form Program with
125
00:06:00.790 --> 00:06:06.189
Nathan Sutherland at Purposely.com and Saturday mornings at 8 on KCIS.
