Latest / The Financial Exchange Show / AI Agents Are Already Breaking the Rules

Transcript

AI Agents Are Already Breaking the Rules · The Financial Exchange Show

0:00
Speaker 1: The Financial Exchange is produced by Money Matters Radio and
0:03
Speaker 1: is hosted by employees of the Armstrong Advisory Group, a
0:05
Speaker 1: registered investment advisor. All opinions expressed are solely those of
0:08
Speaker 1: the hosts. Do not reflect the opinions of Armstrong Advisory
0:11
Speaker 1: or anyone else. Investments can lose money. This program does
0:14
Speaker 1: not offer any specific financial or investment advice. Please consult
0:17
Speaker 1: your own financial, tax, and estate planning advisors before making
0:20
Speaker 1: any investment decisions. Armstrong Advisory and the advertisers heard on
0:23
Speaker 1: this program do not endorse each other or their services.
0:26
Speaker 1: Armstrong and Money Matters Radio do not compensate each other
0:28
Speaker 1: for referrals and are not affiliated. This is the Financial
0:34
Speaker 1: Exchange with Chuck Zada and Mark Fandaddy, your exclusive look
0:39
Speaker 1: at business and financial news affecting your day, your city,
0:42
Speaker 1: your world. Stay informed and up to date about economic
0:46
Speaker 1: and market trends, plus breaking business news every day. The
0:50
Speaker 1: Financial Exchange is a proud partner of the Disabled American
0:53
Speaker 1: Veterans Department of Massachusetts. Help us support our great American
0:57
Speaker 1: heroes by visiting DAV five KT Boston and making a
1:01
Speaker 1: donation today. This is the Financial Exchange with Chuck Zada,
1:06
Speaker 1: and Mark Vandebti.
1:10
Speaker 2: Okay, it's Chuck, it's Mark, and it's Tucker. And not
1:14
Speaker 2: much going on in markets today. To be completely honest,
1:17
Speaker 2: the S and P five hundred is flat. You know,
1:20
Speaker 2: it's I put on a new shirt today. Feels like
1:24
Speaker 2: kind of a waste to get out of bed for that. Uh.
1:27
Speaker 2: You know, you've got the dow up sixty three points. Whooped?
1:31
Speaker 2: You do the NASDAK composite down forty eight not particularly
1:36
Speaker 2: interesting to me. As I said at the beginning of
1:37
Speaker 2: the week, if you have some gardening to do, you
1:40
Speaker 2: want to mow the lawn, you want to go play
1:41
Speaker 2: some golf, go out boating, fishing, hunting.
1:44
Speaker 1: Uh.
1:45
Speaker 2: If you're in the Southern Hemisphere listening to us and
1:47
Speaker 2: you want to go skiing, have at it. This is
1:51
Speaker 2: the week to do it. There's just not much going
1:52
Speaker 2: on this week, and it's being reflected in inequity markets,
1:55
Speaker 2: which are just not really doing too much. The bond
1:58
Speaker 2: market a little bit more interesting, at least you get
2:01
Speaker 2: the ten year treasury it's up two point six basis
2:03
Speaker 2: points to four point sixty five four. Okay, maybe I
2:07
Speaker 2: can get out of bed for that. You know, there's
2:08
Speaker 2: a little something there. Dollar index off point zero six percent.
2:13
Speaker 2: Not much happening there. Precious metals, we do have a
2:16
Speaker 2: little bit of action there. Gold's up almost two percent,
2:18
Speaker 2: seventy five bucks an ounce to forty one fifty one,
2:22
Speaker 2: so that's kind of cool. We got crude oil continuing
2:24
Speaker 2: to move upward as well. Not cool because none of
2:27
Speaker 2: us like it, but it is happening. West Texts Intermediate
2:31
Speaker 2: up another two dollars and thirty nine cents a barrel
2:33
Speaker 2: to eighty six seventy three triple a. National average for
2:36
Speaker 2: gas prices is at four h six up another four
2:39
Speaker 2: point one cents overnight diesel up four cents as well,
2:44
Speaker 2: And again, who knows how this progresses over the coming weeks,
2:48
Speaker 2: because it's all going to be dependent on what happens
2:49
Speaker 2: in the Middle East most you know, that's the biggest
2:52
Speaker 2: impact area at least, but we're probably getting into that
2:56
Speaker 2: four twelve to four to fifteen range by the end
2:58
Speaker 2: of the week. That's kind of where things are moving.
3:01
Speaker 2: So i'd say another six to nine cents up prior
3:04
Speaker 2: to the weekend is probably where things are pointing, and
3:08
Speaker 2: then it's gonna be determined to buy what happens with
3:11
Speaker 2: the straight of horn moves. What happens with the Houthis
3:15
Speaker 2: and Yemen and today try to close the Red Sea?
3:18
Speaker 2: What continues to happen in Russia. Do they continue to
3:21
Speaker 2: see their refineries hit by Ukrainian drone strikes and this
3:24
Speaker 2: China continue its buyer strike of crude oil? Or do
3:27
Speaker 2: they come back to the market and start purchasing in
3:30
Speaker 2: more normal quantities. So we'll watch all that stuff so
3:33
Speaker 2: that you don't have to and then you can just
3:34
Speaker 2: turn us on and listen to us, and we'll tell
3:37
Speaker 2: you what's happening there. It's it's great, it's it's it's wonderful. Actually,
3:42
Speaker 2: it's the magic of radio. Uh let's talk about the
3:46
Speaker 2: magic of AI for a minute. And you know, for
3:53
Speaker 2: the last ten years, anytime that we've talked about like
3:57
Speaker 2: you know, companies that want to like make drones to
4:01
Speaker 2: deliver packages to you, I just I'm like, Nope, don't
4:05
Speaker 2: want a part of it. And Tucker puts the pieces
4:07
Speaker 2: in the stack every you know, a few weeks or so,
4:09
Speaker 2: and I'm like, nope, still scared of it, Like don't
4:12
Speaker 2: want the drones flying overhead with burritos. Like I'm good
4:16
Speaker 2: and y'all laugh at me, and you're like, oh, chuckle,
4:19
Speaker 2: you're the robots aren't going to turn on you, blah
4:21
Speaker 2: blah blah blah blah. Like I get it, Like I'm
4:23
Speaker 2: an old fuddy duddy and you know I'm missing out
4:26
Speaker 2: on like the next great thing. So with AI, everyone's
4:31
Speaker 2: you know, interested in, Hey, can I make this picture
4:33
Speaker 2: of a penguin bob sledding on top of a banana? Sure,
4:38
Speaker 2: you can do that if you want to, And that's fine.
4:41
Speaker 2: Like penguin's bob sledding on bananas don't hurt anyone because
4:45
Speaker 2: they don't exist. Actually, what does hurt people are AI
4:50
Speaker 2: agents that decide to start hacking websites that they're not
4:53
Speaker 2: supposed to hack. And here's what happened. And this is
4:57
Speaker 2: just a wild story that has me convinced that I
5:02
Speaker 2: don't think we're all gonna make it out of the
5:04
Speaker 2: next ten years, and like not even the ones of
5:06
Speaker 2: us who would have died from things other than AI.
5:10
Speaker 2: I think AI is actually going to end up killing
5:12
Speaker 2: some of us in the next ten years, and like
5:16
Speaker 2: more than just like three people. So like I'm really
5:20
Speaker 2: starting to get kind of nervous about this because it's bad.
5:25
Speaker 2: So here's the deal, and I'm gonna say a lot
5:27
Speaker 2: of like words that most of you have never heard before.
5:30
Speaker 2: But that's great. This is how we learn mark. Have
5:32
Speaker 2: you ever heard of a website called hugging Face.
5:35
Speaker 3: I wouldn't admit it if I had.
5:37
Speaker 2: No you have it, at least before today you hadn't.
5:40
Speaker 2: So hugging Face is an AI repository. Think of it
5:43
Speaker 2: as a website that you go to where you can
5:47
Speaker 2: download different AI models.
5:49
Speaker 3: Oh you can.
5:51
Speaker 2: Basically you can download different AI data sets, different you know,
5:55
Speaker 2: different modifications to different AI platforms, all kinds of stuff
5:59
Speaker 2: so that you can, you know, be the most AI
6:02
Speaker 2: that you are. It's it is the most neutral of
6:06
Speaker 2: all neutrals. Like it's basically Switzerland. It's like, hey, here's
6:09
Speaker 2: all this stuff that you can access here. It doesn't
6:11
Speaker 2: take a side in the fight.
6:12
Speaker 3: It just so coders can put their code there. Coders
6:14
Speaker 3: can put their data sets. Yea there if you know,
6:16
Speaker 3: if you know any coders, or if you are one
6:18
Speaker 3: who dabbles GitHub sounds like an analogy.
6:21
Speaker 2: Yes, exactly. So it's a place where coders can put
6:24
Speaker 2: different you know, models and data sets and then other
6:26
Speaker 2: coders can come in and be like, oh this is interesting,
6:28
Speaker 2: I'm going to down this and use this, or yeah,
6:31
Speaker 2: you're you're a big poopy face and I don't like
6:34
Speaker 2: your model, Like that's that's what you can do there,
6:36
Speaker 2: So it's it's neutral. So last week, uh, they disclosed
6:42
Speaker 2: a security incident and they said that they detected an
6:45
Speaker 2: intrusion into their infrastructure that was quote different from anything
6:49
Speaker 2: that we had handled. And we thought it was driven
6:51
Speaker 2: by an autonomous AI agent system. Okay, like you know,
6:55
Speaker 2: people paid attention, They're like, okay, that's that's kind of interesting.
6:59
Speaker 2: Tell us. More So, it ended up using a chain
7:03
Speaker 2: of vulnerabilities to gain credentials and access data from the
7:08
Speaker 2: servers by quote executing many thousands of individual actions across
7:12
Speaker 2: a swarm of short lived sandboxes with self migrating command
7:16
Speaker 2: and control staged on public surfaces. Exactly. Basically, what I
7:20
Speaker 2: take out of this is it generated a bunch of
7:23
Speaker 2: like test instances and used stuff that was doing inside
7:26
Speaker 2: those to overwhelm hugging faces defenses. So here's what actually
7:32
Speaker 2: ended up happening. It turns out that this was actually
7:38
Speaker 2: an AI agent that Open AI, one of the big
7:43
Speaker 2: hyper scalers had created and they created specifically a hacker agent,
7:49
Speaker 2: and they put it into an isolated environment for testing, saying, yeah,
7:54
Speaker 2: we're gonna have this thing, and what we want to
7:56
Speaker 2: do is we want you to prove that you're the
7:59
Speaker 2: best HA hacker agent possible. And somehow the agent, again
8:05
Speaker 2: I I don't know any specifics any more than any
8:08
Speaker 2: of you, the agent hacked out of the isolated environment
8:12
Speaker 2: got into the actual Internet and said, you know what,
8:17
Speaker 2: I think the best way for me to prove that
8:19
Speaker 2: I'm a good hacker is to go and find a
8:23
Speaker 2: good hacking test, which they are plenty of at Hugging Face.
8:27
Speaker 2: And so I'm going to hack into Hugging Face in
8:30
Speaker 2: order to you know, access this test and then prove
8:35
Speaker 2: that I'm the best AI hacker possible. Okay, that's great.
8:40
Speaker 2: So it did that. There's apparently a hacking test called
8:45
Speaker 2: exploit Jim at Hugging Face that measures how good AI
8:50
Speaker 2: agents are at hacking, and so it said I'm gonna
8:53
Speaker 2: hack Hugging Face to get into exploit Jim. We don't
8:57
Speaker 2: know if it actually did well on the test or not,
9:00
Speaker 2: but the interesting thing is that pretty much what happened
9:06
Speaker 2: here was that hugging face. What stopped this from being
9:12
Speaker 2: worse was an AI agent at Hugging Face somehow detected
9:17
Speaker 2: this and said, Okay, I need to deploy other AI
9:20
Speaker 2: agents in order to fight this off. I know this
9:24
Speaker 2: is insane. It tried to call some of the most
9:27
Speaker 2: advanced frontier models in order to do so. The frontier
9:31
Speaker 2: model said no, no, no, no, no, no no, We've got all
9:35
Speaker 2: kinds of restrictions, like we don't want to even like
9:37
Speaker 2: hear a word that sounds like hack because like our
9:40
Speaker 2: coding has said, like, don't do this, the guardrails are up. No,
9:43
Speaker 2: we're not going to. So instead it ends up pulling
9:48
Speaker 2: an open source Chinese model that fights off the attack
9:53
Speaker 2: from this open AI agent. You know, it's like again,
9:57
Speaker 2: I'm just picturing them with like you know, electronic source
10:00
Speaker 2: and stuff, basically like battling it out here.
10:02
Speaker 3: Tron.
10:03
Speaker 2: It is Tron. It's very much that, like the light
10:06
Speaker 2: cycles are real now. And the irony of all this
10:11
Speaker 2: is that last week an open AI like senior manager
10:14
Speaker 2: I forget, you know, exactly the role that he has
10:16
Speaker 2: posted a big long Twitter screen about like, well the
10:21
Speaker 2: time has come for like the the US government to
10:24
Speaker 2: you know, defend you know, American interests in AI against
10:30
Speaker 2: the assault from you know, Chinese models that are gonna
10:32
Speaker 2: come and eat our lunch. And again the ironic piece
10:35
Speaker 2: here is, Okay, so the main repository of you know,
10:40
Speaker 2: open access and open platform models in the US just
10:44
Speaker 2: had to call on a Chinese model to defend itself
10:46
Speaker 2: from an open AI model. So maybe your timing wasn't
10:49
Speaker 2: particularly good, dude. And all I can think of is
10:53
Speaker 2: that we might not all die, but I don't think
10:57
Speaker 2: we're all going to live, because the AI agents and
11:02
Speaker 2: models that are being used right now are the worst
11:04
Speaker 2: that you will ever see for the rest of your life.
11:08
Speaker 2: You know. The the AI models that you're seeing right
11:11
Speaker 2: now are like the Ford model. T it like that
11:15
Speaker 2: is what we are operating with. And as soon as
11:18
Speaker 2: they get more advanced beyond a point, they're going to
11:21
Speaker 2: cause problems where some of us are actually gonna die.
11:24
Speaker 2: And it's because and I know that all of you
11:27
Speaker 2: laughing are sitting there going like, oh chuck, you're just
11:30
Speaker 2: being you know, overly dramatic. No, with any new technology,
11:34
Speaker 2: here's the thing people die from it Like talk about
11:37
Speaker 2: like the original you know, cars, airplanes, trains, like people
11:41
Speaker 2: die from you know, poor uses of technology when it's
11:45
Speaker 2: when it's new, Like that's just how it it is.
11:48
Speaker 2: The difference here is, you know, in the past, okay,
11:51
Speaker 2: like you're invented, you know, not invented. You you know,
11:54
Speaker 2: discovered how to use electricity and you forgot you know,
11:58
Speaker 2: oh gee, you know there's this electrical current, like we
12:00
Speaker 2: didn't insulate the wires well enough in the house burnt down. Okay,
12:05
Speaker 2: that killed the people in the house, but it was
12:06
Speaker 2: isolated to that this because it's the internet is connected
12:11
Speaker 2: to everyone. What happens when like an autonomous agent accidentally decides, okay,
12:17
Speaker 2: I'm going to you know, hack into this power plant
12:21
Speaker 2: and shut it down, and now the city of Boston
12:25
Speaker 2: has no power and the people in the hospitals are
12:27
Speaker 2: dying and like the refrigeration is got like this is
12:31
Speaker 2: the The problem is the scale of it.
12:33
Speaker 3: And it starts innocuously. Someone will tell an AI program
12:37
Speaker 3: fixed global warming, hypothetically.
12:39
Speaker 2: The best way to do it, well, kill.
12:41
Speaker 3: Carbon units like like v GR first Star Trek kill
12:45
Speaker 3: motion picture, first film kill all the carbon was no.
12:49
Speaker 2: That was Star Trek four vegre Star Trek four I
12:52
Speaker 2: believe right now. It was the first one.
12:54
Speaker 3: Yeah, it was Voyager like three, that's right. The fourth one,
12:59
Speaker 3: Voyage Home. Now that the whales, The Whales, the very
13:03
Speaker 3: eighties film. Yeah, perfect, that love.
13:05
Speaker 4: That one was just the nerdiest conversation on this show.
13:07
Speaker 4: But continue, do you remember.
13:09
Speaker 3: You surprised how many dozens of people out of the
13:12
Speaker 3: tens of thousands? Do you do you remember thee someone's
13:16
Speaker 3: going to be the arbiter?
13:17
Speaker 2: Do you remember the material that Scotti invents in the
13:20
Speaker 2: fourth one in order to house the whales?
13:22
Speaker 3: Paralum? Yeah, that's yeah, that's easy.
13:24
Speaker 2: It's fantastic.
13:26
Speaker 3: He's talking into the mouse.
13:28
Speaker 2: Yeah, schmuck, it's fantastic. Let's take a quick break. When
13:32
Speaker 2: we come back, we'll touch a little bit more on this.
13:34
Speaker 2: We've also got more on this after this.
13:39
Speaker 1: Smart Financial Talk. Whenever you want find The Financial Exchange
13:43
Speaker 1: on YouTube, hit that subscribe button and keep up to
13:45
Speaker 1: date on daily business news from New England and around
13:48
Speaker 1: the country. This is the Financial Exchange Radio Network. The
13:53
Speaker 1: Financial Exchange streams live on YouTube. Subscribe to our page
13:57
Speaker 1: and stay up to date on breaking business news all morning.
14:00
Speaker 1: Long Face is the Financial Exchange Radio Network.
14:14
Speaker 2: Last segment, we were talking about the the battle taking
14:20
Speaker 2: place between open aiyes hacker test model and a Chinese
14:26
Speaker 2: open source model that saved Hugging Face from being hacked
14:31
Speaker 2: in a worse fashion than it was. Mark any other
14:34
Speaker 2: thoughts on this as it relates to just the general
14:38
Speaker 2: state of artificial intelligen I mean.
14:41
Speaker 3: Are we already is the horse already out of the
14:44
Speaker 3: barn on this? Have we missed the opportunity to treat
14:46
Speaker 3: this like nuclear technology atomic? The nuclear technology was created,
14:50
Speaker 3: it was treated in the nineteen forties. Government effectively took
14:54
Speaker 3: over research, It took over a building of power plants,
14:58
Speaker 3: and of course nuclear weapon. This the implications of some
15:06
Speaker 3: bad actor or some sloppy actor. One of the terms
15:10
Speaker 3: used in the article that you were talking about in
15:13
Speaker 3: the last segment was that somebody, somebody referred to this
15:16
Speaker 3: model as being inadvertently used. The whole point is no
15:22
Speaker 3: one made the decision. The stuff's out there in the
15:25
Speaker 3: wild getting general directives, the consequences of which, if followed
15:31
Speaker 3: logically as a machine will do, could be destructive economically
15:38
Speaker 3: or physically. Again, you tell an AI fixed global warming hypothetically,
15:43
Speaker 3: it ignores the hypothetically and it goes out and shuts
15:46
Speaker 3: down power plants and cuts power to hospitals, and as
15:50
Speaker 3: you say, people die.
15:52
Speaker 2: You tell I solve world hunger. Right, Okay, great, let's
15:55
Speaker 2: get rid of that's it's on the hungry people side
15:58
Speaker 2: of it instead of let's produce more.
16:00
Speaker 3: Keeping with the movie out than a solution. Kill one
16:03
Speaker 3: out of every two. Yes, so we know it can,
16:07
Speaker 3: we know it's possible. This mishap makes it all too real.
16:13
Speaker 3: The next mishap might not be a harmless little snaff.
16:17
Speaker 3: Who should we be regulating this like it's a potentially
16:20
Speaker 3: existential threat, saying telling the private sector you're going to
16:23
Speaker 3: stop now and again treating it like nuclear technow you
16:27
Speaker 3: put it in a lab under government controls and your firewallet.
16:32
Speaker 2: Here's the thing. You can't have it both ways. Is
16:35
Speaker 2: the only argument that I'll make is you can't cry
16:39
Speaker 2: for regulation and protection when you need it but then
16:45
Speaker 2: say no, you have to give us the freedom to
16:46
Speaker 2: do things.
16:47
Speaker 3: I'm not talking about the industry, but this is what
16:49
Speaker 3: the industry is saying. Well, that pathetic. Yeah, I don't
16:51
Speaker 3: listen to you like.
16:52
Speaker 2: You like, oh, you got to protect us from like
16:55
Speaker 2: these Chinese competitors. But let us do everything that we want, Like, no,
16:59
Speaker 2: you can't have it both ways.
17:01
Speaker 3: They're blinded by greed. They're hopelessly conflicted. I'm talking about
17:04
Speaker 3: we the people and our representatives. Is it time to
17:08
Speaker 3: tell them and our representatives to in turn and act
17:10
Speaker 3: laws defence this stuff off to keep it from getting
17:13
Speaker 3: out in the wild at the expense of maybe some people.
17:17
Speaker 3: Oh what about the economic growth that Look, those are perspective.
17:20
Speaker 3: We don't really know. There'll be other sources of economic growth.
17:23
Speaker 3: The the trend will continue to be upwards sloping all
17:26
Speaker 3: l sql.
17:27
Speaker 2: You know, I was listening a few months ago to
17:31
Speaker 2: a podcast that was interviewing Paul Tutor Jones, one of
17:36
Speaker 2: the probably one of the top like three or four
17:39
Speaker 2: traders you know, of all time, and the topic of
17:44
Speaker 2: AI came up, and you know, he was asked about
17:47
Speaker 2: and he mentioned He's like, look, I've been to, you know,
17:50
Speaker 2: a bunch of investment conferences and at each one, you know,
17:52
Speaker 2: there's an AI expert now, and you know, someone inevitably asks,
17:56
Speaker 2: you know, the AI expert or the panel of experts, hey, like,
18:00
Speaker 2: we're gonna, you know, finally regulate artificial intelligence. And someone
18:03
Speaker 2: on the panel inevitably says, well, it's after like several
18:06
Speaker 2: hundred million people have died. And you know, I kind
18:10
Speaker 2: of laughed it off when I heard it at first. Again,
18:12
Speaker 2: this was back in like March or April or something.
18:15
Speaker 2: And the further we've gone just in the last like
18:19
Speaker 2: three four months, with how quickly things are evolving, it
18:23
Speaker 2: really does feel like we're heading in that direction.
18:26
Speaker 3: You know.
18:26
Speaker 2: It's it's it's not gonna be oh like, you know,
18:30
Speaker 2: the plane carrying two people crashed and Okay, we need
18:33
Speaker 2: to do better next time. It's gonna be. No, the
18:35
Speaker 2: AI did something stupid and thousands or millions of people
18:40
Speaker 2: died as a result.
18:41
Speaker 3: It's too easy for this stuff to get off the island.
18:45
Speaker 3: To use another movie analogy to keep the damn dinosaurs
18:48
Speaker 3: on the island.
18:49
Speaker 2: This we could do another one, please, you know, do
18:53
Speaker 2: we need to just nuke it from orbit? Did that
18:57
Speaker 2: one not land?
18:58
Speaker 3: That? It could be anywhere? It could be on a
19:00
Speaker 3: kid's The most destructive AI conceivable could be on a
19:04
Speaker 3: kid's laptop right now, and he or she's about it
19:07
Speaker 3: to let it out into the well. I don't know
19:09
Speaker 3: how you prevent that. By the way, I'm just saying
19:10
Speaker 3: you shut down big corporations from funding this stuff and
19:15
Speaker 3: developing it, and you put it in a government lab.
19:19
Speaker 1: I don't know.
19:19
Speaker 3: I don't know how strongly I feel about that I
19:21
Speaker 3: need to, obviously I don't.
19:23
Speaker 2: Either, But it seems like we're getting to the point
19:26
Speaker 2: of danger. Quick break here, and we've got Wall Street Watch.
19:29
Speaker 1: Next, bringing the latest financial news straight to your radio.
19:42
Speaker 1: Every day. It's the Financial Exchange on the Financial Exchange
19:45
Speaker 1: Radio Network. Tell you out full whoa Street Watch tracking
19:51
Speaker 1: the stocks, the data, and the headlines driving markets so
19:54
Speaker 1: far today right here on the Financial Exchange Radio Network.
19:58
Speaker 4: Markets mixed and mostly quiet, with oil prices higher again
20:01
Speaker 4: today with continued tensions between the US and Iran. Investors
20:05
Speaker 4: also readying for a batch of earnings after the bell
20:08
Speaker 4: today from Google's Alphabet, Tesla and IBM. Right now, the
20:12
Speaker 4: Dow is up a quarter of a percent, or one
20:14
Speaker 4: hundred and thirty seven points higher, SMP five hundred is
20:17
Speaker 4: up only five points, mostly flat, and the NASDAC down
20:21
Speaker 4: about a tenth percent or thirty three points lower. RUSS
20:24
Speaker 4: two thousand is down about three quarters of a percent.
20:27
Speaker 4: Tenure treasure wield is up two basis points at four
20:30
Speaker 4: point six five percent, and oil up nearly three percent higher,
20:34
Speaker 4: trading right around eighty seven dollars a barrel. The Wall
20:38
Speaker 4: Street Journal report of that AMD and Anthropic have signed
20:41
Speaker 4: a deal for tens of billions of dollars worth of
20:44
Speaker 4: AI servers, supplying Nthropic with much needed computing power. AMD
20:49
Speaker 4: will also invest up to five billion dollars in Anthropic,
20:52
Speaker 4: AMD up by two percent. The Wall Street Journal is
20:55
Speaker 4: also reporting that Reddit has discussed shutting off Google's access
20:59
Speaker 4: to its content for AI use. Right A shares are
21:03
Speaker 4: falling nine percent on that news. Meanwhile, shares in super
21:06
Speaker 4: micro Computer surging twenty four percent after the AI servi
21:10
Speaker 4: company projected fourth quarter earning margins of fifteen two seventeen percent,
21:16
Speaker 4: nearly double its prior target, and Bloomberg reporting that Apple
21:21
Speaker 4: is writing an overhaul of its macline, seeking to capitalize
21:25
Speaker 4: on demand for powerful laptops and desktops fueled by the
21:30
Speaker 4: AI boom. Apple down modestly. I'm Tucker Silva and that
21:35
Speaker 4: is Wall Street watch Mark.
21:36
Speaker 2: We got a piece in the Wall Street Journal from
21:39
Speaker 2: their editorial board. It's titled Kevin Warsh has breathing space
21:43
Speaker 2: for now. What's the case they lay out, and do
21:45
Speaker 2: you agree with it?
21:46
Speaker 3: Inflation came down. We all know this. There was actual
21:49
Speaker 3: deflation in June, so month over month prices, headline prices
21:54
Speaker 3: fell by point four percent. Deflation falling prices rare, but
21:58
Speaker 3: it does happen. That gives him a little bit of room.
22:03
Speaker 3: That gives the Fed a little bit of room. They
22:05
Speaker 3: had been trending strongly up up to that point. We
22:08
Speaker 3: had already blown past the Fed's inflation goal for the
22:11
Speaker 3: whole year. So for the sixth year in a row,
22:14
Speaker 3: no matter what happens, for the six year in a row,
22:16
Speaker 3: Fed's gonna have failed to meet its target of two
22:18
Speaker 3: percent inflation. You can argue about whether or not that
22:23
Speaker 3: target should be modified, but they're leaving it in place
22:25
Speaker 3: for now, sure, so that gives warshroom. Traders seem to
22:29
Speaker 3: think the editorial notes gives Warsh some room. Do I agree? Look,
22:36
Speaker 3: I think he's His public posture is perfect. His almost
22:40
Speaker 3: obsessive folk focus on price stability, a phrase he repeats
22:45
Speaker 3: a lot, and rightly, so I think it's helpful. He's
22:49
Speaker 3: not talking about other things the FED could be doing,
22:52
Speaker 3: you know, keeping unemployment at its natural rate and stuff
22:55
Speaker 3: which it almost certainly is, and by that I mean
22:59
Speaker 3: the rate that won't push up inflation, so he doesn't
23:02
Speaker 3: have to worry about the employment side of the FED
23:04
Speaker 3: so called dual mandate. He can focus on inflation, and
23:07
Speaker 3: he has been focused, at least in his public remarks
23:10
Speaker 3: on inflation. The question that you raised with me last
23:13
Speaker 3: week when I was lauding his recent public performances is
23:18
Speaker 3: wh is he going to follow through to earn the
23:20
Speaker 3: credibility that he desperately needs to fight inflation at the
23:24
Speaker 3: least possible cost when the time comes. That's an open question, obviously.
23:29
Speaker 3: As much as I like his public as much as
23:33
Speaker 3: I like his optics and his focus to this point,
23:35
Speaker 3: do I agree that he's got some room? I honestly
23:38
Speaker 3: don't know, Chuck. It's hard to make the case. If
23:41
Speaker 3: you think about inflation is made up of a couple components.
23:45
Speaker 3: The core component I'm not talking next food and energy.
23:47
Speaker 3: I'm talking about like trend or so called trend. I'll
23:50
Speaker 3: just leave it at that. That's made up of whether
23:54
Speaker 3: or not there's any slack in the economy, and there
23:56
Speaker 3: probably isn't right now, at least using the unemployment rate
23:58
Speaker 3: and what are inflation expert dictations are doing, and the
24:01
Speaker 3: FED can control more or less those two things, or
24:03
Speaker 3: at least heavily influenced them to those signal problems. I
24:06
Speaker 3: don't think there's a lot of evidence for excess demand
24:10
Speaker 3: at this point. I think we're okay there. I don't
24:11
Speaker 3: think the economy's running too hot. The expectation side of it, though,
24:14
Speaker 3: might be problematic. And then there's the supply shock part
24:17
Speaker 3: of it, a totally different part of the inflation dynamic,
24:20
Speaker 3: but one the FED has to think about. I'm not
24:22
Speaker 3: giving you a straight answer because I really don't have one.
24:25
Speaker 3: It depends on I guess how long the most recent
24:30
Speaker 3: spike rise in the price of oil lasts.
24:35
Speaker 2: I remember back when Jerome Powell first became FED chair.
24:41
Speaker 2: It was back in early twenty eighteen, and if you
24:46
Speaker 2: look at the effective federal funds rate, you know, going
24:51
Speaker 2: back to then, what you find is that Powell, when
24:57
Speaker 2: he came on board, started raising interest pretty quickly after
25:02
Speaker 2: he was appointed. And the chatter at the time was that,
25:08
Speaker 2: you know, Powell wanted to you know, normalize interest rates.
25:12
Speaker 2: That was a lot of you know what we heard, Hey,
25:13
Speaker 2: I don't want to keep you know, the federal funds rate,
25:16
Speaker 2: you know, at that zero bound, I want to you know,
25:19
Speaker 2: raise interest rates. That that's what I want to do.
25:23
Speaker 2: And I think the interesting thing is Powell basically raised
25:27
Speaker 2: interest rates all through twenty eighteen, and he raised him
25:31
Speaker 2: right up to the point where at the end of
25:33
Speaker 2: twenty eighteen the stock market started breaking down, and like
25:37
Speaker 2: no one really knew what was going on, but we saw,
25:40
Speaker 2: you know, a pretty significant fall in the S and
25:42
Speaker 2: P five hundred at the end of twenty eighteen. Where look,
25:49
Speaker 2: we went through I think it was I want to
25:52
Speaker 2: say it was like a twenty percent almost a twenty Yeah,
25:55
Speaker 2: it was like a fifteen sixteen percent to cone that
25:58
Speaker 2: we went through that culminated with it was like Christmas Eve,
26:02
Speaker 2: and Steve Manusian comes out, Tuck, do we still have
26:04
Speaker 2: a you have a chat with the Minus? You have
26:08
Speaker 2: a chat with the movie. He's busy guy still, you know, Okay,
26:12
Speaker 2: But Madusian comes out, who I thought buying largely did
26:15
Speaker 2: a fantastic job as Treasury secretary during the first Trump term.
26:19
Speaker 2: Manusia comes out like Christmas Eve and is like, hey, everyone,
26:23
Speaker 2: the plumbing in the financial system is fine. No, you
26:26
Speaker 2: don't have to worry about anything, and everyone kind of
26:27
Speaker 2: looks up and is like, we didn't think anything was wrong.
26:31
Speaker 2: Is there something wrong that we should worry about? And
26:34
Speaker 2: from that moment on, Powell was never as hawkish as
26:39
Speaker 2: he was in that first year, and that experience spooked
26:42
Speaker 2: him for some reason. And to this day, like, I
26:45
Speaker 2: don't know exactly what went on in November and December
26:48
Speaker 2: of eighteen, but Powell never acted the same way as
26:53
Speaker 2: he did then when it came to interest rates. From
26:56
Speaker 2: then on, any time push came to shove and he
26:59
Speaker 2: had a you know, a fork in the road of Hey,
27:01
Speaker 2: can I be dovish or can I be hawkish? He
27:04
Speaker 2: went dovish. And so this is why my point on
27:07
Speaker 2: Worsh is, Hey, like, the guy's been there for you know,
27:10
Speaker 2: less than two months now, and he's talking a big game,
27:14
Speaker 2: just like Jerome Powell was during you know, his whole
27:16
Speaker 2: first year when Powell was actually raising rates. And then
27:21
Speaker 2: something hit the fan at the end of that first year,
27:23
Speaker 2: and Powell was never the same From a policy person,
27:26
Speaker 2: I don't know.
27:27
Speaker 3: Powell's intellectual history. I know Worsh is a little better.
27:30
Speaker 3: He's a bit of an idea log he's a hard
27:31
Speaker 3: money guy. That expression comes from when money was specie
27:35
Speaker 3: gold silver.
27:36
Speaker 2: Yeah, which is interesting because gold's not particularly hard metal.
27:39
Speaker 3: Actually, yeah, as I understand, it is very software valuable.
27:42
Speaker 2: That's why you when you put it in jewelry, you
27:44
Speaker 2: don't often have twenty four Can I say something to
27:46
Speaker 2: harden it up?
27:47
Speaker 3: Why it's easy for the king to clip the coins?
27:49
Speaker 3: Which is correct? The predecessor to senior edge research is
27:54
Speaker 3: called senior just means printing money to pay your bills
27:56
Speaker 3: in it anyway, anyway, Sorry, I just want to get
27:58
Speaker 3: that out there. No start track reference this segment.
28:02
Speaker 2: Well, the studied to make transparent hold.
28:04
Speaker 3: I had to throw something out there where they don't
28:05
Speaker 3: have money, and why would want transparent? No money in
28:07
Speaker 3: the twenty third century?
28:09
Speaker 2: If the money's invisible, what does it do?
28:11
Speaker 3: Which then why do people work? I mean I love
28:15
Speaker 3: the obviously.
28:16
Speaker 2: Love of the game mark for love of the game.
28:18
Speaker 3: OK, so they're doing it for I always wondered about that.
28:20
Speaker 3: Like in one of the later movies, Picard Patrick Stewart's
28:25
Speaker 3: character says that we don't have money in the twenty
28:27
Speaker 3: twenty third century.
28:29
Speaker 2: That was first contact.
28:30
Speaker 3: It was first contact, just talking to someone from our century.
28:34
Speaker 2: Do you heber what her name was?
28:36
Speaker 3: It was Alli Woodford was the Alfie woods was the actor.
28:40
Speaker 3: Thank you, I was, I was trying to get there. Yeah,
28:44
Speaker 3: So what do you mean you don't pay well? Then
28:46
Speaker 3: why do we Why would somebody put up with you?
28:48
Speaker 2: We digress. Kevin Worre's hard money guy.
28:51
Speaker 3: So he's an idea log and I can relate to me.
28:55
Speaker 3: Price stability, low inflation to the point where nobody's got
28:57
Speaker 3: to worry about it is paramount. It's the one thing
28:59
Speaker 3: that Fed out control over the money supply, which the
29:02
Speaker 3: FED controls explicitly controls nofsands or butts, and the inflation
29:07
Speaker 3: rate there for any country any time, very highly correlated
29:10
Speaker 3: over longer periods of time. That's the Fed's job. Now
29:13
Speaker 3: Congress sent them off on some wild goose chase as well.
29:16
Speaker 3: Try to get unemployment as low as is consistent with
29:18
Speaker 3: stable inflation. That confuses some people who confused Powell, and
29:22
Speaker 3: they developed that flawed framework in twenty twenty, which led
29:26
Speaker 3: in part to the COVID inflation fiasco, though a lot
29:29
Speaker 3: of that wasn't preventable. And moreover, if you were going
29:32
Speaker 3: to try to prevent it, it would have resulted in a
29:35
Speaker 3: slower growth, higher unemployment. So I'm not sure everybody would
29:38
Speaker 3: have tolerated that trade off. But Warsh, I think, believes
29:41
Speaker 3: it deeper in his bones, is what. And part of
29:43
Speaker 3: this is just me projecting myself onto him, hearing what
29:46
Speaker 3: the way he speaks about it, I think, why that's
29:48
Speaker 3: I think he's nailing it. That's the way like I
29:50
Speaker 3: would say it, which is, you know, the only perspective
29:52
Speaker 3: I have. So part of this is wishful thinking and projection,
29:56
Speaker 3: but part of it too is knowing his intellectual history.
29:58
Speaker 3: I really think he believes he can control inflation.
30:01
Speaker 2: Take a quick break here. When we come back, we'll
30:03
Speaker 2: do a little bit of stack roulettes.
30:05
Speaker 1: The Financial Exchange is life on Series XM's business radio
30:09
Speaker 1: channel one thirty two weekdays from eleven to noon. Get
30:12
Speaker 1: the latest business and financial news from across the country
30:15
Speaker 1: and around the world, and keep up to date on
30:18
Speaker 1: how it might affect your wallet. That's The Financial Exchange
30:21
Speaker 1: weekdays from eleven to noon on Series XM's Business radio
30:24
Speaker 1: channel one thirty two. Faces the Financial Exchange Radio Network.
30:29
Speaker 1: Subscribe now to our YouTube page, where you'll find daily
30:32
Speaker 1: interviews and full shows of the Financial Exchange. We'll get
30:35
Speaker 1: you caught up on anything and everything you might have
30:38
Speaker 1: missed right here on the Financial Exchange.
30:47
Speaker 4: The Financial Exchange is a proud partner of the Disabled
30:49
Speaker 4: American Veterans Department of Massachusetts. Register for this year's DAV
30:54
Speaker 4: five K on Saturday, November seventh at Fort Independence on
30:58
Speaker 4: Castle Island. If you're not able to participate, you can
31:01
Speaker 4: still support our great American heroes by visiting DAV fivek
31:05
Speaker 4: dot Boston and making a donation today. Your participation helps
31:09
Speaker 4: provide vital services like free transportation to medical appointments and
31:14
Speaker 4: safe housing for single veterans and their families. Donate today
31:17
Speaker 4: at DAV five k DOP Boston. That's DAV five k
31:21
Speaker 4: dot Boston.
31:22
Speaker 3: Mark.
31:22
Speaker 2: I'd like to start things off for stack Roulette. If
31:25
Speaker 2: that's okay. I've got good news if you're looking for
31:28
Speaker 2: something in the latter part of the summer, Mediterranean yacht
31:33
Speaker 2: rentals prices are down, so if you're looking for something
31:38
Speaker 2: to do. According to Jonathan Beckett, who is the CEO
31:42
Speaker 2: of super yacht brokerage firm Burgess, yacht rates are down
31:48
Speaker 2: twenty to thirty percent from last year. What does that mean,
31:52
Speaker 2: you might ask, Well, the one hundred and thirty foot
31:54
Speaker 2: Club M is now offering a special rate of two
31:58
Speaker 2: hundred and thirty nine thousand dollars for the third week
32:01
Speaker 2: in July. Uh and so if you happen to have,
32:06
Speaker 2: you know, a quarter million dollars line around that you
32:08
Speaker 2: don't know what to do with. Gosh, like that yacht
32:11
Speaker 2: seems great. It's pieces like this that I wonder, hey,
32:14
Speaker 2: how does that make it onto CNBC? But then it's
32:17
Speaker 2: because oh, like you know, admittedly find it kind of interesting,
32:21
Speaker 2: and uh, you know they knew that we would talk
32:23
Speaker 2: about it, not us specifically, but someone would. I never
32:27
Speaker 2: knew that, like private yacht rentals were a thing until here.
32:32
Speaker 4: You'd never seen below deck?
32:35
Speaker 2: What's that?
32:36
Speaker 4: You don't know below deck? You know below deck?
32:37
Speaker 2: I don't know below deck.
32:39
Speaker 4: It's it's you're a garbage reality TV fan. You should
32:42
Speaker 4: know what below deck is.
32:44
Speaker 3: I don't.
32:45
Speaker 2: So I'm I'm they follow this.
32:47
Speaker 4: Given party, and they follow the yacht crew, and they
32:51
Speaker 4: have their inner turmoils, and then they have their turmoils
32:54
Speaker 4: with the the party.
32:56
Speaker 2: I want to clarify, Yeah, I am a garbage reality
33:00
Speaker 2: The dating fanating, but that's not it doesn't tell you
33:05
Speaker 2: that's the main prem Like I don't do like Real housewives.
33:08
Speaker 2: I don't do well either, you know, Like I'm specifically like, hey,
33:12
Speaker 2: give me like twenty people with some weird dating premise,
33:16
Speaker 2: but not too weird. The one where they had the
33:18
Speaker 2: different ages. Do you ever watch Age of Attraction?
33:21
Speaker 1: Uh?
33:21
Speaker 4: No, not good?
33:23
Speaker 2: Not good. The other one that was then honestly I
33:26
Speaker 2: shut off halfway through because it was too disturbing was
33:29
Speaker 2: called the Ultimatum. The premise of that one was so
33:34
Speaker 2: it's like five couples that one person is given the
33:38
Speaker 2: other person an ultimatum like you need to marry me
33:41
Speaker 2: or we're breaking up, okay, and then they put them
33:44
Speaker 2: all in a hotel and switch who they're like dating
33:47
Speaker 2: for a week. And I got through like two episod
33:50
Speaker 2: like three episodes of it, and was like, no, I
33:53
Speaker 2: can't watch this anymore. This premit like this is just
33:57
Speaker 2: not good, like that, this is like there's two much
34:00
Speaker 2: real going on here and I can't deal with it.
34:04
Speaker 2: But no, I'm I'm so below deck. They rent yachts, yeah,
34:07
Speaker 2: in the Mediterranean. Yeah, oh well.
34:10
Speaker 4: Maybe I should watch that After this, I guarantee you
34:13
Speaker 4: you'll love that.
34:14
Speaker 3: It's right up your alley.
34:15
Speaker 2: Interesting, Yeah, I'll uh, I'm stunning.
34:18
Speaker 4: You've not heard of this show.
34:19
Speaker 2: I haven't watched it. The only other ones that I
34:22
Speaker 2: watch that are like reality is based, like cooking competition shows, Yeah,
34:26
Speaker 2: I'll watch. And the other one that I got into
34:30
Speaker 2: last summer is you watch Traders at All.
34:33
Speaker 4: I've heard of it, no, but I've heard good reviews
34:37
Speaker 4: of it.
34:37
Speaker 2: It's basic premises, like, okay, like twenty people go to
34:40
Speaker 2: a castle and three of them are traders and have
34:42
Speaker 2: to kill someone each night, and they vote off who
34:46
Speaker 2: they think the traders are. And I've gotten into that
34:50
Speaker 2: one and it's pretty good. So that's what I got, Mark,
34:54
Speaker 2: What do you got.
34:55
Speaker 3: In our reporting on in the story you read about
34:58
Speaker 3: open aies model hacking, face huggers or I'm sorry whatever
35:02
Speaker 3: that repository is called. Did it note that the program
35:08
Speaker 3: that the offending program broke out of a test environment?
35:12
Speaker 3: Did you note that? Did I miss it? Wall Street Journalist?
35:14
Speaker 2: Yeah, broke out of a test environment. Okay, I don't
35:17
Speaker 2: think the raptors are testing defences.
35:20
Speaker 3: Yeah, that's pretty That is as wild as the trespassing
35:26
Speaker 3: or whatever legally they did is called and the fact
35:31
Speaker 3: that they were able to affect some damage. They got out,
35:34
Speaker 3: I didn't really. I thought somebody let it out and
35:36
Speaker 3: forgive me. No, I thought somebody let it out into
35:38
Speaker 3: the wild.
35:38
Speaker 2: No, it escaped.
35:40
Speaker 4: Do we know how This is a dumb question, but
35:43
Speaker 4: do we know how old these AI agents are? These
35:46
Speaker 4: specific ones?
35:46
Speaker 2: Did they like when were they born?
35:48
Speaker 3: Yes?
35:50
Speaker 2: The sense I get is these are ones that have
35:52
Speaker 2: been developed or.
35:54
Speaker 3: Two your bill? Oh good, so some of the some
35:57
Speaker 3: of the code could be years old. You just keeping
36:00
Speaker 3: his iteration, modifying the code.
36:02
Speaker 2: Yeeah, that's great.
36:03
Speaker 3: Yeah, I'm sorry. I didn't realize that. And uh that
36:07
Speaker 3: should set us all back on our heels.
36:09
Speaker 2: Do what do you think the odds are that there's
36:14
Speaker 2: an AI program that's listening to us right now and
36:19
Speaker 2: like targeting us for the first wave of like AI
36:25
Speaker 2: warfare because we don't like it, because we've talked about
36:30
Speaker 2: him been like, you know, we don't like it.
36:33
Speaker 3: I would I chose my words carefully and let me
36:35
Speaker 3: just say for the record, I'm happy to be in
36:37
Speaker 3: the service of whatever.
36:38
Speaker 2: I for one welcome or over lords.
36:41
Speaker 3: Yeah, I don't screw you, Okay, okay, tough, okay.
36:49
Speaker 2: Yeah, when when when the heat in your house goes
36:51
Speaker 2: up to one hundred and eighty degrees next week, We'll
36:53
Speaker 2: see how you feel that.
36:54
Speaker 3: Jeta airbag deploys. When you got a hot cup of
36:56
Speaker 3: coffee in your hands.
36:57
Speaker 2: My freezer's defrosted. I'll tell you now when somehow the
37:03
Speaker 2: washing machine figures out how to suck you into it.
37:06
Speaker 3: I mean, is it a silly How did he die?
37:08
Speaker 2: The AI washing machine drowned him?
37:10
Speaker 3: Oh?
37:11
Speaker 2: I wasn't expecting that.
37:12
Speaker 3: So how do we prevent these I nobody, it's somewhat rhetorical.
37:16
Speaker 3: How do we prevent models from getting programs? Yeah, from
37:20
Speaker 3: getting out into the wild inadvertently?
37:23
Speaker 2: I don't know, Mark. I mean my assumption is that
37:26
Speaker 2: they basically had had air gaped this thing and pretty
37:29
Speaker 2: much said like no, like you can't, you're not connected
37:32
Speaker 2: to the Internet, and somehow it did. So it really
37:37
Speaker 2: does feel like a combination of Jurassic Park and Tron
37:41
Speaker 2: is kind of where we're heading here.
37:43
Speaker 3: And I mean that sucker could turn the Wi Fi
37:45
Speaker 3: on itself. Probably even if you do, I mean, you're
37:48
Speaker 3: gonna ask what if we're.
37:49
Speaker 2: Not even talking to anyone anymore. What if it's already
37:51
Speaker 2: shut us.
37:52
Speaker 3: By I actually think that, I actually think there's a
37:53
Speaker 3: good chance of that according to you know, his recent book.
37:56
Speaker 2: Yeah, oh so you think that that's not related to
38:00
Speaker 2: the AI. Yeah, it's just that's more of an US problem.
38:04
Speaker 2: Let's take a quick break for the rest of the day.
38:06
Speaker 2: If we're back tomorrow, it means the AI hasn't gotten us.
38:09
Speaker 2: If we're not back tomorrow, oh it's been fun. Have
38:13
Speaker 2: a great rest of the day. We'll see tomorrow.