N1HRV — broadcast
20260920 050000 UTC
382 transcript segments
Google Speech-to-Text API Automatic Transcription (Chirp)
Data courtesy of The GDELT Project (https://www.gdeltproject.org/), from the Internet Archive TV News Archive.
Machine transcription. Treat it as a searchable index of what was broadcast, not a verbatim quotation record.

[00:00:04] Did of I do the right thing? Should I
[00:00:07] in go see doctor? What should I cook for
[00:00:10] dinner? People are turning to artificial intelligence as a
[00:00:14] constant companion, chatting with it almost like it's
[00:00:17] a close
[00:00:19] a friend. But what is AI actually good
[00:00:23] at? When should we trust it, and when should we be more
[00:00:40] career decisions, replacing psychotherapist or even a
[00:00:43] of friend,
[00:00:43] in whatever the problem, artificial intelligence always seems to have the
[00:00:47] right answer, but Hamburg-based AI researcher
[00:00:51] Maximilian Keeneer encourages his students to think critically about
[00:00:54] AI responses. Die wachsen jetzt in
[00:00:58] einer. machine instantly answers any
[00:01:01] question, that's Maximilian Keaner's avatar
[00:01:05] talking, the researcher programmed it himself, thereby making
[00:01:08] himself replaceable, at least according to the
[00:01:11] AI. I could handle your
[00:01:14] emails, your
[00:01:16] of reports, even this interview, and honestly I do a pretty good
[00:01:18] job, yeah, maybe emails, that's just
[00:01:21] optimization, but talking to students, interacting, showing
[00:01:25] empathy, taking
[00:01:27] of responsibility, you can't do that, you're
[00:01:28] of just imitating.
[00:01:29] 30
[00:01:45] of
[00:01:52] of
[00:02:05] photos of
[00:02:06] every
[00:02:11] single cherry down to the smallest detail and that's no problem for
[00:02:15] AI, but cherries aren't
[00:02:17] people. what
[00:02:20] AI simply can't do is make value
[00:02:23] of judgements, which career path should I choose, the law, medicine,
[00:02:27] something
[00:02:28] of else? there isn't one correct answer. It's not a
[00:02:30] calculation, the problem we keep
[00:02:33] of running into is that we try to squeeze every complex problem into an AI
[00:02:37] format, das KI Format drücken
[00:02:40] wollen. In Keino's view, ethics doesn't slow down the development
[00:02:44] of AI. It's what makes AI genuinely useful for people in the first
[00:02:48] place. A team from the universities of Hornheim and
[00:02:51] Bamber has developed an AI that doesn't always have a quick answer.
[00:02:55] This socratic AI openly admits when it isn't entirely
[00:02:59] sure.
[00:03:05] a
[00:03:14] questions leads to
[00:03:16] Asking follow-up more balanced responses
[00:03:17] of and helps build trust, especially among professionals.
[00:03:21] A study found that doctors, for example, often remain cautious about
[00:03:24] AI. That's because experts can sometimes underestimate what
[00:03:28] the technology is actually.
[00:03:34] of
[00:03:40] AI of deep fakes like these have become very convincing in recent years and
[00:03:44] as computing power continues to grow they're getting even more
[00:03:47] realistic. the same goes for deep fakes circulating on the dark
[00:03:51] web, but is that really innovation? just because
[00:03:55] something's new doesn't make it innovation, there has to be value in
[00:03:58] it. for
[00:03:59] of example,
[00:04:16] of
[00:04:30] calculations, but Kina says humans still need to make the
[00:04:33] decisions. It's
[00:04:37] easy to forget what's really happening when we ask AI a
[00:04:41] question. Behind every answer is vast network of
[00:04:44] computers housed in
[00:04:47] of massive highly secure facilities protected from dust and
[00:04:50] heat, and nowhere has more
[00:04:53] of them than northern
[00:04:54] of Virginia in the US, just next to Washington
[00:04:56] DC. They take up enormous amount. of
[00:05:00] space consume huge amounts of electricity, and
[00:05:04] they're not just an isore, they're pretty noisy
[00:05:07] too.
[00:05:12] Pirio first stepped outside the home he just bought in 2012,
[00:05:16] he was surrounded by greenery, but where there was once a
[00:05:20] forest, there's now data
[00:05:22] center. It's really frustrating, we you
[00:05:25] know uh, how they how they could have voted to change the
[00:05:29] zoning to allow this so close to a neighborhood is beyond
[00:05:33] my comprehension. It's not just the view that's
[00:05:37] changed, the soundscape has
[00:05:39] too. just such
[00:05:43] annoying sound of angry
[00:05:46] because it will reduce the prop the
[00:05:50] value of our properties here, angry
[00:05:53] because it chase the wildlife away from
[00:05:57] us. the area in northern Virginia has become a
[00:06:00] global hub for data centers where the information that powers giant
[00:06:04] companies such as Uber, Netflix and Amazon as well as from AI
[00:06:08] applications is stored and processed.
[00:06:12] centers in particular require enormous amounts of
[00:06:15] electricity. these are data centers,
[00:06:19] look that direction, that direction behind us, all three
[00:06:22] sides. Elena Schlussberg is also
[00:06:25] frustrated. At Virginia's more than 600 data
[00:06:28] centers, 260 are concentrated in just two residential
[00:06:32] communities. In the 1990s, the area was home to
[00:06:36] AOL, America's first major internet service
[00:06:39] provider. Today it's at the heart the AI boom.
[00:06:43] Look, see another American flag. We're all a part of America when we're building
[00:06:47] out uh, the wealthiest industry in the world and paying for their
[00:06:50] infrastructure. Northern Virginia leads the world in data
[00:06:54] center concentration, accounting for 14% of all data
[00:06:58] centers globally. Running those facilities
[00:07:02] doesn't require many workers, but it does consume vast amounts of
[00:07:06] electricity, and that's having an impact on
[00:07:09] local residents. Communities are in direct. competition
[00:07:13] with industry that has an unprecedented
[00:07:17] need for power and water, and so those things that you normally rely on
[00:07:21] your affordable, clean energy is now being
[00:07:25] consumed by this one
[00:07:27] customer, and that's why you see your utility bills
[00:07:31] increasing. electricity prices here have risen by
[00:07:35] 30% in just five years, the enormous demand for
[00:07:38] energy is one major downside the AI boom. so
[00:07:42] it looked as though it was matter of economic development and a good thing for
[00:07:46] the community, but they did not think through the
[00:07:49] implications for the ordinary rate payer. industry often
[00:07:53] needs a certain amount of power, but it tends to be
[00:07:57] distributed and the increases are predictable and
[00:08:01] measurable. there's almost nothing that compares to
[00:08:04] this in the past. but there are locals who are
[00:08:08] benefiting. some residents have joined forces to sell
[00:08:11] farmland for the... construction of data centers and power
[00:08:14] plants, mine has been a family
[00:08:18] farm for generations and
[00:08:20] unfortunately my kids and grandchildren have
[00:08:24] absolutely no interest in it, and we came up with a win win
[00:08:28] solution, the county is going to make money, the
[00:08:32] um, which helps them pay other people, all their
[00:08:35] employees more money, get better parks, get better schools for
[00:08:38] everyone, and yes we're going to make money.
[00:08:42] more than $1 billion tax revenue every year, and
[00:08:46] construction continues around the clock, but that growth comes a
[00:08:49] cost. Depending its capacity, a single AI data
[00:08:53] center can consume as much electricity as
[00:08:56] an entire
[00:08:56] city.
[00:09:00] Keeping buildings cool already accounts for
[00:09:03] more than 10% of global electricity consumption, and
[00:09:07] demand is only expected to rise. Researchers think they
[00:09:11] may have found a way to dramatically. reduce the energy needs for air
[00:09:14] conditioners. They're developing new cooling technology that
[00:09:18] uses special metals that heat up when stretched and cool down again
[00:09:22] when released. Only enough room
[00:09:26] for a single soda can, but this mini refrigerator is packed with
[00:09:29] innovative technology. Researchers at Zalland
[00:09:33] University in Southwest Germany have built it to demonstrate a new calling
[00:09:37] method that could be more energy efficient and
[00:09:39] climate-friendly.
[00:09:44] We need less energy for cooling, that means lower energy consumption
[00:09:48] overall, which ultimately means lower CO2 emissions
[00:09:51] globally. This cooling system uses
[00:09:55] very little electricity and doesn't require environmentally harmful or
[00:09:59] flammable refrigerants, but how? The key
[00:10:03] lies in bundles of fine wires, a bit like artificial muscles that
[00:10:06] rotate around the cooling chamber. They're made from a shape
[00:10:10] memory alloy. engineered to return to a preset
[00:10:14] shape after being deformed.
[00:10:17] is a professor of smart material systems at Zaland
[00:10:20] University. He uses an infrared camera to show us what's
[00:10:24] happening. The thermal image reveals the hidden superpower
[00:10:28] the nickel titanium wires under tension. It's the same
[00:10:31] phenomenon seen a balloon, a phase
[00:10:34] transformation. The atoms arrange themselves
[00:10:37] differently depending on the temperature and the mechanical stress in the
[00:10:40] material. So when I stretch this bundle of wires, I force it into
[00:10:44] different phase. The atoms reorganize
[00:10:47] themselves. That creates a different energy state, heat is
[00:10:51] released and the material warms up. When I release the tension.
[00:10:55] the atoms rearrange themselves again, we get different energy state,
[00:10:59] heat is absorbed and the bundle cools down again. this is known
[00:11:03] as an elastocaloric effect. it can create temperature differences
[00:11:06] of up to 40 degrees celsius within the material. the phenomenon
[00:11:10] is well known, but only recently have researchers tried to harness it for
[00:11:14] use in refrigerators and heat
[00:11:16] pumps.
[00:11:21] how do we transfer the heat and cold that we generate to where it's actually needed?
[00:11:25] In a fridge, we need to bring the cooling into the interior and release the heat
[00:11:29] into the surroundings. Here,
[00:11:33] circulating air passes over the cooled wires, then flows round the
[00:11:36] cooling chamber circled in blue, causing the contents inside to cool
[00:11:40] down too. The team from Zarbrucken demonstrates the
[00:11:44] technology with this mini refrigerator at trade shows and
[00:11:47] conferences. However, its compact design limits its cooling
[00:11:51] performance. Drinks inside can only be cooled to about 12 degrees.
[00:11:56] The team has also developed more powerful prototypes with cooling
[00:12:00] capacities of up to 250 watts, roughly what an average fridge
[00:12:04] needs, but in computer simulations, electricity
[00:12:07] consumption is about half that of comparable conventional
[00:12:10] systems. The technology is slowly gaining
[00:12:14] momentum. Startups around the world are developing caloric heat
[00:12:18] pumps and refrigerators. Both the European Commission and the
[00:12:21] U.S. Department of Energy see enormous potential.
[00:12:25] air conditioners are becoming more common in households around the world, in India for
[00:12:29] example, and they consume huge amounts
[00:12:33] of energy. The world's roughly 1.6 billion air conditioning
[00:12:37] units account for around 30% of global electricity
[00:12:40] consumption, according to estimates from the international energy
[00:12:43] agency. That figure has at least tripled since
[00:12:46] 1990. Researchers at the center for
[00:12:50] mechatronics and automation technology in Zarbrucken are convinced will need
[00:12:54] innovative.
[00:13:25] But that goal
[00:13:26] is still a long way off. For now, the latest prototype is tackling
[00:13:30] different challenge. The nickel titanium wire bundles wear
[00:13:34] out over time. Depending on the alloy and the amount of force
[00:13:37] applied, they can withstand anywhere from tens of thousands to hundreds of
[00:13:41] thousands of cycles. The new replacement cartridge makes
[00:13:45] swapping them out relatively easy. In this case, just 28
[00:13:49] grams of nickel titanium wire provide roughly 100 watts of cooling
[00:13:53] power, more than enough for the small.
[00:13:59] The technology is very well suited to places that need to be clean, the
[00:14:03] air passes directly over the wires and releases into the room without causing
[00:14:07] any problems. The Zarbrukan team believes it
[00:14:10] cooling systems could find a market in niche applications such as lab
[00:14:14] equipment. They believe the technology could be ready within 2 to
[00:14:18] 3 years, but they estimate it will be at least another
[00:14:22] decade before consumers can start replacing. conventional
[00:14:25] refrigerators with elastocaloric
[00:14:27] systems. As
[00:14:31] temperatures rise around the world, so does the risk of
[00:14:34] wildfires? Particularly vulnerable are canniferous
[00:14:37] forests, dominated by a single species and trees the
[00:14:41] same age. One solution could be more resilient
[00:14:45] mixed forests, but choosing the right trees to plant for the
[00:14:49] future is a complex task. Forest owners have to
[00:14:53] think decades ahead. AI help them make those
[00:14:56] decisions?
[00:15:02] what i plan here today will be harvested by future generations and hope i make the
[00:15:06] right
[00:15:06] decisions.
[00:15:11] if the decision is wrong, i have to make expensive changes, rebuild and basically
[00:15:15] start all over again. to create the forest the
[00:15:19] future you first need a lot of information about its current
[00:15:22] condition. a research team. University of Applied
[00:15:26] Sciences in central Germany is developing an app that allows
[00:15:29] anyone walking in the woods to share their
[00:15:31] observations.
[00:15:37] smartphone horizontally, and then you take three
[00:15:40] photos. first,
[00:15:42] it says here, take a picture angled up into the tree
[00:15:44] canopy, then it tells me to take a second photo straight
[00:15:48] ahead into the forest, and a third one angled down towards the
[00:15:52] ground. after that, i can share my location and upload
[00:15:56] the images. photos taken by forest visitors are
[00:16:00] a real treasure trobe of data, say the researchers
[00:16:02] every measurement,
[00:16:06] every piece. information helps, they've discovered a
[00:16:10] medium-sized fallen deadwood log, heavily decomposed,
[00:16:14] even that will be recorded, the data is meant to
[00:16:18] provide the foundation for a completely new
[00:16:20] approach. we developing an AI
[00:16:24] model and investigating whether AI can assess the condition of a forest
[00:16:28] based on photographs. what
[00:16:31] special is that anyone can take part? as citizen
[00:16:35] scientists, volunteers the app to take photos on their
[00:16:39] walks. The project is called
[00:16:43] Minervalt Kai, meaning My Forest
[00:16:45] AI. The
[00:16:49] researchers trained the AI model behind the app, using more than
[00:16:52] 6,00 their own photos from two
[00:16:55] forests. The next step is to
[00:16:59] encourage enough citizen scientists to take
[00:17:01] part. The first version the free
[00:17:05] myforest AI app is already available. for android. It should
[00:17:09] provide invaluable insights about the forest's current
[00:17:12] conditions, but many forest owners need to make changes
[00:17:15] now, like here in Munich. They need to know which tree species can
[00:17:19] best withstand heat, drought, and pests, and these are big
[00:17:23] decisions. Replacing the ecological benefits of a single mature
[00:17:26] tree, like water storage and carbon capture, can require up to
[00:17:30] 400 young trees a cost of up to €25,00 euros per
[00:17:34] hectare, and they're having to make these decisions.
[00:18:07] dedicated forester. That's why he founded a startup called
[00:18:11] Arbero. Using software and AI, he calculates which
[00:18:15] tree species can best survive in specific locations. To do
[00:18:18] that, he purchases data, including from the European space
[00:18:22] agency. The app
[00:18:25] combines factors from a wide range of data: how
[00:18:29] acidic is the soil, is location expected to receive lot
[00:18:33] of rainfall or very little. How much
[00:18:37] sun? shine and frost is there, how are temperatures
[00:18:41] changing, and is the terrain flat or
[00:18:44] mountainous? all of this is mapped with precision down to just
[00:18:47] few square meters, using data from the year 2000 up to
[00:18:51] now. die plattform, the the
[00:18:54] platform shows existing
[00:18:54] which tree species is most suitable, both today and in
[00:18:58] projections extending to the year 20100. i can also
[00:19:02] change the mix of tree species and see exactly which ones are best suited to
[00:19:06] that location. Today Sylvius is
[00:19:10] inspecting bark beetle damage with forester Thomas Venger whose
[00:19:14] experience helped with the development of our barrow.
[00:19:17] After the outbreak, many beetle infested trees had to be felled and removed,
[00:19:21] leaving large stretches almost empty of
[00:19:24] trees. The
[00:19:27] losses were substantial. In the future, the digital forester he
[00:19:31] created will hopefully reduce that risk, but these
[00:19:35] AI supported systems won't make
[00:20:08] Experts envision something like this: uniform
[00:20:11] monocultures such as spruce forests will be replaced by mixed
[00:20:15] forests made up of deciduous and canniferous trees of different ages
[00:20:19] and heights. This creates distinct forest layers: the
[00:20:22] canopy, middle layer and undergrowth; light and shade alternate
[00:20:26] across the forest floor, encouraging different layers of growth.
[00:20:30] Sustainability will also play a role. Only individual trees will
[00:20:34] be removed, a process known as group selection harvesting.
[00:20:39] the gaps allow the forest to regenerate without losing its cool,
[00:20:43] moist
[00:20:43] microclimate. it
[00:20:47] starts with a single decision, which tree and where? Which species am I
[00:20:51] putting my faith? It's difficult, but we have access to
[00:20:55] huge amount of data and we should make use of
[00:20:57] it. In the future,
[00:21:01] data will also come from citizen scientists, and everyone who
[00:21:05] contributes gets a little something in
[00:21:07] return.
[00:21:14] When someone uploads a contribution, they receive a thank you message, because we're
[00:21:17] grateful for every submission from the public. They also
[00:21:21] receive some information about the forests information.
[00:21:25] Building the forest the future means making decisions today that will affect
[00:21:29] generations to come. No one knows exactly what these forests
[00:21:33] will look like a hundred years, but the process of creating them
[00:21:37] has already
[00:21:37] begun.
[00:21:47] allows people who are deaf or hard of hearing. to feel
[00:21:51] music through their entire body and get out on the dance
[00:21:54] floor. now everyone's picking up good
[00:21:57] vibrations.
[00:22:04] young musicians are competing for recognition, prestige and
[00:22:08] prize money in one of Switzerland's biggest talent contests.
[00:22:12] the crowd is feeling the music, especially the people standing on these
[00:22:16] special platforms. they convert sound into vibrations.
[00:22:20] sending it directly through the
[00:22:21] body that makes the
[00:22:25] concerts a tangible experience even for people with hearing
[00:22:29] impairments.
[00:22:33] it makes a huge difference. when i go to a party, i
[00:22:37] can feel vibrations through my body, but i don't pick up the finer
[00:22:41] nuances the music. here with this platform, they really come
[00:22:45] through. wow, wow,
[00:22:48] incredible. This is the first time I've experienced anything like
[00:22:52] this, the platform, the music, it's
[00:22:54] fantastic. I want more of
[00:22:58] it, much more, a thousand times
[00:23:01] more. It's amazing. They should keep
[00:23:04] doing this.
[00:23:08] Until now, sign language interpreters could only convey song, lyrics and the
[00:23:12] mood the music. Now deaf concert goers are
[00:23:16] getting much more immersive, sensory experience.
[00:23:20] the vibration platform is being used at concerts for the first time and
[00:23:24] still requires close supervision from its
[00:23:26] inventorland.
[00:23:32] this is a live concert, not pre-recorded music. as soon as the
[00:23:36] band gets on stage, you have to be ready to make adjustments and bring out the
[00:23:40] frequencies that are best felt through the
[00:23:42] body.
[00:23:46] event technology specialist roland developed the platform largely from home.
[00:23:53] large coil that creates
[00:23:55] a
[00:24:10] This is the tactile transducer, inside there's
[00:24:12] a magnetic field from the signal we send to
[00:24:14] it.
[00:24:23] The vibration platform came about during preparations for the Eurovision
[00:24:26] song contest in Basel. Roland was part the
[00:24:30] communications team at Swiss Public Broadcaster
[00:24:33] SRG. The moto
[00:24:37] was united by music and we really wanted to embody that
[00:24:40] idea. The goal was to bring as many people together as
[00:24:44] possible, including those can't hear music and let them
[00:24:47] experience it.
[00:25:20] is connected to music, why should anyone be left
[00:25:22] out? as deaf community, we're often
[00:25:26] left out, a concert without a vibration platform isn't nearly as
[00:25:30] engaging, here we can take part on equal
[00:25:32] terms. the goal is to
[00:25:36] make live music more accessible and more inclusive so that more
[00:25:40] people can enjoy it together, sharing that extra
[00:25:43] buzz.