NOTICIAS IA: El terremoto político en USA que amenaza el futuro de la IA — Transcript
Full transcript
- 0:01Hello everyone, I'm John Hernandez, and
- 0:03here I am for another week with your AI
- 0:05news video, because wow, what a massive
- 0:07shake-up we've had in the United States
- 0:09that has completely changed the
- 0:11artificial intelligence landscape in
- 0:12barely 24 hours.
- 0:15It's absolute madness, which obviously
- 0:18has to do with the U.S. midterm
- 0:20elections.
- 0:21I'm going to start the recording because
- 0:23today I have to explain a lot of
- 0:24important things so your agents can stay
- 0:26up to date. The truth is that what has
- 0:28happened in the United States is
- 0:30unprecedented and was not expected.
- 0:33This week were the midterm elections, as
- 0:35they are called in the U.S., which are
- 0:37basically elections held to choose or
- 0:39renew the Congress and the Senate.
- 0:43These are the two chambers that enable
- 0:45all legislation.
- 0:47The President of the government, in this
- 0:49case the United States, can make many
- 0:51decisions, but everything having to do
- 0:53with budgets, appointing people to
- 0:55high-level positions, and so on, must
- 0:57pass through these chambers, so the fact
- 0:59that these chambers change can greatly
- 1:01affect how things can be done within the
- 1:03government.
- 1:05Well, what has happened, to give you an
- 1:07idea, is that Donald Trump has lost by a
- 1:09landslide.
- 1:11Basically, the voters have punished him
- 1:13for many things, mainly the war in Iran
- 1:15and the rise in oil prices, but also,
- 1:17obviously, for others that affect us
- 1:19more directly as an audience of this
- 1:21channel.
- 1:23The fact is that artificial intelligence
- 1:25has played a fairly important role in
- 1:27the result. What has happened, so you
- 1:29get an idea, had practically never
- 1:31happened in all of history.
- 1:34In areas and states where traditionally
- 1:36those from Trump's party had victory
- 1:39almost guaranteed, they have lost to the
- 1:41Democrats.
- 1:43Trump is a Republican, and basically,
- 1:45the Democrats have given them a beating
- 1:47everywhere. And what about the part that
- 1:50affects us?
- 1:51Well, the Democrats are capitalizing on
- 1:54everything they can to take votes away
- 1:56from the Republicans; mainly the price
- 1:58of oil, prices, and inflation in general
- 2:01is what has especially marked this
- 2:03punishment vote against Donald Trump,
- 2:05who is going to have a tremendously
- 2:07difficult time continuing to govern for
- 2:10the two years he has left in his term,
- 2:12since he would basically have to pass
- 2:14almost anything through the enemy, the
- 2:17Democrats,
- 2:18to get his actions approved.
- 2:20And obviously, they are not going to let
- 2:23him work.
- 2:24Now, regarding the part that concerns us
- 2:26, the audience of this channel, which is
- 2:28artificial intelligence, this could have
- 2:30enormous consequences for the
- 2:32development of global AI; because,
- 2:33evidently, the United States is the
- 2:35dominant player along with China, and
- 2:37the fact that things are happening,
- 2:39politically speaking, in the United
- 2:40States could completely change the
- 2:42landscape.
- 2:43And what is the perception of those
- 2:45Democrats regarding artificial
- 2:47intelligence? Well, take a look.
- 2:49Thank you, Speaker Elect Jeffrees.
- 2:51The Trump Administration House
- 2:53Republicans last year made a conscious
- 2:56decision to let the AI industry run wild
- 2:58and that was a bad decision politically
- 3:01and a bad decision on policy.
- 3:04Two things will change next term.
- 3:06We will hold public hearings on AI.
- 3:12AI.
- 3:13>> This is one of the new senators, Ted
- 3:15Lieu, and as you can see, what he is
- 3:17saying is that things are going to
- 3:18change.
- 3:20Artificial intelligence is going to be
- 3:21regulated, many more guardrails are
- 3:23going to be put in place, and CEOs are
- 3:24going to be forced to testify publicly.
- 3:26But he is not the only one. In general,
- 3:29all Democratic candidates have been
- 3:31precisely capitalizing on that sense of
- 3:33fear that exists in the United States
- 3:36regarding artificial intelligence.
- 3:45This is the new senator from Michigan,
- 3:47Abdul El-Sayed, who is also basically
- 3:48the first Muslim in the history of the
- 3:50United States, but he is one of the most
- 3:52negative about artificial intelligence
- 3:54and also one of the most confrontational
- 3:56toward the Trump administration.
- 3:58And in this case, he says to completely
- 4:01stop artificial intelligence.
- 4:04All this fear of AI and capitalizing on
- 4:06AI politically is not new in the United
- 4:08States.
- 4:09In fact, the great advocate,
- 4:18a temporary halt to the construction of
- 4:20data centers in the United States on a
- 4:22national level, that is to say, that no
- 4:24data centers be built until they clarify
- 4:26what they want to do, right?
- 4:29Well, this now seems to have much more
- 4:31strength: keep in mind that all these
- 4:33new, uh, politicians will take office in
- 4:35January.
- 4:36Therefore, we won't see any changes
- 4:38until January, but starting in January,
- 4:39we could see quite a few revolutions.
- 4:42And that is the reason why the stock
- 4:44market has fallen so radically in the
- 4:45United States, primarily driven by the
- 4:47drops in artificial intelligence
- 4:49companies.
- 4:51But what Bernie Sanders has always asked
- 4:53for, as I say, is to stop the data
- 4:55centers because, hey, there are many
- 4:56theories about the water issue, about
- 4:58this and that, and none of them make
- 4:59much sense.
- 5:01I’m telling you honestly, you know from
- 5:03here that I am one of the people who
- 5:05most advocates for AI to be regulated
- 5:07and for things to be done right and not
- 5:08fast, but even so, this whole data
- 5:10center issue makes absolutely no damn
- 5:12sense.
- 5:13Look, back home, this week—this weekend
- 5:15, in fact—in Zaragoza, there was a
- 5:17protest against the data centers being
- 5:19built in Aragon under the slogan "they
- 5:21don't create jobs, they don't pay taxes
- 5:23, and they take our resources."
- 5:26Honestly, this is starting to really get
- 5:28on my nerves. I was on television the
- 5:29other day, and it was more of the same.
- 5:31Everyone is obsessed with how much water
- 5:33data centers consume, and so on. But
- 5:36it’s just not true; we really need to
- 5:38analyze this statement a little better.
- 5:41To give you an idea, in Aragon, this
- 5:42meant 5,000 people blocking the city,
- 5:44trying to rebel and protest against
- 5:46these data centers, which I think is
- 5:48very legitimate and valid; but I would
- 5:50love for them to have the correct
- 5:52information so they can decide what they
- 5:54are protesting about, because, honestly
- 5:56, the level of misinformation on these
- 5:58topics is absurd.
- 6:01Let's analyze this claim from the
- 6:03protest a bit and break down each of
- 6:06their arguments step by step, shall we?
- 6:09In this case, the first one is "they
- 6:11don't create jobs."
- 6:12As for the claim that they don't create
- 6:14jobs, I honestly don't know where the
- 6:15hell that came from.
- 6:17First of all, about 80 small data
- 6:18centers are being planned for
- 6:19construction in Aragon; they aren't a
- 6:20huge deal.
- 6:22All of them combined don't even add up
- 6:24to one of Elon Musk's data centers, but
- 6:25in any case, they are 80 data centers.
- 6:28It is estimated that there are
- 6:30approximately 10,000 jobs over 36 months
- 6:32during the construction of a data center
- 6:35. 80 times 10,000 isn't very hard to
- 6:37calculate, right?
- 6:38We are talking about some 120,000
- 6:40high-quality, high-paying jobs for the
- 6:43construction of these data centers over
- 6:45the next 3 years.
- 6:47That is no small thing. Now, it is true
- 6:49that data centers don't need many
- 6:51workers to operate. These data centers
- 6:53will run with around 100 or 150 people,
- 6:55and with that, they can keep running for
- 6:57a lifetime.
- 6:59So we end up with relatively few—though
- 7:01they aren't actually that few—maybe
- 7:03we're talking about 10,000 to 12,000
- 7:05more long-term, extremely high-quality
- 7:07jobs, okay?
- 7:08We aren't talking about crappy jobs, we
- 7:10are talking about quality jobs, so
- 7:11overall employment.
- 7:13Whether it's a lot or a little is
- 7:15certainly up for debate, but in these
- 7:17times, bringing 120,000 jobs to Aragon
- 7:19is not bad at all, considering the pay
- 7:22those jobs offer.
- 7:24All these figures I'm giving you, I'm
- 7:26not making them up; you have them in
- 7:27this article here which talks about how
- 7:29much work a data center generates, and
- 7:30this is from the middle of this year, in
- 7:32August, okay?
- 7:33And it clearly explains, based on
- 7:35capacity and what it generates, and
- 7:36taking into account what is being built
- 7:38in Aragon, this would be roughly what I
- 7:40told you is going to be generated.
- 7:42Second, the issue that it uses up all
- 7:43the water, okay? That it consumes all
- 7:45the resources in the area. It's just not
- 7:48true, it's absolutely absurd. Obviously
- 7:51, a data center consumes water, but it
- 7:53doesn't consume that much water when we
- 7:55put it into perspective with the rest of
- 7:58the things humanity does. It's very easy
- 8:00to say, "No, John consumes a lot of
- 8:01water."
- 8:03Well, damn, it depends on how much I
- 8:04consume proportionally to other things
- 8:06we do.
- 8:06Well, with the data center, it's exactly
- 8:08the same. And once again, don't believe
- 8:10it just because I say it, believe the
- 8:12official data. In this case, I've looked
- 8:15up a website for the Florida water
- 8:17management district, okay? Florida water
- 8:20operations control, okay?
- 8:22An official organization, and it has a
- 8:23super extensive article that I highly
- 8:25recommend you take a look at because the
- 8:27issue of water consumption in data
- 8:29centers—or at least people's perception
- 8:31of it—is truly wild.
- 8:33It has a super extensive article on how
- 8:35they use it, why they use it, and what
- 8:37it implies that they use it, okay?
- 8:39Without any kind of bias, telling you
- 8:41directly the myth versus reality, that
- 8:44is, what a data center actually does.
- 8:48And the real data is that a data center
- 8:50of the style of those being built in
- 8:52Aragon can consume about as much as
- 8:541,000 homes.
- 8:56And you'll say, "Damn, John, 1,000 homes
- 8:58showering, washing dishes, and so on.
- 9:01That's a lot of water." Well, hey,
- 9:02that's the water it is, there's nothing
- 9:04more to it. Do you know how many homes
- 9:05there are in Aragon? 840,000.
- 9:08Okay? So, using the same as 1,000 homes
- 9:11isn't going to be too much, considering
- 9:13there are 840,000 homes.
- 9:15Do you think it would have any impact on
- 9:17the water levels of the Ebro or Aragon
- 9:19if there were suddenly 1,000 more homes
- 9:21, in addition to the 840,000 that
- 9:22already exist? Well, probably not. It
- 9:25would have practically no impact.
- 9:28Well, that is the water impact that data
- 9:30centers will have, because data centers
- 9:32have been self-contained for a long time
- 9:34; they recycle the water, so they fill
- 9:36up with water and, from then on, that
- 9:38water, well obviously, there is a loss,
- 9:40but it's not notable, it's not
- 9:41substantial, it's not a big deal.
- 9:44But let me put it to you another way. A
- 9:46data center consumes more or less what
- 9:49is about 90 hectares of corn, that is,
- 9:52to create uh corn we have to plant
- 9:55hectares of corn, okay? And that corn we
- 9:58consume uses water to be able to give us
- 10:00corn. It’s as plain as day.
- 10:03Well, more or less the equivalent of 90
- 10:05hectares of corn is precisely what a
- 10:07data center consumes.
- 10:09And you might say, "Hey, is that a lot
- 10:10or a little?" John, well look at the
- 10:12data.
- 10:13This year, only 100,000 hectares of corn
- 10:15have been planted in Aragon.
- 10:17And if we add that to the rest of the
- 10:19crops, it turns out that a data center
- 10:21will account for 0.03%
- 10:25of the total water we use for crops.
- 10:280.03%
- 10:31of the water we use for crops. From
- 10:33there, does a data center consume water
- 10:35? Yes, it's a lot, it's significant. No
- 10:38, it's as simple as that.
- 10:40Okay, I don't know how else to tell you
- 10:42, but I'm going to give you one more
- 10:43piece of information, and that is this
- 10:44news item, which I think is very clear,
- 10:46where this guy here, Satya Nadella, who
- 10:47is the CEO of Microsoft, the most
- 10:49important software company in the world
- 10:50, said it in a conference.
- 10:53A data center consumes the same amount
- 10:55of water as a restaurant.
- 10:58I haven't seen them banning the
- 10:59construction of restaurants in Aragon,
- 11:01in Zaragoza, or anywhere else because of
- 11:03the water impact these restaurants have
- 11:05.
- 11:06Well, the same goes for data centers.
- 11:09We have the opportunity of a lifetime to
- 11:11put data centers next to our cities that
- 11:13will generate quality jobs and will give
- 11:15us the computing power we are going to
- 11:17need so much for the new economy.
- 11:19And instead of that, we are throwing a
- 11:21wrench in the works. You can be for or
- 11:23you can be against artificial
- 11:25intelligence.
- 11:27It makes absolutely no difference; we
- 11:29need data centers or our country is
- 11:31going to hell, and from there, if we
- 11:32don't build computing power, we will go
- 11:34from being providers to being customers
- 11:36.
- 11:37You tell me what is better in the
- 11:39economy: being the one who receives the
- 11:41money or the one who puts it in. Well,
- 11:43that is precisely what this is about.
- 11:45And right now, we are at a point where,
- 11:47thanks to this situation in the United
- 11:49States, we could benefit from all the
- 11:51data centers that Google, Microsoft, and
- 11:53OpenAI will have to build, so they come
- 11:55here instead of going elsewhere, like
- 11:57India, for example.
- 12:00There was a third thing they wanted to
- 12:01bring to the table in this demonstration
- 12:03, besides the issue of resources and the
- 12:05issue of jobs. And the third one was
- 12:07that they don't pay taxes. And you say,
- 12:09"Hey, well that will depend on who the
- 12:10hell is in government, because here it
- 12:12will depend on each one." But in the end
- 12:13, you have to see it as a situation of
- 12:15competitiveness. If we want them to come
- 12:17here to set up a data center, we have to
- 12:19make it easy and competitive with other
- 12:21places. And do you know what is
- 12:23happening in other places? Well, for
- 12:24example, in India, they don't make them
- 12:26pay taxes for 20 years. If you put a
- 12:29data center in India, you save on taxes
- 12:31until 2047.
- 12:34That is the level of how clearly India
- 12:36has seen it so that data centers are
- 12:37used, that data centers are built in
- 12:39their country to generate jobs precisely
- 12:41and to give them that edge in computing
- 12:43that allows them to dominate the
- 12:44artificial intelligence market a little
- 12:46more.
- 12:49And that, I repeat, even if AI
- 12:50development stops, as the American
- 12:52Democrats are asking, even if we stop
- 12:53developing artificial intelligence, we
- 12:55are going to need a lot of computing
- 12:57power that we don't have today.
- 13:00Remember that Europe has 5%of the
- 13:01world's computing power, while the
- 13:03United States has 75%.
- 13:06We are in an extremely unfavorable
- 13:08situation. The United States is building
- 13:10data centers like crazy to see what
- 13:12happens after January, but here in
- 13:13Europe, we are not doing it and we
- 13:15should be doing it for the prosperity it
- 13:17brings us, especially in Spain.
- 13:19I am going to leave you this article
- 13:21about India because I think you need to
- 13:22understand what the competitive market
- 13:24for building data centers is and
- 13:25understand that the tax issue is
- 13:26possibly the least of it.
- 13:29By the way, look at how complicated this
- 13:31is getting in the United States, that in
- 13:33a city in Indiana, a small town of
- 13:35barely 80,000 inhabitants, they have
- 13:36already banned super artificial
- 13:38intelligence.
- 13:40It's about Bloomington, and they’ve
- 13:43basically created a new local regulation
- 13:45where they are banning superintelligence
- 13:48.
- 13:49Be careful here, because Donald Trump
- 13:50has really messed things up with the
- 13:52term "superintelligence," and it’s been
- 13:53very confusing during this election. We
- 13:55don’t even know what they’re referring
- 13:57to anymore, because Trump says that what
- 13:58we’ve always called AI is now called
- 14:00superintelligence, even though the
- 14:01concept of superintelligence meant
- 14:03something else. It was usually when we
- 14:04reached that AGI, or something even
- 14:06further beyond, that was superior to all
- 14:07humans.
- 14:08In this case—and don’t let yourselves be
- 14:10confused because this has been muddled a
- 14:12lot—what they have banned is precisely
- 14:14artificial superintelligence, okay?
- 14:16And that means intelligence that is
- 14:18superior to all humans and capable of
- 14:19doing more than humans themselves.
- 14:22So what they are doing is preparing for
- 14:24the future; they aren't banning the AI
- 14:27of today, like ChatGPT, but they are
- 14:29banning any use of that artificial
- 14:31superintelligence in the Indiana city we
- 14:33mentioned.
- 14:35Which means that if we reach a point
- 14:37where it's declared that these tools are
- 14:39"superintelligent," as Donald Trump is
- 14:41doing, we could find that citizens in
- 14:42that city won't be able to use ChatGPT
- 14:44or Gemini, and that is a pretty serious
- 14:46problem for competitiveness.
- 14:49But it’s that level of hatred toward
- 14:51artificial intelligence we're seeing in
- 14:53the United States that is leading to
- 14:55outright bans instead of just regulation
- 14:57, which would be reasonable, but banning
- 14:59it makes absolutely no sense.
- 15:02I’ll leave the article for you to take a
- 15:03look, but really, be careful, because
- 15:05they are publishing a lot of things
- 15:06about this town that don't make sense,
- 15:08and what they’ve done is explained here
- 15:09by them.
- 15:11Ultimately, we have to see what ends up
- 15:12happening with the election, because you
- 15:14know that many promises change once
- 15:15people are in office, but we'll start to
- 15:17see it come January.
- 15:19There is certainly going to be a major
- 15:21shift in the United States because, even
- 15:23if Donald Trump wants to, he won't be
- 15:25able to pass everything he wants or take
- 15:27every action he desires.
- 15:28He cannot, for example, subsidize data
- 15:30centers right now without going through
- 15:33Congress for approval, so they are going
- 15:35to deny it, and from there, AI companies
- 15:37are going to have a problem. The second
- 15:39problem they have in the United States
- 15:41is that all the CEOs have fallen to
- 15:43their knees before Donald Trump and done
- 15:45whatever he wanted, thinking they were
- 15:47in control. And now, suddenly, there's a
- 15:49shift in the political trend, and with
- 15:50people blaming Donald Trump a bit, they
- 15:52are all suffering for having supported
- 15:54him. Probably the one who will benefit
- 15:56the most here is Anthropic, which has
- 15:57been the most resilient against Trump's
- 15:59policies.
- 16:00But both Zuckerberg and Elon Musk are
- 16:02suffering a great deal as a result of
- 16:04all this.
- 16:05In fact, Elon Musk is the one who has
- 16:07lost the most because he spent 500
- 16:09million trying to help Republican
- 16:11candidates win these midterms in each of
- 16:13the areas that interested him, for
- 16:14example, in Texas, and he lost, so he
- 16:16lost all that money.
- 16:18All the candidates he supported have
- 16:19lost. So, imagine how well things turned
- 16:22out for Mr. Elon Musk.
- 16:23In any case, we will have to see, but
- 16:26please, I ask you to be critical and
- 16:29reasonable. We may like
- 16:30superintelligence, we may be afraid of
- 16:32the fact that AI could get out of hand,
- 16:35whatever you want. Computing has nothing
- 16:37to do with that. We need computing, no
- 16:39matter what. We are very short on it.
- 16:41We must become providers as soon as
- 16:43possible to reduce our bills as
- 16:45customers, and from there, we need to
- 16:47invest in data centers.
- 16:50And I think there is no doubt that the
- 16:52best thing that can happen to your
- 16:53community right now is having a data
- 16:54center built next to your house, which
- 16:55will guarantee jobs for your children.
- 16:58So, in any case, let's get to it, let's
- 17:00pay attention, and let's see what else
- 17:02has happened this week.
- 17:05What has happened, and will surely
- 17:07impress you, is the offer Hostinger has
- 17:08for Black Friday, the sponsor of this
- 17:10video, which brings you the option to
- 17:12host your websites on a hosting service
- 17:14that is, as the name implies, incredible
- 17:16, for an absolutely unbeatable price.
- 17:20It's on offer with an 87%discount, and
- 17:22you can get it starting at 1.79 a month
- 17:24.
- 17:25I think it's the lowest price I've ever
- 17:26seen from Hostinger. The version that
- 17:28includes everything related to AI for
- 17:29business, etc., which is the best one, I
- 17:31tell you, ends up being 2.49.
- 17:33Really very cheap. From here, all you
- 17:35have to do is choose how many months you
- 17:37want it for. And to take advantage of
- 17:39the offer, I recommend that you choose
- 17:40the maximum number of months, which
- 17:42would be 48 months, since, after all,
- 17:43you will be using your hosting anyway.
- 17:45Uh, if you select 48 months, that's the
- 17:47price you get.
- 17:48If you want less, you can set it from 12
- 17:50months and the price goes up a little,
- 17:51but it's still much cheaper than what
- 17:53you have today.
- 17:54You can migrate your websites here and
- 17:56save a bundle. Also, to help you save a
- 17:58little bit more, I'm going to give you a
- 18:00discount coupon. Enter "inteligencia"
- 18:02here, which I love that it even works on
- 18:04Black Friday, and it will reduce the
- 18:05price even further.
- 18:06And for just € 130, you would get 48
- 18:09months of hosting, a domain, and
- 18:11everything you want to put in there for
- 18:14your websites.
- 18:16If you decide to build your website with
- 18:17Hostinger's tools, you can create it in
- 18:18minutes, but you can also do it with a
- 18:20WordPress you already have, so it’s very
- 18:21easy to migrate everything.
- 18:23If you use the builder, you'll see that
- 18:25you can edit it however you like later,
- 18:27without needing to know anything about
- 18:29programming, even seeing how it looks on
- 18:31mobile and changing the design for those
- 18:33devices, which is super useful and
- 18:34Hostinger handles the setup.
- 18:36You have a ton of AI tools to generate
- 18:39images and posts.
- 18:40And it can even create what would be
- 18:42your e-commerce products.
- 18:44So you don't just have to use it for
- 18:46static websites, but also for websites
- 18:48with e-commerce.
- 18:50So if you have a website or want to
- 18:52create one, Hostinger is your best
- 18:53option at the best price; take advantage
- 18:55of Black Friday.
- 18:57You have the link in the description and
- 18:58remember to enter the code "
- 19:00inteligencia" because the offer is truly
- 19:01absolutely incredible.
- 19:03I think it's the best one we've seen in
- 19:04all the time Hostinger has been
- 19:06sponsoring this channel.
- 19:08Alright, let's continue with the news
- 19:10because the AI issue isn't just a
- 19:12problem for data centers, but we are
- 19:13seeing many cybersecurity problems.
- 19:16I also have to say, and I want to speak
- 19:18up for this, that not everything you are
- 19:20seeing regarding cybersecurity has to do
- 19:22with artificial intelligence.
- 19:24We've seen all these cases of
- 19:25refrigerators, both in supermarkets and
- 19:27blood banks, but there is nothing to
- 19:29indicate that it was done with
- 19:30artificial intelligence.
- 19:32I am convinced that hackers use AI just
- 19:35as everyone else does, but I don't think
- 19:37that was actually made possible thanks
- 19:39to AI. Okay, I'm telling you this
- 19:41because it's important that we start to
- 19:42differentiate. Also, the other day, a
- 19:44week ago, Revolut went down and it had
- 19:45nothing to do with AI. It was some
- 19:47Google servers that went to hell, okay?
- 19:49So, at the end of the day, let's not mix
- 19:51things up, but we are in a very
- 19:52complicated situation.
- 19:55We are seeing more and more cases of all
- 19:57kinds of cybersecurity issues, and
- 19:58obviously, AI has had something to do
- 20:00with it, directly or indirectly, but the
- 20:02worst is yet to come.
- 20:05This week we heard from the head of the
- 20:07Spanish cyberspace command, which is
- 20:08basically a military role that oversees
- 20:10it—just like we have someone for the sea
- 20:11, someone for the sky, and someone for
- 20:13land, well, we have one for cyberspace,
- 20:15okay? And he made sure to lay out on the
- 20:17table that things are getting quite
- 20:19complicated.
- 20:21The person you see here is Javier Roca,
- 20:23who is literally the commander of the
- 20:25Spanish Joint Cyberspace Command, and he
- 20:27basically said that we must prepare for
- 20:29the worst-case scenario.
- 20:32In fact, he compared the impact of
- 20:34cybersecurity due to artificial
- 20:35intelligence—that is, the impact AI will
- 20:38have in this space—to something more
- 20:40dangerous than when artillery or
- 20:42airplanes were invented.
- 20:45In other words, that is the level of
- 20:47increasing problems that is coming our
- 20:49way.
- 20:50Just imagine, back in whatever year it
- 20:52was, when we were militarily prepared to
- 20:54fight on land and suddenly planes
- 20:56started arriving; it completely changed
- 20:57the security of nations because suddenly
- 20:59someone could fly over and drop a bomb
- 21:01on you.
- 21:03Well, he says that is less impactful
- 21:04than the effect artificial intelligence
- 21:06will currently have on the security of
- 21:08nations.
- 21:09So, just imagine how massive this really
- 21:11is. We have to take this seriously.
- 21:13Honestly, these things make me think
- 21:15back to that time in February, I don't
- 21:17know if you remember, when Anthropic
- 21:19announced its models and said, "Guys,
- 21:21things are going to get rough regarding
- 21:22cybersecurity because these models are
- 21:24starting to be very powerful." And
- 21:26people started saying, "Marketing, they
- 21:28are just trying to pump up their
- 21:29company's valuation."
- 21:31Well, look at that marketing—now the
- 21:33head of cybersecurity command is telling
- 21:35us that some very rough times are ahead
- 21:37.
- 21:38And what will those rough times be like
- 21:39?
- 21:40Well, he didn't say, but I can tell you
- 21:42what Anthropic's director of
- 21:43cybersecurity said, which I mentioned a
- 21:45few days ago, and basically, they said
- 21:47we have about 12 to 18 months of
- 21:48struggling ahead, and after that, things
- 21:50will improve.
- 21:52Their reasoning is that, for the moment
- 21:54, frontier models—the most powerful
- 21:56models—are incredibly capable in
- 21:58cybersecurity. And what happens? That
- 22:00the average person doesn't use those
- 22:02models.
- 22:03So, until smaller models are capable
- 22:05enough to defend against large models—
- 22:08which happens over time as they are
- 22:10distilled from the larger ones and learn
- 22:12to do things—we will remain very
- 22:14vulnerable.
- 22:15That means someone with bad intentions
- 22:17and a good AI can spend 20 or 30,000
- 22:19bucks and hack something, okay?
- 22:22Until we have our defenses raised, we’re
- 22:24going to have a hard time, and they
- 22:25estimate it will take those 12 to 18
- 22:27months, a period where we’ll struggle
- 22:29significantly.
- 22:31We’ll see how bad it gets and where it
- 22:33ends up breaking, but I have a theory
- 22:34that in the next 6 to 12 months
- 22:36something serious will happen in terms
- 22:38of cybersecurity that will make
- 22:40everyone’s hair stand on end, and I’m
- 22:42talking about something likely affecting
- 22:44banks or power grids.
- 22:46I believe there will be something
- 22:47stemming from artificial intelligence
- 22:49that will trigger a major political
- 22:50reaction.
- 22:52And also a great social awakening, which
- 22:54is about time, because we’ve been warned
- 22:56about this for months, at least since
- 22:57February, but I certainly think even
- 22:59longer because they’ve been pointing at
- 23:01the Moon for a long time and we keep
- 23:03looking at the finger.
- 23:06And speaking of cybersecurity, ChatGPT
- 23:08has launched a new model, in this case
- 23:10GPT6 Cyber, one of the new models
- 23:12specially trained to defend against
- 23:14cyberattacks.
- 23:16And what is this?
- 23:17Well, basically, they are those models,
- 23:19like what were once myths, that are only
- 23:21available to trusted companies—
- 23:23companies that register in programs,
- 23:25which in OpenAI's case is called
- 23:27Daybreak—and they basically give them
- 23:29these defenses before these models are
- 23:31released.
- 23:32They are models that are basically
- 23:34uncapped.
- 23:35They are models that, if we compare them
- 23:37to the models available to the public,
- 23:38are much better than the previous ones.
- 23:41If you look here, for example, at GPT6
- 23:43Astra, which is the model we all have
- 23:44publicly, as you can see, it scores 0%
- 23:46on this test.
- 23:47Conversely, these models like GPT6 Cyber
- 23:50allow for a score of 48.5%on that test.
- 23:54The difference is that this one is
- 23:55capped, which is the one we all have,
- 23:57and this one is uncapped, which is the
- 23:59one only those trusted companies have,
- 24:00allowing them to shield their systems
- 24:02against cybersecurity attacks.
- 24:05The main problem is that not everyone
- 24:07only has access to those restricted
- 24:09models, as Anthropic has already said:
- 24:11models like Z and GLM 5.3 have the
- 24:13capabilities that Mythos had in February
- 24:15, and therefore, they are very capable
- 24:17models that are advancing very quickly
- 24:19and are open source, meaning those
- 24:21guardrail limitations do not exist in
- 24:23them.
- 24:25You download the model and do whatever
- 24:27you want with it, so from that point on
- 24:29, those cybersecurity capabilities are
- 24:31not being sufficiently blocked because
- 24:33people can access them with open source
- 24:35models.
- 24:37Well, we continue with more news, and
- 24:38this is one of those that literally seem
- 24:40like Black Mirror, but they are not.
- 24:43And it is that in Delaware, the cradle
- 24:45of putting companies to save on taxes,
- 24:46they have created the option or are
- 24:48about to approve the option for an
- 24:49artificial intelligence to be able to
- 24:51create a company.
- 24:54The motivation for Delaware is very
- 24:55simple: they want to attract capital,
- 24:57and they are going to do it as they have
- 24:59done until now, by creating options so
- 25:00that you can create your company in that
- 25:02state more cheaply and by paying fewer
- 25:04taxes.
- 25:05Here, basically also regarding the part
- 25:07of liability, if they allow an AI to
- 25:09create the company, there is no human
- 25:11executive who has to basically, well,
- 25:14deal with the consequences of that
- 25:15company.
- 25:16The liability is shifted directly to the
- 25:18owners of the company, which would be
- 25:20the artificial intelligence.
- 25:22This, the idea is that your agent can
- 25:24create a company that you can exploit,
- 25:26obviously, because you exploit people.
- 25:28We will see for how long and we will see
- 25:29if that is a good idea, which certainly
- 25:31seems to me like one of the worst.
- 25:32I mean, it is one of those things where
- 25:34, when you consider what things we
- 25:35should not do with AI just in case and
- 25:37you say: "Hey, maybe we shouldn't
- 25:38connect it to the internet, we shouldn't
- 25:40put it in the hands of children," we
- 25:42have already done all of that.
- 25:44"Maybe we shouldn't make it so it can
- 25:45have its own company, make decisions of
- 25:47that level, and be able to execute
- 25:48things," well look, they are working on
- 25:49it, okay?
- 25:50So, it is one of those quite stupid
- 25:52decisions, but one that is going to
- 25:53bring them a lot of capital, and that is
- 25:55what Delaware is looking for in this
- 25:56case.
- 25:58For the moment it is not approved, for
- 25:59the moment it is a proposal for a 30-
- 26:01month pilot program, so there is quite a
- 26:02bit of time left until your artificial
- 26:04intelligence agent can create a company
- 26:05.
- 26:07And keep an eye on this response from
- 26:09Elon Musk to a tweet, which is more
- 26:11important than it seems, as he says the
- 26:13cost of serving Grokbot was becoming
- 26:14impossible to sustain relative to the
- 26:16revenue it was generating.
- 26:18Okay? Why do you see this response?
- 26:21Well, because Elon Musk has cut the
- 26:22usage limits within Grokbot plans, for
- 26:24those personal assistants, which are
- 26:26actually the only ones currently
- 26:28available in Spain today.
- 26:30Actually, someone is already telling me
- 26:32they’ve started to see uh OpenAI GPTs
- 26:34appear, so check your accounts. If you
- 26:35have a pro account in Spain, it seems
- 26:37like GPTs are starting to reach you,
- 26:38okay?
- 26:40In any case, what he is telling us here
- 26:42is that the amount of usage allowed
- 26:44within the Grokbot account is being
- 26:46radically reduced, and that obviously
- 26:48affects things directly because these
- 26:50agents consume a huge amount.
- 26:54This is also an important issue on the
- 26:55OpenAI side because, in theory, starting
- 26:57November 18th, they were supposed to
- 26:58start charging us for GPTs, and we'll
- 26:59see what business model they end up
- 27:01implementing, since they have been free
- 27:02until now.
- 27:04It’s a real shame they haven't arrived
- 27:05in Spain yet, so we can't take advantage
- 27:07of that, although as you know I have
- 27:09early access—I have it right here on the
- 27:10watch—so I’ve been using it and I think
- 27:12it’s absolutely incredible. In fact, I
- 27:14think it’s the best thing ChatGPT has
- 27:16released since they launched ChatGPT
- 27:17itself.
- 27:18In any case, for now, we are seeing that
- 27:20these agents are going to get expensive
- 27:22, I have no doubt, they are very useful
- 27:24and we will have to see if it’s worth it
- 27:26.
- 27:27In the case of GPTs, it's rumored—
- 27:29because it appeared on a page that
- 27:30shouldn't have been public, but was
- 27:32there for a few hours, etc.—that the
- 27:34enterprise cost, we’ll see what the B2C
- 27:36cost is, but for companies it was
- 27:38rumored that each GPT will cost $ 50 a
- 27:40month per employee.
- 27:42Meaning, from here on we’ll see if
- 27:44that’s the case or not, but it is a
- 27:46fairly substantial price increase
- 27:47compared to ChatGPT licenses because
- 27:49it’s something additional.
- 27:51So, if companies weren't already
- 27:53thinking that we were starting to move
- 27:55from $ 20 accounts to $ 100 ones, now we
- 27:57add another $ 50 if we want the employee
- 27:59to have a GPT. So we'll see if this is
- 28:01really the case or if it was a mistake,
- 28:03but it certainly seems quite, quite
- 28:05expensive.
- 28:06Well, I’ll leave you with this news
- 28:08that’s going to cause a stir: it seems
- 28:09Google is going to search in
- 28:11schools that don't use artificial
- 28:12intelligence.
- 28:15All of this has an explanation, but it's
- 28:16a big deal because it's something that
- 28:18is hitting Google with backlash from all
- 28:20sides.
- 28:21Okay? Here’s the story.
- 28:23A series of schools in the United States
- 28:25decided they wanted to avoid the use of
- 28:26artificial intelligence by their
- 28:27students, for example, in tools like
- 28:29Google Classroom, which my kids use at
- 28:30school.
- 28:31Okay? Well, suddenly they said, "Hey,
- 28:33we're going to remove AI from Google
- 28:35Classroom and all of Google's smart apps
- 28:37because we truly believe kids should
- 28:39study without AI," and we can more or
- 28:41less agree on that. I completely
- 28:43disagree, but in any case, that was the
- 28:45approach taken by the schools. And what
- 28:47was Google's response? Well, fine, no
- 28:49problem. You can deactivate the AI, you
- 28:51have the option, but if you deactivate
- 28:53the AI, you also deactivate search, so
- 28:54you're left without Google entirely. I
- 28:56mean, basically, they've told them the
- 28:58product is so unified, so fused together
- 29:00, that they can't separate it.
- 29:02And so, if they want to remove the AI,
- 29:04they are also going to remove all Google
- 29:05search functions.
- 29:07They will be able to retain some of the
- 29:08features within Classroom, etc., but
- 29:10they are basically wiping out something
- 29:12very important, which is the ability for
- 29:14students to use Google Search.
- 29:16Obviously, there are alternatives, and
- 29:19this is causing a tremendous backlash
- 29:21for Google's image for forcibly using
- 29:23search as leverage to ensure artificial
- 29:25intelligence is used in schools where we
- 29:28might not know what the effect will be
- 29:30on children.
- 29:32We have spoken at length about the topic
- 29:33of education. You have an incredible
- 29:34podcast with Sonia, who is one of the
- 29:36people who knows the most about this,
- 29:37and you’ve had it on my channel for a
- 29:39while now. Which, by the way, went quite
- 29:40unnoticed, that podcast, and I think
- 29:42it's very interesting, so I recommend
- 29:43you watch it if you haven't seen it yet
- 29:44.
- 29:45But I also think that, without a doubt,
- 29:47we have many doubts, but in the field of
- 29:49education, we have to do things; I mean
- 29:50, it doesn't matter if we make mistakes
- 29:52.
- 29:53If we realize in a couple of years that
- 29:54it wasn't the right path; the kids
- 29:56aren't going to be ruined, okay? So, I
- 29:57think it’s important and that the
- 29:59biggest risk we have right now in
- 30:00education is doing nothing. So I don't
- 30:02support Google's move; I think people
- 30:04should have the freedom to choose, but
- 30:05in this case, I think they are making a
- 30:07mistake trying to remove AI from schools
- 30:08.
- 30:10And watch out, because there is a lot of
- 30:12news like this, like Apple cutting
- 30:14worker spending, and it's being
- 30:16perceived alongside the political noise
- 30:19of the Democratic victory, Trump's loss
- 30:21in the Senate and Congress, the fact
- 30:23that Elon Musk has lost a lot of money,
- 30:26and the issue that right now at Nvidia,
- 30:28they're in a very complex moment because
- 30:30Jensen Huang has been caught cheating on
- 30:33his wife with a Chinese spy, and all
- 30:35these things.
- 30:37It is being seen as the bursting of the
- 30:39artificial intelligence bubble, okay?
- 30:41And I think that is absolutely wrong. I
- 30:44think the view is really very
- 30:45short-sighted.
- 30:47A lot of things have happened all at
- 30:49once in one week, but that doesn't mean
- 30:50a bubble has burst, especially not the
- 30:52underlying technology, which is
- 30:53artificial intelligence.
- 30:55We are going to see how this technology
- 30:57keeps growing and the stocks of these
- 30:59companies keep rising.
- 31:01Maybe not at the same pace as before,
- 31:03but I think we are far from having built
- 31:05too many data centers, so I think we are
- 31:08very far from a bubble bursting.
- 31:10Mind you, the stock market drops have
- 31:12been very significant, but there have
- 31:13been others in the past. Remember when
- 31:15Dipsi came out, the US stock market took
- 31:17an epic hit, and we recovered and have
- 31:19been at all-time highs a million times.
- 31:21I am sure the stock market will continue
- 31:23to move forward and things will keep
- 31:24going up.
- 31:26Because look at this same thing with
- 31:27Apple, the news says Apple is cutting
- 31:29the supply for its workers, the use of
- 31:31AI by its workers, and you say, "Wow,
- 31:33Apple isn't going to use it anymore."
- 31:35No, that's not it. If you look at the
- 31:38news, until now, Apple was spending $
- 31:42300 a day per worker on AI. I mean, $
- 31:45300 a day. These people have tens of
- 31:48thousands of employees. That's a damn
- 31:49fortune. And what are they doing? They
- 31:51are lowering it to 200.
- 31:53I mean, it is by no means crazy. In fact
- 31:56, they say there are plans to eventually
- 31:57reduce it to 100.
- 31:59For now, with those 200, it means Apple
- 32:01spends 4,000 bucks a month per employee
- 32:03on artificial intelligence in tokens.
- 32:06It's insane. No one has ever spent this
- 32:09much money on any SaaS or anything that
- 32:11has provided a company with a tool to
- 32:13work better. $ 4,000 per employee is a
- 32:16tremendously booming market, and
- 32:17artificial intelligence companies are
- 32:19going to do very well.
- 32:21$ 4,000 is a fortune, so let's tone it
- 32:23down a bit, the talk about the bubble
- 32:26bursting.
- 32:27It’s a lot of clickbait. The reality is
- 32:29that, for now, the AI business remains a
- 32:32profitable venture. We will see how far
- 32:34it goes and where it takes us. I have
- 32:36always told you that I have no doubt it
- 32:38is a bubble; like any type of technology
- 32:41, this kind of technological explosion
- 32:43will generate a bubble, and AI is doing
- 32:45the same.
- 32:47Therefore, many people will lose their
- 32:48money, but that does not mean that the
- 32:50technology won't change all of our lives
- 32:52.
- 32:53Well, and finally, I want to talk to you
- 32:55about this company I found called Shinke
- 32:57, which is one of those examples of a
- 32:59company that can really do things with
- 33:01artificial intelligence that couldn't be
- 33:03done until now. This is crazy, because
- 33:05we are talking about a fishing company,
- 33:07or related to fishing, where basically
- 33:10the theorem here is that when you catch
- 33:12a fish and take it out of the water, the
- 33:14fish—at a massive level, when fishing
- 33:16boats catch them—the fish suffocates on
- 33:18the boat, okay? And, basically, since it
- 33:20doesn't have water, the fish suffocates
- 33:22and dies.
- 33:23When the fish suffocates in such a cruel
- 33:26way, honestly, the fish releases a
- 33:28series of things from its brain and body
- 33:30that make the meat's quality worse, okay
- 33:32?
- 33:34It might be a matter of chemical
- 33:35reactions, whatever you want to call it
- 33:37, but the point is that what happens is
- 33:39that the quality of the fish is reduced
- 33:40, and that is the fish we all eat all
- 33:42over the world. Okay? Well, this company
- 33:45, using artificial intelligence, has
- 33:47developed a machine where fishermen put
- 33:50these fish in, and it detects where the
- 33:52fish's brain is, gives it a puncture
- 33:55that instantly kills it before the fish
- 33:57can release all that and before the fish
- 34:00continues to suffer, which is a very
- 34:02decent way to dispatch the fish and, at
- 34:05the same time, improves the quality of
- 34:08the fish meat.
- 34:09So, this would basically be the typical
- 34:11thing hunters do when they catch an
- 34:13animal and kill it instantly; well, it's
- 34:14basically the same thing but with fish.
- 34:17In this case, the machine you'll be
- 34:19seeing in the videos is mind-blowing.
- 34:22You put the fish in on one side; with a
- 34:24series of camera sensors driven by
- 34:25artificial intelligence, it detects
- 34:27where it needs to strike based on the
- 34:29fish species, inserts the pin, and from
- 34:31there, it kills it instantly, making the
- 34:33fish meat much better for consumption
- 34:35than what we were eating before.
- 34:38One more proof, as I was telling you,
- 34:39that artificial intelligence is
- 34:41completely transversal. It can be used
- 34:43in practically any industry, even in
- 34:45things that seem absolutely crazy to us
- 34:47. Well, here you have it applied to one
- 34:50of the most manual things we do, like
- 34:52fishing. Well, in short, as you can see
- 34:54, artificial intelligence continues to
- 34:55be tremendously important. I'm going to
- 34:57stop the plot so you can have all this
- 34:59for your agents. You have the link in
- 35:00the description and I'll see you in the
- 35:02next video.
- 35:03If you have any questions, remember, you
- 35:05can leave a comment down below and we'll
- 35:07answer as soon as possible.
- 35:08Bye, family. M.
About this transcript
This page contains the full transcript of NOTICIAS IA: El terremoto político en USA que amenaza el futuro de la IA by Inteligencia Artificial, generated from the public captions YouTube serves with the video. The transcript has 6,930 words across 1,096 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.
What you can do with it
Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.
Free YouTube transcript tool
YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.