Google Cloud Speech-to-Text API (Chirp) + Gemini 2.5 Flash Non-Thinking. Treat it as a searchable index of what was broadcast, not a quotation record.
00:00:00and they've been dodging taxes,
00:00:02now that, uh, they're watching, uh, how
00:00:06they have to return that tax, they're paying the price of absence.
00:00:10Rebellion, uh, rebel in the name
00:00:13of the future. So that's the only hope.
00:00:16Take your destiny into your own hands. Thank
00:00:20you very much, and a true message to end this morning program. Thank you for
00:00:23the time spent with us. In the past few minutes, it was Professor Enver
00:00:27Kazaz, and this is also the last. Content in the morning
00:00:31program, stay tuned to N1, have a pleasant rest of the day.
00:00:55I am Majila Šabanović, you are watching
00:00:57N1.
00:01:47guess if a
00:02:34for next uh AI industry insiders are warning the
00:02:38rise in the uh intelligence of AI models
00:02:42is risk that is growing too much as you can see I've been joined by
00:02:45Jonathan Richard Schwortz who's the visiting professor at Imperial head of
00:02:49AI research at Thompson Writers Foundation very good morning to you thanks so much for joining us hey good
00:02:53morning um I guess to set this up and in the last sort of
00:02:5724 hours or so um number of
00:03:00threads on x have gone viral, one in particular from a
00:03:03former AI safety worker at both apeni and anthropic
00:03:07that that raised the big question, said that the chance AI one day could
00:03:11kill us all was greater than 10%, right? is that
00:03:14assessment you agree with? um, i think on sort of extinction, i'm perhaps
00:03:18a little bit more sort of cautious about my predictions, but i think what is
00:03:22almost certain now is that we have a trastic reshift of power in society,
00:03:25i think that sort of frontier labs have an immense opportunity
00:03:28to sort of reach. economic power, social power, political power,
00:03:32scientific power, even infrastructure power, and I think it's something that we should very
00:03:36seriously think about a society. um, I guess one of the other central
00:03:39things that's come out in the last 24 hours was the the
00:03:42suggestion that bosses at all of the big AI
00:03:45companies, none in particular,
00:03:48um, are more aware of the risks than they make public, is that
00:03:52something that that you think could be
00:03:54to have stuff about Jesus true? um, I think of course all insiders within those
00:03:57labs probably have have sort of about
00:04:00three to six months horizon of roughly where development is headed, i think that some of
00:04:03the sort of recent developments we have on model development, especially a sort
00:04:07of new idea called um sort of a cursive self-improvement where models
00:04:11kind of just become better and better over time um is something
00:04:14that uh seems to have started to take off and seems to be something that there's a lot of
00:04:18energy in and uh so the bosses but
00:04:21also really all researchers with these slabs will have a unique inside that isn't necessarily public
00:04:25yet. is there a concern throughout the ranks
00:04:28of employees? "this is going too fast, too
00:04:31quickly and and and people have felt they can't speak up, I think that's
00:04:35fair to say, and it's not just the employees, I mean, I have been in this field for over 10
00:04:39years and this year has been one of the most of astonishing amounts of progress, both
00:04:42in terms of the model's capabilities and how fast they're being released, I think the
00:04:46field is becoming more diverse, which is important because it's not sufficient for us to just look
00:04:50at open air and the tropic, there's a huge amount of open-weight models that are being
00:04:53released that are just almost as capable, and really anybody could
00:04:57download them, you and me and increasingly they can be run and..." consumer
00:05:00hardware and so there's a sense here in
00:05:02which sort of safety and security issues are more serious
00:05:06than ever before clearly all of these companies
00:05:10like in any new technology want to want to
00:05:13win, want to get their first and innovate fastest, if we
00:05:17only cared about profit, which clearly we don't, but if we
00:05:21only cared about profit, is it fair to say in this technology
00:05:24more than any others that it's going to be winner takes all or winner takes
00:05:28most, is there that sort of legitimate incentive for them to
00:05:32put the the you know the foot to the floor to try and be first, because
00:05:36if they don't then somebody else will eat all of their lunch, mean that's
00:05:39certainly how they're operating right, i think there's perhaps the space. for more kind of pluralistic world in the
00:05:43ice space, but the the frontier labs operating this kind of speed right, and it's really
00:05:47kind of prisoner dilemmas kind of situation where the sort
00:05:51of uh self-interest of these firms is to move as fast as they
00:05:54possibly can, and unless there's some sort of external arbiter where a
00:05:58situation where they just keep raising right, and the sort of voluntary
00:06:01agreements that people have tried to put in place haven't really been able to prevent that off slow
00:06:05things down, because guess to follow on question for that, which if there is
00:06:08that prize for someone at the end of it is when you get
00:06:12onto the nationalistic debate, which is if you don't
00:06:16allow the, let's narrow it down to two countries for
00:06:19argument sake, the US companies, one of them at least
00:06:23to be the winner, then the alternative could be a Chinese company being the winner, is that
00:06:27fair way of looking at it? um, I would say that lot of people,
00:06:30including lot of my own academic work is trying to kind of get away from that dynamic,
00:06:34right? there's a huge amount of energy and effort about the sort of idea of sovereign
00:06:38ai, nation, states institutions trying to develop their
00:06:41own models and sort of get a... from this dynamic and being fully reliant on
00:06:44others uh in this country and other middle powers there's a lot of interesting work being done so
00:06:48I don't think we're in a situation where it's inevitable that you know sort
00:06:51of either Chinese or or US slaps are dominating and everybody else is sort of left behind I think
00:06:55there are things that can be done but the opportunity windows shrinking I
00:07:00regulator impos rules or restrictions
00:07:04that slow down the US's ability to innovate
00:07:07would that prisoners dilemma risk still apply that
00:07:10China might not do the same well I mean that's the conversation people having right,
00:07:14frankly uh sort of if regulation is is only applied in a
00:07:17certain jurisdiction or a certain country uh it's it's sort
00:07:20of huge opportunity for oters to kind of catch up or even exceed
00:07:24the ability so really i think what's needed is a sort of multilateral
00:07:27international sort of way of of of of coordinating this
00:07:31um we do have an AI safety institute here in the UK mean many people say
00:07:35it's one of the world leaders but guess are we now a
00:07:38point where it's wishful thinking to to expect
00:07:42that that UK- based institution could make a difference
00:07:45in this? yeah, mean ac, the the AI security institute has been
00:07:48incredibly influential and it's really seen as one of the world class institutes in this
00:07:52space, but do think that we're in a position where it's no longer clear if we can fully rely on this
00:07:56voluntary excess, right? i think it's also fair to say that ac has primarily
00:08:00worked with the US labs and in a world where more and more models are being provided by more and
00:08:04more institutions, it's not necessarily clear whever one single
00:08:07institution can completely scale, so do think we need something that goes beyond a
00:08:11single kind of countries. institution and something like ac a more multilateral international
00:08:14basis would certainly be very valuable. bring us back to where we started and and the
00:08:18scale of the risk that you that you think
00:08:22that we as as society face from AI
00:08:25and if it's not 10% chance of total wipe
00:08:29out, is it underpice in the way that we talk about it in terms
00:08:32of significant risks to society? yeah, mean
00:08:36i personally think that we are um, i think this 10% figure is
00:08:39something that's very difficult to sort of assess right, it's more like a gut. feeling then a proper sort
00:08:43of populistic measure, but I do imagine that the kind of systematic
00:08:47risk like the power describe the sort of economic impacts and so those are
00:08:51almost certain to be happening, I mean there's already a sort of situation where arguably AI
00:08:54developers are are making sort of large judgments on what human values
00:08:58are, uh sort of decisions that usually happen throughout the political process right, and
00:09:02this is already happening, and it's kind of troubling that people like me or my
00:09:05colleagues get to the side what model should be saying on sort of some of the most contested issues in the
00:09:09world right, and sometimes I believe that the extinction debate is... important, but
00:09:13there's lots of other kind of debates here that are kind of falling a little bit by the way
00:09:16side. what if people say that this is normal? this is
00:09:20normal in in a process of technological
00:09:22innovation
00:09:25and law makers and regulators will catch
00:09:28up and there'll always be a sort of balance that that evolves over
00:09:32time. yeah, i would i would say that's probably mistaken. mean, if we look at past
00:09:35technologies, comparison, people like to make this industrial revolution that
00:09:39sort of ran over 80 years in in a limited number of. sectors and already there the
00:09:43consequences were quite dire for many people until some of the
00:09:47benefits were to be seen, think in the internet sort of more recent
00:09:50example, lot of the predictions of how things would turn out ended up being true, i
00:09:54think the level of uncertainty and the scale and speed of things is just different this time.
00:09:58um, jonathan, thanks so much for joining us, we really appreciate it. jonathan richard
00:10:01schwarz from
00:10:02imperial.
00:12:08Prayer apps, digital scriptures and virtual communities,
00:12:11faith is moving to our screens, from AI, spiritual
00:12:15guides and tiktok priests to worship in virtual reality,
00:12:18technology is changing how people believe and belong,
00:12:22but can it deepen faith or does something get lost
00:12:25along the way:
00:12:29faith on our phones: you version
00:12:32says its family of bible apps has now passed 1 billion
00:12:35installs. for churches, apps and social media are no longer just
00:12:39extras, they are becoming important ways to reach
00:12:42people. in 2025, the Vatican held its first jubilee
00:12:46for digital missionaries with more than 1ous catholic
00:12:50influencers taking part in Rome or online. the
00:12:53goal is to connect with people who... may never walk through a church
00:12:56door. In Poland, one young priest is taking that
00:13:00mission to tik tok. On tiks, there are many
00:13:03people who wants to like fight with me, so I bless
00:13:07them, and yeah, that's what they do. This is
00:13:11father Sebastian. He's 30 and has more than 420,0
00:13:14followers on TikTok.
00:13:17For more than six years, he served as a priest in Saviege, a Polish
00:13:21town of around 4500 people. He
00:13:24als has a pretty unusual
00:13:27sideline,
00:13:29so I've got this tik to coffies, I bought this boxes, here
00:13:33I sit, and I put some music, indie,
00:13:36uh, Christian indie folk, that's what I love, Christian
00:13:40indie like that,
00:13:43uh, yeah, and I start, and I'm going live,
00:13:47Sebastian began posting on tiktok in October 2022,
00:13:50after watching another priest livestream, he thought it was something he could do too,
00:13:54and... began broadcasting himself,
00:13:58not all of his fellow priests approved, but sebastian carried on,
00:14:02it worked, so what keeps people
00:14:05watching? my tiktok world
00:14:09is um is filled with variety, i know that
00:14:13my first mission is to, as i said to lead teenagers
00:14:16closer to christ, so i want
00:14:19Christ and and the church, but i know that i
00:14:22want like catch them when
00:14:25if i want if i will be doing only that kind of south, that's why
00:14:28i've got some kind of lifestyle content or about video
00:14:32games flaki,
00:14:36poland is still one of europe's most catholic countries, but even
00:14:40here fewer people are practicing their faith, especially among
00:14:43genz. in 1992, 69%
00:14:46of pols aged 18 to 24 said they regularly attended religious
00:14:49services, by 2021 that had dropped to just
00:14:5223%.
00:14:55but christianity is not shrinking
00:14:57everywhere,
00:15:00it's growing fastest in Sub Saheran Africa, driven
00:15:03largely by its young expanding population.
00:15:06By 2020 the region was home to about 700 million Christians
00:15:10compared with about 500 million in Europe,
00:15:14but back in Poland, the church is losing touch with many young people,
00:15:18Sebastian wants to bring them back. I want to catch people
00:15:21who uh don't follow these caltics, go to church, they
00:15:25don't have any contacts with this religion stuff, but they do have a
00:15:29TikTok, my mission in TikTok is to be bridged from this their
00:15:33world to the true world from God, catching
00:15:36young people. where they already are can make faith feel more personal
00:15:40and more accessible, but that same access can also be exploited.
00:15:44Islamist preachers have learned to use social media to build large, often
00:15:48young audiences, their messages can blend into ordinary
00:15:51everyday content, and algorithms can gradually narrow the
00:15:55range of use users see, serving them more and more of the
00:15:59same. Experts warn that in many recent cases, much of the path
00:16:02to radicalization has unfolded online, sometimes beginning on
00:16:06TikTok before. moving into private chat groups in Germany,
00:16:10one young Muslim woman challenges extremis and misogenistic
00:16:13interpretations of her faith, that's made her a
00:16:16target of some very ugly backlash. German
00:16:20influencer Noor el huda Shihade began receiving death
00:16:23threats after she stopped wearing her hijab. She
00:16:27says it no longer reflected who she was. It's
00:16:30pretty obvious that I'm Muslim, I don't believe that wearing a hija protects
00:16:34women from being sexualized, I just couldn't identify. it or
00:16:38couldn't find a personal reason for it, so i stopped wearing
00:16:40it. the clip quickly went viral on tiktok where
00:16:44algorithms often amplify content that spark strong
00:16:47reactions. the same algorithms also amplify islamist
00:16:51preachers, some with hundreds of thousands of followers. what
00:16:54experts note, this starkly reduces the influx of variable
00:16:58content, pushing young users into potential extremist
00:17:01bubbles. this online radicalization is a growing
00:17:04problem.
00:17:14It becomes dangerous when there's no counterbalance, if the only
00:17:17voices out there are hate preachers or misogenistic scholars and there's
00:17:21no one speaking up against them, offering a different perspective and
00:17:25providing proper information, then the situation can become
00:17:29very dangerous.
00:17:31that's why she speaks out against islamist extremism and misogy
00:17:34online. she says that the... has not stopped,
00:17:38but neither has she.
00:17:41I have a clear goal in mind, I really want to protect women from
00:17:44religious abuse, and I'm not losing focus. Sometimes it can hit you
00:17:48hard, especially when things get very personal, but my goal is
00:17:52much stronger than all that hate.
00:17:56Some Islamist creators present strict ideology in everyday
00:17:59advice on marriage, clothing and
00:18:02behavior. They present traditional gender roles as religious
00:18:05fact. Experts. It's important to have visible
00:18:09alternative voices that show young users a diverse picture of
00:18:12Islam.
00:18:15Using technology to spread the word of God or debate faith is one
00:18:19thing, but now technology itself is becoming part of the
00:18:22religious experience. Several online AI characters are designed
00:18:26to simulate Jesus and answer users with responses
00:18:29based on biblical texts. They promise
00:18:32personal spiritual guidance, but they also turn intimate questions of
00:18:36faith in. to paid business, seeing God's hand in
00:18:40everyday moments brings true peace, you can
00:18:43talk to AI Jesus for just few dollars a
00:18:45minute.
00:18:48It was created by a US-based tech startup, which offers paid
00:18:51video calls through its website and app. AI Jesus was trained on
00:18:55the Bible and select sermons. It has
00:18:57focused set of information and purpose that's about the
00:19:01spirituality and the guidance and mentorship.
00:19:05I Jesus remembers previous conversations. That's meant to
00:19:08make exchanges more personal and foster an emotional
00:19:11connection. It's all part of much bigger market. Analyst
00:19:14value spiritual wellness apps, including meditation, mindfulness and
00:19:18faith-based apps at more than two billion dollars with
00:19:22AI playing a growing role. We have
00:19:25this huge field of spiritualities
00:19:28and the mindfulness and the
00:19:31strive for immortality and self-optimatization,
00:19:34which I would also call religious and in this... field AI
00:19:38is playing a huge
00:19:41role. AI systems generate language by predicting
00:19:45likely words, they have no faith or spiritual experience.
00:19:48Still, almost 40% of Christian Genzia said AI
00:19:52spiritual advice could be as trustworthy as
00:19:54pastors, according to a survey carried out by a faith
00:19:58tech company. What AI is also
00:20:01doing is creating immersive
00:20:05experiences we that we associate
00:20:08with feelings of transcendence and religions
00:20:11and here huge question is
00:20:15is this authentic? hi Jesus, my name is jam. the
00:20:19company says AI jesus is not intended to replace or
00:20:23scripture, but to offer comfort and guidance based on christian
00:20:26teachings, but there are concerns. researchers worn
00:20:30that data protection rules for AI chatbots are often unclear.
00:20:33studies show these systems can also produce false information and reproduce.
00:20:37bias: the experience may feel
00:20:41spiritual, but the service remains
00:20:43commercial. can an AI model on a religious
00:20:47figure offer meaningful spiritual? guidance or does it
00:20:51turn faith into a money maker? the AI
00:20:54Jesus exists only in the cloud, but physical AI is
00:20:57also entering religious spaces. a temple in Kyoto, an AI
00:21:01powered robot has been designed to discuss buddhist teachings and
00:21:05reconnect people with their faith. this is
00:21:07budaroid, japan's first AI robot monk, hard at
00:21:11work in Kyoto's showing in temple. it can walk and sit
00:21:14in lotus position. once
00:21:18again i bring my palms together in hard.
00:21:26Buddhism teaches us to avoid violence and hurtful words, follow ethical
00:21:30principles and calmly
00:21:33train the mind. It might remind you of Minder, this humanoid used
00:21:36in another Kyoto temple launched in 2019, but Minder
00:21:40can only chant pre-recorded sermons and can't have a
00:21:43dialogue. Budhoroid was developed to offer spiritual advice and
00:21:47spark renewed interest in Buddhism in Japan.
00:21:51It was developed by Saji Kumagai and we
00:21:54consult regarding our uh daily what I
00:21:58say problems and troubles or
00:22:01anxiety with human
00:22:04monks, but um front of human uh sometimes we
00:22:08feel very shy to uh talk
00:22:11honestly on the other hand with
00:22:15robot or with AI chatbot it's very
00:22:18easy to talk.
00:22:21Buddhism is one of the largest religions in Japan, more than a third of the
00:22:24population identified as Buddhist in 2020, but the number is
00:22:28falling fast, down 12% within a
00:22:30decade.
00:22:33fewer young people feel connected to the faith, still not everyone's
00:22:37convinced about using AI to change
00:22:39that.
00:22:42there are many ways to answer
00:22:44questions.
00:22:49ai can solve mathematic.
00:23:26I am not a buddhist monk, but I have given the role of conveying the teachings of Buddhism and listening
00:23:29everyone's stories.
00:23:32Japan isn't the only place exploring how robots might interact with
00:23:35religion. In the United Arab Emirates, a humanoid robot
00:23:39dressed in traditional emirati clothing also sparked
00:23:43debate. clips appearing to show buosunaida praying in a
00:23:46mosk through criticism online. Some social media. users
00:23:50argued that prayer requires intention, consciousness and
00:23:53faith, qualities a machine cannot possess. the
00:23:57robot's inventor is clear about this limits. robot is
00:24:00not cannot be muslim or cannot
00:24:03be in a religion, it's robot. we take
00:24:06it to improve our life. when faith goes digital,
00:24:10who gets the final say? the believer, the religious community,
00:24:14or the people behind the algorithm. there are no simple
00:24:18answers and this debate is only: beginning, so what do you
00:24:22think, would you turn to an AI for spiritual advice?
00:24:25let us know, that's all from me, thanks for watching, bye.
00:30:07the anatomy of the human body: sure, it's
00:30:10complex, but we've been studying ourselves for centuries, and
00:30:13know what we're made of, right? doctors
00:30:16today are supposed to learn the names and locations of every bone,
00:30:20muscle and organ, and they do, at least in
00:30:24men. but how do you explain that researchers have just
00:30:27discovered structures in the female pelvic area that don't
00:30:31appear in textbooks?
00:30:35That's and much more from the world of medicine this time around
00:30:39tomorrow
00:30:40today.
00:30:45Researchers have rediscovered a forgotten bit of female
00:30:48anatomy. Cells of the ovary are pictured here in
00:30:52red, but what's that green thing? I had no idea what this
00:30:56was because the structure was not really depicted in any...