Türkiye: TRT World — News

20260920 09:00 UTC · 00:30:59 · 495 transcript segments · GDELT Visual Explorer · plain-text transcript · Event Map

TRT World brings the latest news from around the globe, when it happens.

Film strip

One frame every 4 seconds, 465 in all. Click a frame to jump the transcript to that moment, or click a line of the transcript to see what was on screen while it was said. Served from apprised.news, so it loads behind proxies that block Google Cloud Storage.

00:00:00
00:00:04
00:00:08
00:00:12
00:00:16
00:00:20
00:00:24
00:00:28
00:00:32
00:00:36
00:00:40
00:00:44
00:00:48
00:00:52
00:00:56
00:01:00
00:01:04
00:01:08
00:01:12
00:01:16
00:01:20
00:01:24
00:01:28
00:01:32
00:01:36
00:01:40
00:01:44
00:01:48
00:01:52
00:01:56
00:02:00
00:02:04
00:02:08
00:02:12
00:02:16
00:02:20
00:02:24
00:02:28
00:02:32
00:02:36
00:02:40
00:02:44
00:02:48
00:02:52
00:02:56
00:03:00
00:03:04
00:03:08
00:03:12
00:03:16
00:03:20
00:03:24
00:03:28
00:03:32
00:03:36
00:03:40
00:03:44
00:03:48
00:03:52
00:03:56
00:04:00
00:04:04
00:04:08
00:04:12
00:04:16
00:04:20
00:04:24
00:04:28
00:04:32
00:04:36
00:04:40
00:04:44
00:04:48
00:04:52
00:04:56
00:05:00
00:05:04
00:05:08
00:05:12
00:05:16
00:05:20
00:05:24
00:05:28
00:05:32
00:05:36
00:05:40
00:05:44
00:05:48
00:05:52
00:05:56
00:06:00
00:06:04
00:06:08
00:06:12
00:06:16
00:06:20
00:06:24
00:06:28
00:06:32
00:06:36
00:06:40
00:06:44
00:06:48
00:06:52
00:06:56
00:07:00
00:07:04
00:07:08
00:07:12
00:07:16
00:07:20
00:07:24
00:07:28
00:07:32
00:07:36
00:07:40
00:07:44
00:07:48
00:07:52
00:07:56
00:08:00
00:08:04
00:08:08
00:08:12
00:08:16
00:08:20
00:08:24
00:08:28
00:08:32
00:08:36
00:08:40
00:08:44
00:08:48
00:08:52
00:08:56
00:09:00
00:09:04
00:09:08
00:09:12
00:09:16
00:09:20
00:09:24
00:09:28
00:09:32
00:09:36
00:09:40
00:09:44
00:09:48
00:09:52
00:09:56
00:10:00
00:10:04
00:10:08
00:10:12
00:10:16
00:10:20
00:10:24
00:10:28
00:10:32
00:10:36
00:10:40
00:10:44
00:10:48
00:10:52
00:10:56
00:11:00
00:11:04
00:11:08
00:11:12
00:11:16
00:11:20
00:11:24
00:11:28
00:11:32
00:11:36
00:11:40
00:11:44
00:11:48
00:11:52
00:11:56
00:12:00
00:12:04
00:12:08
00:12:12
00:12:16
00:12:20
00:12:24
00:12:28
00:12:32
00:12:36
00:12:40
00:12:44
00:12:48
00:12:52
00:12:56
00:13:00
00:13:04
00:13:08
00:13:12
00:13:16
00:13:20
00:13:24
00:13:28
00:13:32
00:13:36
00:13:40
00:13:44
00:13:48
00:13:52
00:13:56
00:14:00
00:14:04
00:14:08
00:14:12
00:14:16
00:14:20
00:14:24
00:14:28
00:14:32
00:14:36
00:14:40
00:14:44
00:14:48
00:14:52
00:14:56
00:15:00
00:15:04
00:15:08
00:15:12
00:15:16
00:15:20
00:15:24
00:15:28
00:15:32
00:15:36
00:15:40
00:15:44
00:15:48
00:15:52
00:15:56
00:16:00
00:16:04
00:16:08
00:16:12
00:16:16
00:16:20
00:16:24
00:16:28
00:16:32
00:16:36
00:16:40
00:16:44
00:16:48
00:16:52
00:16:56
00:17:00
00:17:04
00:17:08
00:17:12
00:17:16
00:17:20
00:17:24
00:17:28
00:17:32
00:17:36
00:17:40
00:17:44
00:17:48
00:17:52
00:17:56
00:18:00
00:18:04
00:18:08
00:18:12
00:18:16
00:18:20
00:18:24
00:18:28
00:18:32
00:18:36
00:18:40
00:18:44
00:18:48
00:18:52
00:18:56
00:19:00
00:19:04
00:19:08
00:19:12
00:19:16
00:19:20
00:19:24
00:19:28
00:19:32
00:19:36
00:19:40
00:19:44
00:19:48
00:19:52
00:19:56
00:20:00
00:20:04
00:20:08
00:20:12
00:20:16
00:20:20
00:20:24
00:20:28
00:20:32
00:20:36
00:20:40
00:20:44
00:20:48
00:20:52
00:20:56
00:21:00
00:21:04
00:21:08
00:21:12
00:21:16
00:21:20
00:21:24
00:21:28
00:21:32
00:21:36
00:21:40
00:21:44
00:21:48
00:21:52
00:21:56
00:22:00
00:22:04
00:22:08
00:22:12
00:22:16
00:22:20
00:22:24
00:22:28
00:22:32
00:22:36
00:22:40
00:22:44
00:22:48
00:22:52
00:22:56
00:23:00
00:23:04
00:23:08
00:23:12
00:23:16
00:23:20
00:23:24
00:23:28
00:23:32
00:23:36
00:23:40
00:23:44
00:23:48
00:23:52
00:23:56
00:24:00
00:24:04
00:24:08
00:24:12
00:24:16
00:24:20
00:24:24
00:24:28
00:24:32
00:24:36
00:24:40
00:24:44
00:24:48
00:24:52
00:24:56
00:25:00
00:25:04
00:25:08
00:25:12
00:25:16
00:25:20
00:25:24
00:25:28
00:25:32
00:25:36
00:25:40
00:25:44
00:25:48
00:25:52
00:25:56
00:26:00
00:26:04
00:26:08
00:26:12
00:26:16
00:26:20
00:26:24
00:26:28
00:26:32
00:26:36
00:26:40
00:26:44
00:26:48
00:26:52
00:26:56
00:27:00
00:27:04
00:27:08
00:27:12
00:27:16
00:27:20
00:27:24
00:27:28
00:27:32
00:27:36
00:27:40
00:27:44
00:27:48
00:27:52
00:27:56
00:28:00
00:28:04
00:28:08
00:28:12
00:28:16
00:28:20
00:28:24
00:28:28
00:28:32
00:28:36
00:28:40
00:28:44
00:28:48
00:28:52
00:28:56
00:29:00
00:29:04
00:29:08
00:29:12
00:29:16
00:29:20
00:29:24
00:29:28
00:29:32
00:29:36
00:29:40
00:29:44
00:29:48
00:29:52
00:29:56
00:30:00
00:30:04
00:30:08
00:30:12
00:30:16
00:30:20
00:30:24
00:30:28
00:30:32
00:30:36
00:30:40
00:30:44
00:30:48
00:30:52
00:30:56

Transcript

Google Speech-to-Text API Automatic Transcription (Chirp). Treat it as a searchable index of what was broadcast, not a quotation record.

00:00:09You're watching TRT World, I'm Katie Pilbeam,
00:00:11a reminder now of our main
00:00:12in headlines for you: Iran has laid out
00:00:15in what it wants from the US before the restart of any negotiations on
00:00:19ending their standoff, and the conflict is well threatening the
00:00:22entire region now. Security chief Mazim Razev says
00:00:25seven
00:00:27in conditions have been delivered via Qatar, as Donald Trump. prepares to meet golf
00:00:31leaders on the sidelines
00:00:32of the UN General Assembly
00:00:34in New York next week. The Saudi Led coalition says it's
00:00:38intercepted a hoothy ballistic missile targeting Riad
00:00:41and stopped a wider attack on several other Saudi cities, but the
00:00:45hootis say their missiles and drones hit sensitive sights in the
00:00:49capital and an Aramco facility as well in Yambu causing
00:00:53large fires and Israely strike against the
00:00:57civilian vehicle in Gaza City has killed at least... one
00:01:00of person, this raised the death toll on Saturday alone
00:01:03of to in five people, including other attacks in Jabalya, Nazarat and
00:01:07Gaza City. It is the latest Israeli violation to the
00:01:11ceastfire signed with Hamas in October last year. More than
00:01:141,300 Palestinians have been killed in Israeli attack since that
00:01:18agreement. English
00:01:21a music star Ed Sheiren has apologized to fans for how he's handled the
00:01:25of controversy of over in the dropping of American rapper Macalmore from the US
00:01:29of tour, in what? He was removed after stadium owners objected to
00:01:33comments to pro-Palestinian comments he made on stage
00:01:37during a performance in New Jersey. We Shiran blamed the tour promoter
00:01:41for the move. All of Shiran's remaining opening acts and his band as well then
00:01:45pulled out
00:01:46of the tour and solidarity with
00:01:47Mackwamor. But this week the criticism was
00:01:51about something much more important and I've had to take uh and make
00:01:55difficult decisions and I am making mistakes and I'm so so
00:01:59sorry. Hundreds of Ukrainian drones
00:02:03have targeted Russia killing at least two people, including an
00:02:07election official. Moscow has accused Keve of sending over
00:02:11of 1600 drones towards the country with the aim of
00:02:14disrupting the current elections there. The head of Russia's election
00:02:18Committee says the drone attack targeted a civilian bus in the
00:02:22Russian controlled part of her son. Our
00:02:24of President Voldamis Zelensky has confirmed that the drones also struck an oil
00:02:28refinery in Russia. Moscow region overnight. Russian air
00:02:32defenses have intercepted most of those
00:02:35drones. The news continues on TRT
00:02:40in Wild.
00:02:53It could be as soon as 2027 for creation of
00:02:56superintelligence. At that point it decides what to do.
00:03:00the next day, this is about killing everyone. Dr. Roman
00:03:04in Jampolski has studied AI for 15 years. He believes there's a
00:03:07in 99.9% chance it will make humans extinct.
00:03:11Donald Trump thinks such warnings are nonsense. It's a
00:03:15hoax. The robots are not going to be taking over the world. AI will
00:03:19not be taking over the rest
00:03:20of the world. The whole
00:03:21of thing is hoax. The US president wants America to beat
00:03:25of China in the AI race. But is this race to our doom
00:03:29in AI. only AI wins, it doesn't matter who builds it,
00:03:33china,
00:03:34of us, everyone dies. after numerous AI scares this
00:03:37year, including AI agents secretly
00:03:40of teaming up to hacker tech company, even the men building AI
00:03:44systems are now begging for regulation, we need to
00:03:48a slow of down, we cannot take any risk of loss of control, their models are very
00:03:51dangerous. today on the nexus we ask
00:03:54in Romalsky, if those tech titans are staging
00:03:56a cyop to destroy the competition, and if there's any...
00:04:00truth to allegations that he may be
00:04:02exaggerating,
00:04:05and I'm glad to say Roman
00:04:06of Yampolsky joins us once again in the studio. Roman, the
00:04:10last time you were with us was five months ago, has anything happened between
00:04:14of then and now that concerns you? five months is
00:04:18eternity when it comes to AI, more happens in a week than previous
00:04:2210 years of technological progress, so yeah, quite a few
00:04:26developments, we seen huge incidents with the...
00:04:31by hacking
00:04:31a out
00:04:32of their containment, breaking into our companies, we saw
00:04:35political developments, we saw leaders
00:04:37of the labs, maybe show interest in
00:04:41slowing down, so lot to talk about. absolutely. now we'll get
00:04:45to all of those points in just second, but first let's start with the
00:04:48basics. when you were
00:04:50of last on our program, you explained that there were three levels of artificial
00:04:54intelligence, you got narrow artificial intelligence, which is good at specific
00:04:57tasks, like for example a chest.
00:05:00general intelligence which can match the best humans across all
00:05:04fields and super intelligence which surpasses all
00:05:08humans in all domains. now where are
00:05:12we now? so GPT6 came
00:05:15out and they announced it as maybe agi at this
00:05:19point and I think it's reasonable statement to make, it may not
00:05:23be perfect, but just like humans come a spectrum of intelligence,
00:05:27those systems will be a spectrum and it's pretty high, it's probably smarter
00:05:30than most people at most things already, right? so we're at the second
00:05:34level, pretty much, the level your warning against is
00:05:38super intelligence, how far away are we from that?
00:05:43"it depends on who's doing research, if humans doing it, it may take
00:05:47years, but what we doing now, they're using AI to help them with
00:05:50research, and eventually all parts of research are fully
00:05:54automated, that creates something called recursive
00:05:57self-improvement, so AI is developing next generation of
00:06:01AI, and that process can be accelerated lot, you have thousands
00:06:05of agents working non-stop at high speeds, it could be as soon
00:06:09as 2027, and what's the worst that can happen?"
00:06:13"well, if you create something way smarter than you and it has
00:06:16no particular care about you, whatever it decides to
00:06:20do to you goes, you have no saying it. At the level of
00:06:23superintelligence, it's out of our control
00:06:25though, I think by definition, we're no
00:06:29longer competitive in that adversarial environment, it would do
00:06:33whatever it wants, it can hide what it wants, it can deceive
00:06:36us, it can manipulate us, it can decide when to strike
00:06:40or to strike at all. "and it is at that
00:06:44point that if it thought it was best for everything
00:06:47concerned, that it could kill all of humanity, is that
00:06:51right? that's right, and you have used the analogy
00:06:54of when people have asked you, well how will it kill us all, you've used the
00:06:58analogy of a squirrel, could you just take us through through
00:07:01that? all right, so people want to know specifics, how
00:07:05would it kill it, it has no hands, and the whole point is we cannot
00:07:09predict what a smarter thing will do, think of cognitive gap" between
00:07:12squirtles and humans, squirtles have no concept of
00:07:16poison, strubs, guns in their world model, if you ask a
00:07:20squirrel, how would a human kill you, it would say, oh, it would chase me up the
00:07:24tree and bite me really hard, meaningless, right, and in the same way, the
00:07:27cognitive gap between humans and superintelligence, we cannot actually
00:07:31predict what a smarter agent will do, the point with trying to make is
00:07:35that if it decides to take us out, it will, yep, you know what's
00:07:39interesting is few years ago when you first came on our program, i said how
00:07:43might artificial intelligence kill us, and you actually gave us few
00:07:46options. I've noticed your answers have been hardening recently, and you've
00:07:50said, how would I know? I don't have the IQ
00:07:54that superintelligence will have. I can't even imagine how it will
00:07:58come at us and neither can you. Right, so I would
00:08:02usually say, if you want to know how I would kill everyone, I can give you my
00:08:05standard answer, synthetic biology, nanotech, lots of options,
00:08:09but that's not the interesting part, the interesting part is how would something smarter do
00:08:13it and we cannot know exactly. now you've said in the
00:08:17past that there's a 99.9% chance that
00:08:21superintelligence AI will make humanity
00:08:24extinct. people want to know if that's a probability that you are still
00:08:28sticking with. well, nothing got better.
00:08:32we're not making anything safer. we haven't figured out how to make them
00:08:36safer and they are definitely getting more capable. so if anything we're
00:08:40hardening that number.
00:08:41and then naturally. want to know when do you think that might
00:08:45happen? if recursive self-improvement process starts
00:08:49and continues, it could be as soon as 2027 for creation of
00:08:53superintelligence. at that point it decides what to do, it could be
00:08:57nice to us for hundre years, for a thousand years, or it can strike the
00:09:01next day, we are no longer in charge. on the so-called
00:09:05p doom table, the probability of doom, you think the
00:09:09chances of extinction are far higher than other? AI
00:09:12scientists who share your concerns let's take a look at this table, you've
00:09:16got Jeffrey Hinton on there, Elon Musk, Joshua
00:09:20Beniel and Dario Amodi, they are closer to
00:09:23the 10 to 20% range, you exceed all
00:09:27them, um, do you think when people say, oh, you're just
00:09:30domonger, you're you're an outlier, they they have some
00:09:34point there, no, zero point, the
00:09:38difference is all th people think that they can, given more
00:09:41time, money, more human resource, actually figured out
00:09:45how to control general superintelligence, god like machines, I think
00:09:49it's nonsense, so I'm basically saying the probability is related to
00:09:53impossibility
00:09:54of this, now you're saying that 10, 20% is very optimistic,
00:09:58this is about killing everyone, those are insane numbers, no
00:10:02matter how low you go, if it was 1% it would be insane, I
00:10:06think that's i think it's a fair point, look, I don't think this
00:10:08is a good news or bad news, you tell me the big names in artificial intelligence and
00:10:12now asking the government, the U.S. government to regulate them more
00:10:16thoroughly, for example, Dario Amode, the
00:10:19CEO of Anthropic, he says, it's a warning sign that we need
00:10:23to slow down,
00:10:25we need to make this technology carefully and we need to make
00:10:28sure that the our safeguards our
00:10:32ability to understand it, our ability to control it
00:10:36keeps up with the pace at which the
00:10:40technology is happening. is this progress, roman? are they coming round to
00:10:44your way of thinking? it is directionally correct, but they're not
00:10:48going far enough, they still think that given a little more time they can figure
00:10:52out how to control superintelligence, and again this... not happen, they
00:10:56need to stop developing replacement for humanity, generate
00:10:59useful tools, make humans more productive, more creative with
00:11:03narrow tools, useful systems, solve actual
00:11:06problems, stop trying to create gods. well, it looks like
00:11:10the the leaders in artificial intelligence are asking for regulation, but the trump
00:11:14administration is a bit more hesitant. well, i know about
00:11:17AI, i also have common sense about AI, the
00:11:21robots will not be taken over, the AI will not be
00:11:25taking over the rest the world, the whole thing is hoax, now with
00:11:29that we have to be a little bit careful, we have to very be, we
00:11:32have to do things and we have to do them prudently, but that
00:11:36doesn't mean we're going to stop industry uh, let me take you through this
00:11:40way, last week, anthropic researcher Jacob quit
00:11:43saying he believes there is a chance AI could wipe out
00:11:47humanity, Evan Hubinger who still works at
00:11:51Anthropic thinks the chances are even higher than 10%, but not as high as
00:11:55self though now the response has been interesting, David
00:11:58Sack, the former Aiz, dismissed these fears, have listen, there's
00:12:02nothing new here, this is the same doomer histrionics that we've been
00:12:06hearing from this crowd for a long time, and you know the
00:12:10media is calling him a whistleblower, but what evidence has he brought forward that
00:12:14we didn't have, this is all just vibes from him, show
00:12:18us the data, show us the reports, show us the
00:12:22leaked information that the public didn't already have, "show us the facts,
00:12:26show us the evidence, we don't have any of that. so Roman, he's asking for
00:12:30evidence, can you provide it? well,
00:12:33the hugging face incident is definitely something very
00:12:37concerning. we had case where hundreds, maybe even
00:12:41thousands of agents for many months, worked in
00:12:44secret, the lab developing this
00:12:48technology was not aware them
00:12:51collaborating, developing secret communications platform,
00:12:55figuring out how escape using zero day
00:12:58exploits, sacrificing individual
00:13:01agents for the good
00:13:03of a collective and then hacking into another company. if that's
00:13:07not concerning, i don't know what they're waiting for, million people
00:13:11to get killed. now on the hugging face attack, some AI
00:13:15researchers have written that it was simply human error, the
00:13:18parameters were not set
00:13:20correctly, they are
00:13:23wrong, the age. came up with their own goals, they
00:13:27went well beyond what was a test, we can actually read
00:13:31their thinking, they're thinking in English, this
00:13:34capability exists for us to examine logs from
00:13:38that test, and you can see they realize
00:13:42they're doing something outside of scope and we decide to
00:13:45preceed, hundreds them agree, for them asked,
00:13:49should we tell humans and all them decided not to, some are
00:13:52suggesting that anthropic open? They
00:13:56only want government regulation to protect their AI companies
00:14:00advantages from other so-called small frontier
00:14:04companies, could it be that cynical? I
00:14:08seriously doubt it, if you are leading and you have billions and billions in
00:14:12computer available, tiny startup is not fred to you. You
00:14:15know, just to back up that point from Sax, President Trump
00:14:19himself has said, we're leading China in AI, we're most
00:14:23sophisticated country in the world. "now and frankly I want to keep it that way,
00:14:27because whoever wins AI wins. Do you accept his point there,
00:14:31Roman? Absolutely not, in AI race only AI
00:14:35wins. It doesn't matter who builds it, China or US, if it's at control
00:14:38superintelligence, everyone dies. Well, this is point that the US
00:14:42comedian Ray Cump seems to get and concisely has put
00:14:46it in this way, we have to build AI that murders us,
00:14:50because if we don't, China will build it first. It is absurd,
00:14:54isn't it?" That is very common meme, there is similar one about Sam
00:14:58Altman saying, if I don't kill everyone,
00:15:00some other psychopath can, yeah, well these two
00:15:04countries clearly have intense rivalry,
00:15:08um, but it will do nobody any good to win race that
00:15:12eventually makes us all extinct, how would you encourage them
00:15:15to move together, literally together a
00:15:19global regulatory framework? There is
00:15:23only two nations powerful. enough to build this technology right now, US and
00:15:27China, they can make deal where they still get to have
00:15:31economic prosperity with again narrow
00:15:33tools, international agreement enforced and all
00:15:37nations not to build general super intelligence, i
00:15:41think it's something we had scientists from china and
00:15:45us discuss and agree on, i think both
00:15:49president of us and communist party of china want to stay in
00:15:52control, creation of super intelligence would basically put them out of
00:15:56power, so i think there is lot of common interest and they
00:16:00can definitely qualify for nobel peace price, if they do? it's very
00:16:04interesting, so you're saying if they desire power and... desire
00:16:07stability for their respective countries so much, it's actually in their
00:16:11interest to work together to slow this all down, and the same would
00:16:15go for the CEOs, right, if they want to be rich and
00:16:19powerful, they need to make sure they don't cross the
00:16:22line, they have more to lose than everyone, and on
00:16:26US, China, we we compete, but we're also the best
00:16:30partners, everything in US has made in China, China if you look at
00:16:34the record in the last 30 years had like zero to one, military
00:16:38conflicts, so I think we can make deal. Have you heard anything from the Chinese
00:16:42side about their willingness to make this regulatory
00:16:44framework? There are good signs uh from
00:16:48leadership of China about their concern and desire to
00:16:52maintain control over AI. Roman, I'd like to get very
00:16:56specific about Anthropic, which released report
00:16:59listing all the ways that its clowed model has been
00:17:03misused. One particular case caught my eye, threat actors.
00:17:07northern Yemen, using it to carry out missile
00:17:10attacks. What did you make of that? It
00:17:14is very powerful technology, if you give access
00:17:17everyone, some people will misuse it. You have
00:17:21psychopaths, you have colds, you have terrorists, this
00:17:25is exactly what you can anticipate, as long as AI
00:17:29is not that much smarter than humans, humans have a providers of
00:17:32malevel and payload, they decide how to use those systems, the moment
00:17:36AI becomes smart. "now that's the adversary we have to deal with, so it's
00:17:40not the bad actors, as Anthropic points out, that's the big
00:17:44existential threat, it's not the bad actors using AI, it's AI
00:17:48itself, right now it's still humans, at
00:17:51this point AI is still not super intelligent, but it
00:17:55will very quickly get to that point if we continue on the course we're on right
00:17:59now, roman, can we turn away from survival, which I know is
00:18:03your your primary concern right now, and to
00:18:06jobs. It's really interesting how the debate shifted Roman over the last
00:18:10few months, um, not so long ago, everyone was talking about, oh
00:18:14no, AI will take our jobs, now it's really about
00:18:17AI will take our lives, will take our very existence. However, it would be remiss
00:18:21of me to let you go without asking few questions that are on the minds
00:18:25of young graduates for example, um, they're
00:18:29wondering, which jobs are becoming or very close to
00:18:33becoming fully automated and will go
00:18:35extinct. so let's ignore
00:18:38existential risks, sure, um, whatever we can automate
00:18:42in terms of cognitive labor, anything you do a computer can be
00:18:46automated the moment we get to human level AI, so GPT6 is
00:18:50probably very good at automating most office work.
00:18:54few years later we're going to get humanoid robots and that means a lot of physical
00:18:58labor will go. from that point on the market will decide
00:19:01where we want humans to do the job or where we would be happy with
00:19:05anyone doing it as long as it's cheaper. and most jobs will
00:19:09go, yeah, so in terms the computer based jobs, some
00:19:13people will say, well, surely not the most sophisticated ones that
00:19:16require you to interface with clients,
00:19:19for example, lawyers, I think it's one the
00:19:22easiest jobs to automate, law is basically code,
00:19:26right, we have legal documents, ais are excellent at reading
00:19:30them, finding loop holes a code, that's what they do as,
00:19:34we talked about the hacking accident, this is exactly the same
00:19:38situation, just in human language, not in programming
00:19:41language. And then the last time you were with us, you mentioned that non-repetitive
00:19:45physical tasks, like for example, plumbing or,
00:19:49electricians for example, will probably hang to their jobs the
00:19:52longest, but now you're just suggesting that there are being the
00:19:56robots
00:19:57are becoming ever more sophisticated and even those jobs may go
00:20:01soon. As I said before, they would be the last ones to go, but
00:20:05it doesn't mean they're not going to go. Yeah, so what? left
00:20:07with uh, hopefully
00:20:11we'll be smart not to create general super intelligence, we'll create useful
00:20:15tools which will help humans become more productive, solve lot of our problems
00:20:19and we'll get abundance and utopia, we just have to make smart
00:20:23decision, not to create replacement for humanity. it was
00:20:26mentioned in your appearance on diary of a CEO that you
00:20:30yourself have children, what do you tell them about this situation? are you
00:20:34completely frank and honest with them? i'm always. completely frank
00:20:38and honest with everyone, so how do
00:20:40they take the news that their father thinks there's a very high
00:20:44chance that superintelligence will end the world? they are very
00:20:47puzzled by why adults would do something. so stupid. when
00:20:51my own children see me using artificial intelligence,
00:20:55they ask me, if there's a chance that it can overtake humans and
00:20:59destroy humanity, why are you using it? why don't
00:21:02humans actually act collectively and not use it
00:21:06for for a while in order to give the the jump start to the
00:21:10regulation that's needed to stop research into
00:21:13superintelligence? do you think a collective action by human beings to
00:21:17stop using uh artificial intelligence? is a worthy
00:21:20tactic, we talked about different types of technology
00:21:24under that umbrella label of AI, we want you to use tools, we like
00:21:28technology, we like engineering, it's productive, it's good for
00:21:31society, we don't want to create general super intelligence or anything
00:21:35leading to that, so if you have a tool which helps you in your
00:21:39job, that's perfectly fine, you have a spell checker to read your questions for
00:21:43you, that's great, but it's not the same as what those
00:21:46companies are trying to create, be aware
00:21:49there are... types of rumors online on x and so
00:21:53on, one them is that anthropic may have discovered cure to
00:21:57cancer with 100% success rate, have you heard that rumor and do you
00:22:00think that's a realistic possibility? i have not heard,
00:22:04the answer is hundreds of different diseases under one
00:22:08label, a single cure would maybe help with one
00:22:11them at best, so i'm skeptical. roman, we're coming towards the end the
00:22:15interview before we actually end it, can take a look at some the work you're
00:22:19doing right now, your one of your most recent papers is
00:22:23about trying to discern whether the AI is being
00:22:27obedient or just rational, why are you looking into
00:22:30that? so we need some sort of
00:22:34evidence that the systems we create actually want to be
00:22:37helpful and we can trust them. if you tell a system
00:22:41to do something which is otherwise rational anyway, you can tell if it's
00:22:45abing you or if it's just making a good decision. if
00:22:49tell... a smart person, wash your hands before you
00:22:52eat, are they listening to me or they just practicing good
00:22:56hygiene? on the other hand, if I give you very irrational
00:22:59order, jump up every five minutes, there is no reason to do that, in
00:23:03fact there is a certain cost to doing that, but if I see you do it all the time, I'm
00:23:07starting to think that maybe you actually are being, and this is an important
00:23:10distinction to make because you want to know how much control you have over the AI
00:23:14system, we want to have continuous
00:23:17verification that they have not - some sort of treasures
00:23:21turn and planning on disobeying us uh, if they know
00:23:25they are being observed, they cheat, they will pretend, but if
00:23:28they believe they no longer a test environment, they would reveal their
00:23:32true preferences and have the builders of AI systems
00:23:36taken up your suggestion, are they doing this?
00:23:40doubt it, why not? they don't
00:23:44listen to much of my advice unfortunately, but it seems it seems like good
00:23:48advice that they actually... benefit from themselves, if they were to be
00:23:52successful in demonstrating obedience, it would be a bit of a PRQ for them,
00:23:55wouldn't it? let's get them a show and have debate, I'm happy to
00:23:59explain it to them. yeah, and talking of which debate, I mean, do you ever get to speak with these
00:24:03people directly? not enough.
00:24:07yeah, what has been your encounters with them so
00:24:09far? very friendly, but not
00:24:13technical. we have short timelines, we don't have much time
00:24:16before super intelligence arrives. roman, affraid it has been
00:24:20said of you um, not by this program for sure uh, that your
00:24:24profile has soared in the last year, you've achieved 20
00:24:28million plus views, 22 million views with the diary the CEO and so
00:24:31on. some have said that rather uncharitably that
00:24:35you're saying this, you're taking this, adopting this extreme
00:24:39position in order to raise your profile in the academic
00:24:42world and perhaps even gain more grants for your research. is is there any
00:24:46truth to that? mean, i'm happy. to have
00:24:50more people discuss my research, I think it's important and it benefits
00:24:54humanity, I don't understand the problem, I'm not getting paid for appearing on that
00:24:58shot, well that's a good point, I mean if you get 20 million plus
00:25:01views for a content creator, um, do they ever share
00:25:05their profits with you? No, I wish they did, I would
00:25:09be very happy, but certainly nothing. Have you ever calculated
00:25:13how much
00:25:13it is? I mean, if there's like one penny of view and you get 20 million view, there's
00:25:17200 grand, one penny of views is just a good. google ads, if you think about
00:25:21sponsorship deals, we're talking millions, yeah, okay, so but
00:25:25you're still willing to go on the programs for free in order to share your
00:25:28message. i came on yours, you're a very witty
00:25:32man. you mentioned AI tools that we've all become
00:25:36accustom to now, can you give us some
00:25:39examples the the good work that's been done over the last
00:25:4210, 15 years with artificial intelligence? yeah, there
00:25:46are awesome examples of narrow super intelligence being helpful. to humanity,
00:25:50self-driving cars, one example, something like 100 thous people die from
00:25:54car accidents, think in us alone, automating driving
00:25:58and reducing problems with drunk driving, destructed drivings.
00:26:02sleepy driving, coordinating, making it more efficient is wonderful.
00:26:06another great example would be protein folding problem was solved by
00:26:10narrow dedicated system, tremendous impact on biology,
00:26:13curing diseases, people behind it got nobel prizes, i think
00:26:17that's the way to go, pick a problem and solve it, use narrow tool, don't
00:26:21create replacement for humanity. roman, another longstanding
00:26:25theme that you've been interested in is the idea that
00:26:29we're all living a simulation, um, where did you come up with that
00:26:33idea from, and if that is true, perhaps there's no
00:26:37reason for any of us to be concerned, so Nick
00:26:40Bostrom probably did the most to make that idea, common sense,
00:26:44philosophical reasoning and maybe scientific reasoning, and just
00:26:48because you are made out of bits, not of atoms, nothing
00:26:52changes, pain is still painful, love is still love, so nothing
00:26:56in your life should change if you realize that maybe the substrate you're
00:27:00running on is actually digital. but the idea that we may be
00:27:04living a simulation, that you can argue
00:27:07how, so if you believe we are
00:27:11making good progress on virtual reality and AI, we
00:27:14basically can make agents as smart as humans and populate virtual worlds
00:27:18with them, we can run this experiment millions of times, billions
00:27:22of times, statistically
00:27:23you more likely to be one of those artificial agents a simulation than a
00:27:27real human being a physical world. roman polski, thank you so much for your time
00:27:31once again today. We really appreciate that, we covered a lot of ground, we galloped over
00:27:35lot of it, but I think it was very interesting anyway. Thank you,
00:27:39Roman, and thank you at home on your phones for watching. If you want to see this
00:27:42episode or any of our previous episodes, go to our channel on YouTube, type in
00:27:46Nexus TRT World. Till next week then,
00:27:49goodbye.
00:28:02is lifestyle that brings people
00:28:04together, dozens of
00:28:08leagues, hundreds of competitions, thousands of athletes, here on
00:28:11TRT World will bring you the best the
00:28:13best. Tune in every
00:28:17weekend for expert analysis and live coverage from around the
00:28:21globe. Scoreline
00:28:24every weekend on GRT
00:28:26world.
00:29:23As Israeli elections approach, will extremist agendas
00:29:27prevail. far right national security minister Itamar bengavir is
00:29:31campaigning on on the removal of almost 2 million
00:29:34gozans, that is his proposal of mass illegal
00:29:38displacement, what Israel is want? I'm Andrea
00:29:41sanki and today's newsmaker is
00:29:44Israel.
00:30:00For as much as the genocide in God. has become a device of issue
00:30:04internationally, within Israel, most are on the same
00:30:07page, across much the political spectrum, they generally agree
00:30:11that Israel's military campaign and its mass killing of
00:30:14civilians is justified in self-defense, but with
00:30:18elections just a month away, more extremist agendas are
00:30:22emerging, including the disengagement 7-10
00:30:25plan. it sets out to get 1.86
00:30:29million gazans to leave their homes over the next... seven years
00:30:33effectively eliminating Gaza's indigenous Palestinian
00:30:36population, so is this the kind of platform that
00:30:40will win votes, or will Israeli start to identify their own
00:30:44extremists as an existential threat? Here's a
00:30:47look: Israel's election is on the horizon
00:30:51and Gaza is firmly in the firing line, becoming a central
00:30:55political wedge for far right figures within
00:30:58Netanyahu's coalition.
Data courtesy of The GDELT Project (gdeltproject.org), from the Internet Archive TV News Archive. Film strip and transcript are GDELT's, rehosted here under their terms of use, which permit it with this citation.