Latest / Gospel Tech with Nathan Sutherland / AI, the Supreme Court, and Peter's Rebuke (Part 1)

Transcript

AI, the Supreme Court, and Peter's Rebuke (Part 1) · Gospel Tech with Nathan Sutherland

0:01
Speaker 1: Purposely Your Life, God's Purpose Listen at purposely dot com.
0:14
Speaker 2: Hello everyone, and welcome to the Gospel Tech Podcast. My
0:17
Speaker 2: name is Nathan Sutherland and this podcast is dedicated to
0:20
Speaker 2: helping families love God and use tech. Today we are
0:23
Speaker 2: jumping into a conversation on AI. I am excited for
0:27
Speaker 2: this conversation because it's both progressing from where we have
0:31
Speaker 2: been discussing and it picks up with the news. So recently,
0:35
Speaker 2: there's a very famous lawyer named Neil Katiel, and he
0:41
Speaker 2: recently had a tweet because I can't call them an X,
0:45
Speaker 2: I just won't, so had a tweet where he well,
0:49
Speaker 2: let me just read a little bit of it for you.
0:53
Speaker 2: He argued, the tariff's case is very excited. The tariff's
0:55
Speaker 2: case one, and I'll read you portions of the time
1:00
Speaker 2: of Wednesday night of this past week. This had one
1:03
Speaker 2: point four million views. Five months ago, I argued before
1:07
Speaker 2: the President's for trillion dollar tariffs at the Supreme Court
1:10
Speaker 2: in two hundred and thirty seven years, the court has
1:12
Speaker 2: never struck down a sitting president's signature initiative. Legal scholars
1:16
Speaker 2: said it was impossible. Some of my own colleagues said
1:18
Speaker 2: it was impossible. We won six'. Three he goes on
1:21
Speaker 2: to talk ABOUT i had a mindset, coach an improv,
1:24
Speaker 2: coach a meditation, coach And, Harvey harvey predicted many of the,
1:29
Speaker 2: questions was, brilliant, tireless occasionally. Insufferable here's the, Catch harvey
1:34
Speaker 2: isn't a person goes down to say, Tomorrow ted Releases
1:41
Speaker 2: ted The talk forum releases my talk about what really
1:45
Speaker 2: happened and WHAT i learned stand at the. PODIUM ai can,
1:48
Speaker 2: PREDICT ai can. Analyze WHAT ai cannot do is what
1:51
Speaker 2: actually won the? Argument connect read the, room here the justice.
1:54
Speaker 2: Words so go find. Yours go deeper in this. Age
1:59
Speaker 2: that's where your edge, lived is the human. Edge so
2:02
Speaker 2: he posts this tweet about how awesome he did and
2:06
Speaker 2: HOW ai was his superpower that got him. There the
2:10
Speaker 2: Actual Ted talk, itself, which by the, way this was
2:13
Speaker 2: like lambasted as on so many levels of like what
2:16
Speaker 2: are you? Doing like we get, it you're very, smart
2:18
Speaker 2: you're very.
2:18
Speaker 1: Good.
2:19
Speaker 2: Congratulations but What ted, talk which you should actually, watch
2:24
Speaker 2: is much more interesting than that. Tweet The Ted talk
2:28
Speaker 2: basically comes down, to, HEY ai levels the playing field for,
2:33
Speaker 2: everyone and as someone who's experienced a lot of academic.
2:36
Speaker 2: Success Mister katiel is attempting to, say, like, Yes i'm
2:42
Speaker 2: good in front of people and that's my, strength and
2:44
Speaker 2: like that's WHAT i play to and LIKE ai helped
2:49
Speaker 2: me get. This and it's actually very interesting the QUOTE
2:53
Speaker 2: i like the most from this presentation and it really
2:56
Speaker 2: needs to be referenced because otherwise we can just make
2:59
Speaker 2: fun of this really random. Situation but he, says now
3:04
Speaker 2: this is this is bigger than just. Law it's about
3:08
Speaker 2: all of. Us for, centuries the expert was the person
3:10
Speaker 2: who read the, room who read the, most who remembered the,
3:13
Speaker 2: most and who'd seen the, most the season, doctor the experienced.
3:17
Speaker 2: Lawyer their edge was accumulated. KNOWLEDGE ai is making that
3:21
Speaker 2: edge nearly, worthless not because humans no longer, matter but
3:24
Speaker 2: because the particular advantage pattern recognition across vastata and breadth
3:28
Speaker 2: of knowledge is now available to. Anyone and, THAT i
3:32
Speaker 2: believe for us is the. Takeaway so if you've heard
3:34
Speaker 2: about This Neil katilla thing or you have any strong
3:37
Speaker 2: takes on, it like that's. Great but the point for
3:39
Speaker 2: us as parents and for followers Of christ and for
3:41
Speaker 2: individuals living in AN ai environment to the extent that
3:45
Speaker 2: we even understand what that. Means Mister katiell's point is.
3:50
Speaker 2: Huge this has leveled the playing. FIELD i, mean if
3:53
Speaker 2: you think about, it, like it doesn't take away the human.
3:56
Speaker 2: Element there still needs to be a person to apply that.
3:57
Speaker 2: Info but this is the next step for us growing
3:59
Speaker 2: up With, google and if we just wanted to know a,
4:02
Speaker 2: fact we could just google, It, Like, AH i don't remember,
4:04
Speaker 2: that google. It IF i can't remember A bible, VERSE
4:06
Speaker 2: i can type in the. Phrase IF i don't remember the,
4:09
Speaker 2: PHRASE i can give it the. Context what Did jesus
4:12
Speaker 2: say about seeds while talking to a large? Crowd, like
4:16
Speaker 2: and it'll give me the seven instances of?
4:18
Speaker 1: That.
4:19
Speaker 2: Like it is just a more advanced version of a
4:21
Speaker 2: skill we've. Had, Right you could always go to some
4:24
Speaker 2: kind of biblical study tool and you could look up
4:27
Speaker 2: keywords and sometimes one of the back of Your bible
4:29
Speaker 2: would have five, instances but you could find a bigger
4:32
Speaker 2: one and it would have. Fifty And google just gives
4:35
Speaker 2: you all of that at your. Fingertips now is that
4:37
Speaker 2: bad for human? Understanding, yes it's terrible because the time
4:40
Speaker 2: it took you and the intentionality required to go find those.
4:43
Speaker 2: Books we're an important part of being, deliberate and this
4:48
Speaker 2: has leveled the playing field so someone without the, time
4:50
Speaker 2: so that someone who's a stay at home parent with
4:52
Speaker 2: kids at their ankles can still get that info quickly
4:55
Speaker 2: and move on with their reference or their, study or their,
4:58
Speaker 2: sharing or their encouragement or their personal reflection like they
5:01
Speaker 2: can do that because it's leveled the playing. Field no
5:04
Speaker 2: longer do you have to have these tones stashed in
5:07
Speaker 2: a back, corner or have access to a library that has,
5:09
Speaker 2: them or have the time to flip through them and
5:11
Speaker 2: store them like this is its leveled playing. Field, similarly
5:16
Speaker 2: our children no longer need and hear me out to
5:20
Speaker 2: the end of the, stop they no longer need the
5:22
Speaker 2: ability to write a five paragraph essay on their, own
5:26
Speaker 2: not because it's not an important, skill but because they
5:28
Speaker 2: have a very capable tool that will do that product for.
5:31
Speaker 2: Them but the downside is very. Similar they actually don't
5:35
Speaker 2: know what they, know and they don't know very similar to.
5:37
Speaker 2: Us if we can just look up A bible verse
5:38
Speaker 2: when we want, it there's a chance that we don't
5:40
Speaker 2: actually know That bible. Verse we remember there was a reference,
5:43
Speaker 2: there but we may not know that. Verse and if
5:46
Speaker 2: we don't have it in our, hearts it does get
5:48
Speaker 2: more difficult to apply it and to reference it and
5:51
Speaker 2: to make sure it was done, accurately because we need
5:54
Speaker 2: to know some of that. Context you get no context
5:56
Speaker 2: When google pulls you up a. Verse it's just that,
5:59
Speaker 2: verse sometimes in a list of other verses related to
6:02
Speaker 2: the topic you. Searched so now we're missing. Out and
6:05
Speaker 2: actually today's first will be an example of that of, like,
6:09
Speaker 2: ooh the context, Matters so, SORRY i didn't mean to
6:13
Speaker 2: tease A bible, verse but that context doesn't, matter and
6:16
Speaker 2: we lose. That, similarly when our children write a paper
6:19
Speaker 2: with An internet assistant to say CHAT, gpt yes they
6:23
Speaker 2: get a good, product but did they produce the? Product
6:27
Speaker 2: like maybe maybe they did that fully knownly and they
6:29
Speaker 2: just don't have the time and they just need a little. Support,
6:31
Speaker 2: okay maybe they just prompted THE ai went and changed some,
6:36
Speaker 2: words so it looks more like they did, it and
6:38
Speaker 2: they don't actually have the ability to synthesize the, ideas
6:40
Speaker 2: and they may not even know that. Yet they just
6:43
Speaker 2: KNOW i don't. KNOW i know how to do, this
6:45
Speaker 2: and every time someone asks me to do, It i'm
6:46
Speaker 2: going to come back here and do it again and
6:48
Speaker 2: there is a concern of a loss of. Skill so
6:51
Speaker 2: then we have Mister cutiel who's, like, OH i USED
6:53
Speaker 2: ai and it's leveled the playing. Field that doesn't mean
6:57
Speaker 2: the product always gets. Better it means a very smart
6:59
Speaker 2: person who's incredibly capable of his job already has said
7:03
Speaker 2: WHOA ai is really good at noting data over massive
7:09
Speaker 2: sum of. Space in, fact he goes on to say,
7:10
Speaker 2: this he's named THIS Ai. Harvey they custom built it
7:14
Speaker 2: for this particular. Case and he goes over twenty five
7:16
Speaker 2: years of court everything basically went before The Supreme court
7:21
Speaker 2: and then ask THE, ai, hey here's what we're arguing.
7:23
Speaker 2: On here's kind of my talking. Points what SHOULD i?
7:26
Speaker 2: Expect and it went justice by justice and said this
7:28
Speaker 2: one's going to be concerned about this stuff this one
7:31
Speaker 2: and was able to. Predict he even goes through and,
7:33
Speaker 2: predicts like here's What harvey. Said this particular justice would
7:36
Speaker 2: say here was her response and it's almost, verbatim and that,
7:40
Speaker 2: right like that is something a person could have done
7:43
Speaker 2: by reviewing twenty Five like that's a doctoral degree to
7:46
Speaker 2: go over twenty five years Of Supreme court on a
7:48
Speaker 2: specific topic and then start writing out this is the
7:51
Speaker 2: way each of the Nine justice justices would. Respond but
7:54
Speaker 2: AN ai is really a large language, model is really
7:58
Speaker 2: good at. That like that is what it. Is superpower
8:00
Speaker 2: Is so give that to someone who's already, capable artis,
8:02
Speaker 2: smart already has the, relationships already understands how he wants
8:05
Speaker 2: to handle this. Case, Okay like that's going to be
8:09
Speaker 2: a powerful. Resource BUT i, so, yes it's. Around humans
8:13
Speaker 2: are going to use. It think about where it's an
8:15
Speaker 2: ethical way to use. This and again this is as
8:18
Speaker 2: he his words. Bespoke this is a CUSTOM, ai so
8:21
Speaker 2: it's not publicly available. Data he's not losing his personal
8:25
Speaker 2: edge on, This he's not all those. Things this individual
8:29
Speaker 2: is also a very prickly. Person loves himself some, attention
8:33
Speaker 2: so he goes on and does things like this tweet
8:37
Speaker 2: that has gotten, him, well it's got him a lost of.
8:39
Speaker 2: Attention if there is truly no bad, publicity then he
8:42
Speaker 2: must be very happy with. It but our point for
8:45
Speaker 2: this IS ai is, out it's in the, classroom it's
8:49
Speaker 2: in our. Jobs AND i actually want to focus a
8:52
Speaker 2: little bit on HOW ai is trained because this is
8:55
Speaker 2: something THAT'S i think we need to just be a
8:57
Speaker 2: little more aware of what actually is, happening AND i
8:59
Speaker 2: want to take on a little bit of a journey
9:00
Speaker 2: to get. There so THIS ai case where it showed
9:04
Speaker 2: up in The Supreme court and was used effectively by
9:08
Speaker 2: an individual who, says this is a mage, deal does
9:10
Speaker 2: an eighteen MINUTE ted talk and his takeaway is the
9:13
Speaker 2: playfield is now. Leveled, actually BEFORE i go into, training
9:16
Speaker 2: let's talk about that plain field that's. Leveled anyone can
9:19
Speaker 2: make a mona art just by typing it. In you
9:22
Speaker 2: can go see this for. YOURSELF a mid, journey for,
9:24
Speaker 2: example would be capable of doing. That you give it,
9:27
Speaker 2: words it gives you. Pictures anyone now can design. Music
9:33
Speaker 2: sons Of glory is available On spotify and it's A
9:37
Speaker 2: christian heavy metal band that sings just the, psalms and
9:42
Speaker 2: it's AN ai. Band there is no actual. Singer it's
9:44
Speaker 2: not modifying someone's. Voice someone went in and prompted it and,
9:48
Speaker 2: said use this kind of, instrument sound like, this here's your,
9:52
Speaker 2: lyrics and go and it made these heavy metal songs
9:56
Speaker 2: that are verbatim the first eight psalms The, bible with
10:00
Speaker 2: ridiculous guitar riffs and drum part like very talented you'd
10:06
Speaker 2: have to be very talented to recreate, this but it's ALL.
10:08
Speaker 2: Ai so anyone now has access to incredible music, production
10:13
Speaker 2: to incredible art, capabilities to very nefarious opportunities to essay,
10:19
Speaker 2: writing to massive levels of research and data. Recognition we
10:23
Speaker 2: all have the. Ability do you still need a sharp
10:25
Speaker 2: mind behind it to say was this worth? Creating? Yeah
10:27
Speaker 2: and you still need a conscience and you still need
10:30
Speaker 2: to be. Ethical and the ability is now. There so
10:34
Speaker 2: that's important because how did we get this? Ability, parents
10:42
Speaker 2: as we strive to raise kids who Love god and use,
10:45
Speaker 2: tech it's intimidating to look at this next stage of
10:48
Speaker 2: life and think about what academics may look like for.
10:51
Speaker 2: Them when you're thinking about, college Considered George Fox university
10:55
Speaker 2: for your children because of three. Things, first it's going
10:57
Speaker 2: to give you a more, intimate known environ where they
11:00
Speaker 2: have a focus on being known, personally academically and. Spiritually
11:04
Speaker 2: your child will grow in their. Faith ninety three percent
11:07
Speaker 2: of undergrad students At George fox last year said that
11:10
Speaker 2: being At George fox has helped them grow in the
11:12
Speaker 2: relationship With. Jesus we also want them to grow, academically
11:17
Speaker 2: which means being a part of the university with rigorous
11:19
Speaker 2: academic standards and opportunities to be. Engaged they have nationally recognized, engineering,
11:25
Speaker 2: nursing and education, programs as well as chances to study
11:28
Speaker 2: abroad here within THE us and then internationally and finally,
11:31
Speaker 2: personally to be part of a vibrant campus atmosphere where
11:34
Speaker 2: you can play intramural sports or part of THEIR Ncaa
11:38
Speaker 2: division three athletic programs with twenty three sports. Available go
11:42
Speaker 2: to a place that's going to support your, goals not
11:45
Speaker 2: try to make you support, theirs to have that goal
11:48
Speaker 2: be to raise you up to be All god's called
11:51
Speaker 2: you to, be and to truly equip your young person
11:54
Speaker 2: to Love god and then from their love Of god
11:56
Speaker 2: to go Wherever god has called, them and that includes
11:58
Speaker 2: being equipped with a degree that will helping do, that
12:00
Speaker 2: to be missionally equipped to love and serve, others to
12:04
Speaker 2: lead Where god has planned the, man to know The
12:06
Speaker 2: gospel for themselves and be able to share it with.
12:07
Speaker 2: Others So George Fox university is an amazing resource for our,
12:11
Speaker 2: kids for our community here in The, northwest and for
12:13
Speaker 2: anyone who's looking to raise up a young person who
12:16
Speaker 2: can Love god and from that love. OTHERS i want
12:25
Speaker 2: to take just a look at Open ai and Then
12:28
Speaker 2: anthropic Open. Ai if you go back to twenty, eighteen
12:32
Speaker 2: they had spent a long time creating something that could
12:37
Speaker 2: play a video. Game so in twenty, EIGHTEEN i actually
12:40
Speaker 2: don't even know the. Name i'm assuming it was the
12:43
Speaker 2: Open ai four because the next year it's OPEN a.
12:46
Speaker 2: Five BUT i actually don't have the name of what
12:47
Speaker 2: the product. Was they picked a game Called doda and
12:51
Speaker 2: all you really need to know about this this a video.
12:53
Speaker 2: Game all you really need to know about this video
12:55
Speaker 2: game is there's five players on a team and their
12:58
Speaker 2: job is to march across this map and defeat the
13:00
Speaker 2: other team's main tower in the. Middle there's little mini.
13:03
Speaker 2: Towers there's you, know dozens and dozens of different heroes
13:07
Speaker 2: you can pick. From every hero has special abilities and
13:09
Speaker 2: ways to support the other heroes or to lead with
13:13
Speaker 2: superpowers that they need the support of other. Heroes so
13:16
Speaker 2: it's very. Complex SO i, mean just imagine. Chess but
13:19
Speaker 2: if you can change your queen out for a different,
13:21
Speaker 2: character that's super, strong. Right but now there's five people
13:25
Speaker 2: playing chess on each, team and you can get items
13:28
Speaker 2: and levels and points and like it's this whole. Thing
13:31
Speaker 2: but the whole point is you can go to the
13:33
Speaker 2: top side of the, map the middle of the, map
13:34
Speaker 2: at the bottom of the, map and there's five of
13:36
Speaker 2: you and you need to figure out how to defeat
13:38
Speaker 2: the other. Team so with that as its, premise it
13:43
Speaker 2: has become very. Popular doda is the name of the
13:45
Speaker 2: Game dota two, specifically and these tournaments are. Huge there's
13:49
Speaker 2: a twenty million dollar price perse at one of these,
13:51
Speaker 2: tournaments meaning you could win twenty million. Dollars, ACTUALLY i
13:56
Speaker 2: think you win twelve.
13:57
Speaker 1: If you take.
13:58
Speaker 2: FIRST i forgive me if that's not. True so The
14:01
Speaker 2: international is one of the two large ones twenty. Eighteen
14:05
Speaker 2: this OPEN ai team has trained a computer to go
14:09
Speaker 2: play in this. Tournament and the way that works is
14:12
Speaker 2: BECAUSE i was like that seems like. Cheating it runs
14:14
Speaker 2: off the exact same premise a player would run. Off
14:17
Speaker 2: so it controls each hero and as the heroes move,
14:20
Speaker 2: around it gets field of. Vision it can only see
14:22
Speaker 2: what that player can see and then if it can
14:25
Speaker 2: react to. That and the way they did that is they, said,
14:28
Speaker 2: great play this game for a really long time and
14:30
Speaker 2: then we'll give you. Feedback and it doesn't have to
14:32
Speaker 2: use real world time because it's a. Supercomputer so it's
14:34
Speaker 2: just like in the Show The, flash where if you're
14:37
Speaker 2: moving at the speed of, light everyone else looks like
14:39
Speaker 2: they're frozen in. Place the computer does the same. Thing
14:42
Speaker 2: so it can play one thousand hours in a few,
14:44
Speaker 2: minutes and so it plays it thousands hours and it's
14:48
Speaker 2: fitted or let's say in a, day plays a thousand.
14:50
Speaker 2: Hours at the end of a, day they go back
14:53
Speaker 2: through and they said this was, good that was, Bad
14:54
Speaker 2: and they get very specific feedback when you were this
14:56
Speaker 2: hero and you did that, thing that was. Good so
14:58
Speaker 2: they have pro players now giving analog inputs and saying yes, no, yes,
15:04
Speaker 2: no and building this flow chart of decisions and in
15:07
Speaker 2: a transit trains and they ended up training it for
15:08
Speaker 2: ten thousand days was the amount of in game time
15:11
Speaker 2: the computer. Had it took them three months to get,
15:14
Speaker 2: that so obviously they didn't wait ten thousand, days but
15:17
Speaker 2: the computer was able to get just tons and tons
15:19
Speaker 2: of tons of reps because it has tons of processing.
15:21
Speaker 2: Power they eventually end up with the train. Model it
15:23
Speaker 2: goes to the twenty Eighteen dota two tournament and it
15:26
Speaker 2: gets smoked by the. Protein they go, back they go
15:29
Speaker 2: back to make Open ai, five and twenty, nineteen they
15:32
Speaker 2: show up at the other Major dota to. Tournament at this,
15:34
Speaker 2: point they have trained it for three, months they have
15:38
Speaker 2: put forty five thousand days worth of playtime into, it
15:42
Speaker 2: and it comes back significantly. Stronger it even does a
15:45
Speaker 2: little trash, talking WHICH i. Appreciate it doesn't matter for
15:47
Speaker 2: our greater, conversation but it notices, like, oh people seem
15:51
Speaker 2: to trash talk to try to make their opponents, upset
15:53
Speaker 2: and so it does at one point about five, minutes
15:56
Speaker 2: and it's, LIKE i predict a ninety percent win, ratio
16:00
Speaker 2: and sure, enough it absolutely trounces the protein twice too
16:04
Speaker 2: if it wins two and. Up that was an incredible
16:06
Speaker 2: amount of. Input and Open AI's kind of reflections on
16:09
Speaker 2: this as a company is the same, thing, like, yes
16:12
Speaker 2: it did something very. Complicated, yes that was a milestone for.
16:14
Speaker 2: Us and Open ai five took so much manual feedback
16:19
Speaker 2: and so, much, like so much intensive time that it
16:22
Speaker 2: wouldn't be practical in real world solutions because it just
16:26
Speaker 2: it would take too much to tell. It yes that
16:27
Speaker 2: was good, info no that was bad. Info so then
16:31
Speaker 2: we jump forward and we get our twenty twenty Two
16:35
Speaker 2: OPEN ai products that lead us to a twenty twenty
16:38
Speaker 2: three CHAT gpt and kind of the advent OF ai
16:43
Speaker 2: as we think of. It blast super forward then to
16:49
Speaker 2: twenty twenty six and twenty twenty. Five Actually amazon, announces,
16:55
Speaker 2: hey we've got a warehouse In. Michigan it's twelve hundred.
16:59
Speaker 2: Acre so for, context that would be if you walked
17:02
Speaker 2: in a straight line for forty, minutes you would end
17:06
Speaker 2: you would walk from one corner all the way to
17:08
Speaker 2: the next corner of this. Warehouse that is one point
17:11
Speaker 2: eight miles, square is what twelve hundred acres. Is then
17:16
Speaker 2: you could take a ninety degree, left you'd walk for
17:18
Speaker 2: another forty, minutes and you've now walked half of the.
17:21
Speaker 2: Building you've ended up on the far corner of that.
17:23
Speaker 2: Building so it took you eighty minutes just to walk
17:26
Speaker 2: the span of two. Walls and they say we have
17:29
Speaker 2: twelve hundred. Acres it's filled with, servers and the servers
17:33
Speaker 2: are filled with Our trainium to. Chip So trainium two
17:37
Speaker 2: this does. Matter trainum two is a chip that is
17:42
Speaker 2: very good at processing. Data it's incredibly efficient at, it
17:46
Speaker 2: and part of that comes from just the way it's
17:48
Speaker 2: designed so we gauge processing speed and flops all caps,
17:55
Speaker 2: flops and it's basically how many processes can this run
17:59
Speaker 2: in a. Second so my, COMPUTER i have AN m
18:02
Speaker 2: three processor in A MacBook. Pro this can run a
18:06
Speaker 2: max of five terraflops at max, peak so that would
18:11
Speaker 2: be a terraflop would be a ten with twelve zeros after.
18:16
Speaker 2: It so basically it can run five trillion zero ones
18:22
Speaker 2: in a. Second and that comes thanks to teeny tiny
18:25
Speaker 2: little transistors that allow it to process an incredible amount of.
18:28
Speaker 2: Data WHEN i was growing, UP i remember very vividly
18:32
Speaker 2: when the first dual core processor came, out meaning my
18:36
Speaker 2: chipset has two process two processing, cores but it can
18:40
Speaker 2: run more data now because at some point we was
18:43
Speaker 2: talking there was a dual thread for a while, anyway
18:45
Speaker 2: or quad, thread there was the ability to get it
18:48
Speaker 2: to do. More, well we don't have dual cores or
18:50
Speaker 2: even quad. Cores my processor has fourteen cores in it
18:54
Speaker 2: and it can run five, terraflops so five times ten
19:00
Speaker 2: to the, twelfth, like that's a. Lot that's a whole
19:05
Speaker 2: lot of, numbers and that's a lot of processing. Power
19:08
Speaker 2: so a trainium two can run twenty. Petaflops so for our,
19:14
Speaker 2: purposes that's just four thousand times more processing in a.
19:17
Speaker 2: Second that's one. Chip those chips are set into. Racks
19:22
Speaker 2: those racks are set into server, racks and those server
19:25
Speaker 2: racks are put into a twelve one hundred acre. Facility
19:30
Speaker 2: that's important Because anthropic is AN ai company we've talked about.
19:35
Speaker 2: Before they have a product Called. Claude claude is this
19:38
Speaker 2: large language. Model it is different from THIS ai that
19:41
Speaker 2: was trained to play this video game that now seems
19:43
Speaker 2: archaic because instead of having to say yes no in
19:47
Speaker 2: some kind of flow chart for THE, ai you simply
19:50
Speaker 2: give it a, property Say, claude here's what we want
19:52
Speaker 2: you to. Do and in order to train, it you
19:54
Speaker 2: do something very. Similar if you want A claud to
19:57
Speaker 2: get good at, something you give it a very complex
20:00
Speaker 2: set of, prompts and then you give it, food, right
20:04
Speaker 2: you give it energy and you electricity in this, case
20:07
Speaker 2: and you give it data and you let this thing
20:09
Speaker 2: just go. Nuts so they did. This they Took claud
20:15
Speaker 2: and they had Made, opus and they're very proud Of,
20:17
Speaker 2: opus and they, said this is a great. Thing opus
20:19
Speaker 2: four point six is a big. Deal we are now
20:22
Speaker 2: going to make a new version of claud and we're
20:26
Speaker 2: gonna call It mythos and so they give it it's
20:29
Speaker 2: training time in This so they started with five hundred
20:32
Speaker 2: thousand of These trainium two. Chips so just run. That
20:37
Speaker 2: if one of these chips is four Thousand MacBook proium three.
20:41
Speaker 2: Processors we have five hundred thousand of these. Things and
20:46
Speaker 2: then by the end of twenty twenty five they upgraded
20:48
Speaker 2: to a million of, These so we're going to use
20:50
Speaker 2: a million of. These just hear the. Scale we're going
20:56
Speaker 2: to use a million of, these and we're going to
20:57
Speaker 2: use them for, months and we're going to Let claude
20:59
Speaker 2: train themselves for. ITSELF i should not personalize. It please
21:02
Speaker 2: don't do. That that was a slip. Up and at
21:04
Speaker 2: this POINT i just want to kind of LIKE i
21:06
Speaker 2: want to leave that, here that this idea that we
21:09
Speaker 2: have this MASSIVE ai that is going to be training
21:11
Speaker 2: itself using a huge amount of computing. Power like we're
21:13
Speaker 2: just going to wrap up the conversation right, there because
21:15
Speaker 2: otherwise it will expand to a massive. Conversation AND i
21:19
Speaker 2: want to make SURE i can put a bow on
21:20
Speaker 2: the second. Part so so, far what we know.
21:22
Speaker 1: Is THAT.
21:24
Speaker 2: Ai is being used in the real. World That neil
21:27
Speaker 2: kept you all this last week was LIKE i made
21:28
Speaker 2: a BESPOKE ai and it's. Amazing and not only AM i,
21:31
Speaker 2: amazing but MY ai is the wave of the, future
21:35
Speaker 2: and that this is true for our, children that we
21:37
Speaker 2: want to make sure that we recognize that this is
21:40
Speaker 2: the way their world is going to be. Working so
21:42
Speaker 2: let's do it with, Integrity let's do it with, Creativity
21:44
Speaker 2: let's do it with, conscience both in how we make these,
21:47
Speaker 2: things as we'll talk about next, week and how we
21:51
Speaker 2: use them. Afterwards this conversation is bigger than any single
21:56
Speaker 2: individual or even a single company's. Product But i'm using
21:59
Speaker 2: these specific so we're just noting that back in twenty,
22:02
Speaker 2: Seventeen Open ai was making a product that in twenty
22:05
Speaker 2: eighteen lost video game tournament and twenty nineteen won. It
22:09
Speaker 2: that currently we have products that aren't even trained by.
22:12
Speaker 2: Poking they're being trained by these massive, facilities which we'll
22:16
Speaker 2: get into next. Week and our takeaway from this for
22:19
Speaker 2: this week is a deliberate focus on not gaining the
22:25
Speaker 2: world and losing our. Souls jesus says In mark that
22:28
Speaker 2: the premise of following him is to die to ourselves
22:31
Speaker 2: and to follow. Him he talks about being, ashamed as
22:34
Speaker 2: we'll cover next. Week but the shame is Not i'm
22:37
Speaker 2: scared to mention. Him the shame IS i don't Like god's.
22:39
Speaker 2: Plans it's the shame That judas. Felt it's the shame
22:42
Speaker 2: That peter used to Rebuke jesus and, say, no you're
22:44
Speaker 2: not going to, Die you're gonna be. Amazing we're all
22:47
Speaker 2: going to. Win and he's, like, yeah, yeah, no we
22:49
Speaker 2: should probably set on that for a, Minute, peter because
22:52
Speaker 2: that's not actually the. Plan and therefore you're Fighting god's.
22:55
Speaker 2: Will let's not USE ai to Fight god's. Will let's
22:59
Speaker 2: USE ai to Do god's. Will and so, yes we'll
23:01
Speaker 2: get more into this training look At mythos and what
23:04
Speaker 2: that could mean for the, future because y'all's, like we
23:06
Speaker 2: should all USE ai at all the times and maybe all,
23:10
Speaker 2: right but we also need to be delivered about is
23:13
Speaker 2: this the time to use this resource or is this
23:15
Speaker 2: the time to see that What god's calling us to
23:18
Speaker 2: his Hard and it may not be the kind of
23:20
Speaker 2: thing we solve WITH. Ai it might be the kind
23:22
Speaker 2: of thing we deliberately step into because people are needed
23:25
Speaker 2: who carry the spirit Of god as jars of, clay
23:27
Speaker 2: and When god breaks, us he spills out not bits
23:32
Speaker 2: and lips from AN. Ai so with that in, MIND
23:36
Speaker 2: i hope this was. ENCOURAGING i hope you join us
23:39
Speaker 2: next week as we continue this conversation ABOUT ai used
23:42
Speaker 2: at a high, level used by our, children used in the,
23:44
Speaker 2: classroom used in our daily, lives and what this looks
23:48
Speaker 2: like both on the production, side but then on THE u,
23:51
Speaker 2: side how we can make sure we use it well
23:53
Speaker 2: to look more Like jesus and not distract us from.
23:55
Speaker 2: It hope this is encouraging for. You if you have any,
23:57
Speaker 2: questions reach Out nathan at gospeltech dot net And i'll
24:00
Speaker 2: look forward to joining you next week for this conversation
24:03
Speaker 2: about how we can Love god and use. Tech i'll
24:05
Speaker 2: talk with you then