N1SRP — broadcast
20260920 020000 UTC
370 transcript segments
Google Cloud Speech-to-Text API (Chirp) + Gemini 2.5 Flash Non-Thinking
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.

[00:02:06] image reveals the hidden superpower of Did I do the right
[00:02:09] thing? Should I go see
[00:02:11] doctor? What should I cook for dinner?
[00:02:15] People are turning to artificial intelligence as a constant
[00:02:18] companion, chatting with it almost like it's a close
[00:02:21] friend.
[00:02:24] But what is AI actually good at? When should we
[00:02:27] trust it, and when should we be... more
[00:02:30] skeptical.
[00:02:32] Answers to those questions and much more, this time on DW
[00:02:36] Science Show. Welcome to Tomorrow
[00:02:39] Today.
[00:02:42] Helping with career decisions, replacing psychotherapist, or even
[00:02:46] a friend. Whatever the problem, artificial intelligence always seems
[00:02:50] to have the right answer. But Hamburg-based AI researcher
[00:02:54] Maximilian Keene encourages his students to think critically about
[00:02:58] AI responses.
[00:04:27] Which career path should I choose, the law, medicine?
[00:05:12] my own solution to a problem, instead of just giving me
[00:05:15] answers.
[00:05:17] asking follow-up questions leads to more balanced responses and helps
[00:05:21] build trust, especially among professionals. a study
[00:05:24] found that doctors, for example, often remain cautious about
[00:05:27] AI. that's because experts can sometimes
[00:05:30] underestimate what the technology is actually able to
[00:05:33] do.
[00:05:35] the A3berg has been closed after an Antonov aircraft made
[00:05:39] emergency landing. fortunately, no one was injured.
[00:05:43] AI deep fakes like these have become very convincing in recent years
[00:05:46] and as computing power continues to grow, they're getting even
[00:05:50] more realistic. the same goes for deep fakes circulating on the
[00:05:54] dark web, but is that really innovation?
[00:05:57] just because something's new doesn't make it innovation, there has to be value in
[00:06:01] it. for example, find it very difficult to see deep
[00:06:05] fake sexualized images of children as innovation. it may be
[00:06:08] technically new, but it's not innovation, it's something else. innovation.
[00:06:42] happening when we ask AI a question, behind every
[00:06:46] answer is vast network of computers housed in
[00:06:49] massive, highly secure facilities protected
[00:06:53] from dust and heat, and
[00:06:55] nowhere has more of them than northern Virginia in the U.S., just
[00:06:59] next to Washington DC. They
[00:07:01] take up enormous amounts of space, consume huge amounts
[00:07:05] of electricity, and they're not just an
[00:07:08] isore, they're pretty noisy too.
[00:07:15] 'When Greg Pirio first stepped outside the home he just bought in
[00:07:18] 2012, he was surrounded by greenery, but where there
[00:07:22] was once a forest, there's now data
[00:07:25] center. It's really frustrating, we, you
[00:07:29] know, uh, how they how they could have voted to change the
[00:07:33] zoning to allow this so close to a neighborhood is beyond my
[00:07:37] comprehension. It's not just the view that's
[00:07:40] changed, the soundscape has'.
[00:07:42] too,
[00:07:45] it's just such annoying sound
[00:07:48] of angry because
[00:07:52] it will reduce the prop the value of our properties
[00:07:55] here, angry because it's chase the
[00:07:59] wildlife away from us. the
[00:08:02] area in northern Virginia has become a global hub for data centers
[00:08:06] where the information that powers giant companies such as Uber, Netflix
[00:08:10] and Amazon, as well as from AI applications... is stored and
[00:08:13] processed. AI data centers in particular
[00:08:17] require enormous amounts of electricity.
[00:08:21] th are data centers, look that direction, that direction, behind
[00:08:24] us, all three sides. Elena Schlussberg is also
[00:08:28] frustrated. At Virginia's more than 600 data
[00:08:31] centers, 260 are concentrated in just two residential
[00:08:35] communities. In the 1990s, the area was home
[00:08:39] to AOL, America's first major internet service.
[00:08:42] provider, today it's at the heart of the AI
[00:08:45] boom. look, see another American flag, we're all a
[00:08:49] part of American when we're building out uh, the wealthiest
[00:08:52] industry in the world and paying for their infrastructure.
[00:08:56] northern Virginia leads the world in data center concentration,
[00:08:59] accounting for 14% of all data centers
[00:09:02] globally. running those facilities doesn't require
[00:09:06] many workers, but it does consume vast amounts of
[00:09:09] electricity,
[00:09:11] and that's having a impact on local residence: communities are in
[00:09:15] direct competition with
[00:09:18] industry that has an unprecedented need for power and
[00:09:22] water, and so those things that you normally rely on your
[00:09:25] affordable, clean energy is now being
[00:09:28] consumed by this one
[00:09:31] customer, and that's why you see your utility bills
[00:09:34] increasing. electricity prices here have risen
[00:09:38] by 30% in just 5 years, the enormous demand
[00:09:41] for energy. is one major downside of the AI
[00:09:44] boom, so it looked as though it was a matter of
[00:09:48] economic development and a good thing for the community, but they did
[00:09:52] not think through the implications for the ordinary
[00:09:55] rate payer. industry often needs a certain amount
[00:09:58] of power, but it tends to be distributed and the
[00:10:01] increases are predictable and
[00:10:04] measurable. there's almost nothing that compares to
[00:10:07] this in the past, but there are
[00:10:10] locals who are benefiting. Some residents have joined forces to
[00:10:14] sell farmland for the construction of data centers and power
[00:10:17] plants.
[00:10:20] Mine has been a family farm for generations and
[00:10:24] unfortunately. "my kids and grandchildren have
[00:10:28] absolutely no interest in it, and we came up with a win, win
[00:10:31] win solution, the county's going to make money, the
[00:10:35] um, which helps them pay other people, all their employees,
[00:10:39] more money, get better parks, get better schools for everyone, and yes, we're going
[00:10:43] to make money. data centers generate more than one billion
[00:10:47] dollars in tax revenue every year, and construction
[00:10:50] continues around the clock, but that growth comes a
[00:10:53] cost, depending on it" capacity, a single AI data
[00:10:57] center can consume as much electricity as an entire
[00:11:00] city.
[00:11:03] Keeping buildings cool already accounts for more than 10% of
[00:11:07] global electricity consumption, and demand is only expected to
[00:11:11] rise. Researchers think they may have found a way to
[00:11:15] dramatically reduce the energy needs for air conditioners.
[00:11:18] They're developing new cooling technology that uses special
[00:11:22] metals that heat up when stretched and cool.
[00:11:57] This cooling system uses very little electricity and doesn't require
[00:12:01] environmentally harmful or flammable refrigerants, but
[00:12:04] how? The key lies in bundles of fine
[00:12:08] wires, a bit like artificial muscles that rotate around the cooling
[00:12:11] chamber. They're made from a shape memory
[00:12:14] alloy, material engineered to return to a preset shape
[00:12:18] after being deformed. Paul Motski is a professor of smart
[00:12:22] material systems at Zarland University. He uses
[00:12:25] an infrared camera to show us what's happening. The thermal
[00:12:31] the nickel titanium wires under tension. It's the same
[00:12:35] phenomenon seen in a balloon, a phase transformation.
[00:12:39] Die atome. The atoms arranged themselves differently depending on the temperature and the
[00:12:43] mechanical stress in the material. So when I
[00:12:45] stretch this bundle of wires, I force it into a different
[00:12:48] phase. The atoms reorganize themselves. That
[00:12:52] creates a different energy state. Heat is released. and the material
[00:12:55] warms up. when i release the tension, the atoms rearrange
[00:12:59] themselves again. we get a different energy state: heat is absorbed and
[00:13:03] the bundle cools down again. this is known
[00:13:06] as an elastocaloric effect. it can create temperature
[00:13:09] differences of up to 40 degrees celsius within the material.
[00:13:13] the phenomenon is well known, but only recently have researchers tried to
[00:13:17] harness it for use in refrigerators and heat
[00:13:19] pumps.
[00:13:23] wie führen wir jetzt transfer the heat and cold that we generate to where
[00:13:27] it's actually needed. in a fridge we need to bring the cooling into the
[00:13:30] interior and release the heat into the
[00:13:33] surroundings.
[00:13:35] here circulating air passes over the cooled wires, then flows
[00:13:39] round the cooling chamber circled in blue, causing the contents
[00:13:43] inside to cool down too. the team from
[00:13:46] Zarbrukan demonstrates the technology with this mini refrigerator
[00:13:49] at trade shows and conferences. however, its
[00:13:52] compact design limits its cooling performance. drinks inside can
[00:13:56] only be cool to about 12 degrees celsius. The
[00:14:00] team has also developed more powerful prototypes with cooling
[00:14:03] capacities of up to 250 watts, roughly what an average fridge
[00:14:07] needs. But in computer simulations,
[00:14:10] electricity consumption is about half that of comparable conventional
[00:14:14] systems. The technology is slowly gaining
[00:14:17] momentum. Startups around the world are developing caloric
[00:14:21] heat pumps and refrigerators. Both the
[00:14:23] European Commission. and the US department of energy see enormous
[00:14:27] potential, air conditioners are becoming more common in
[00:14:30] households around the world, in India for
[00:14:33] example,
[00:14:35] and they consume huge amounts of energy, the world's roughly
[00:14:38] 1.6 billion air conditioning units account for around
[00:14:42] 30% of global electricity consumption, according to
[00:14:45] estimates from the international energy agency. That figure has
[00:14:49] at least tripled since 1990.
[00:14:52] Researchers at the center for mechatronics and automation technology
[00:14:56] and convinced will need innovative technologies like this
[00:14:59] to bring rising global energy demand under
[00:15:02] control.
[00:15:05] this is where elastocaloric technology can help. it gives us the
[00:15:09] ability to provide cooling with much lower energy
[00:15:12] consumption, and it does so without producing harmful gases,
[00:15:16] making it safer for both the environment and
[00:15:18] people. the goal is to allow everyone to cool their
[00:15:22] homes without contributing to climate change.
[00:15:37] over time, depending on the alloy and the amount of force
[00:15:41] applied, they can withstand anywhere from tens of thousands to hundreds of
[00:15:45] thousands of cycles. the new
[00:15:47] replacement cartridge makes swapping them out relatively easy. in this
[00:15:51] case, just 28 grams of nickel titanium wire provide
[00:15:54] roughly 100 watts of cooling power, more than enough for the small
[00:15:58] laboratory
[00:15:59] refrigerator.
[00:16:02] the technology is very well suited to places that need to be clean.
[00:16:06] the air passes. over the wires and releases into the room
[00:16:09] without causing any problems.
[00:16:12] the Zarbrukan team believes its cooling systems could find a market in niche
[00:16:16] applications such as lab equipment. They believe the
[00:16:20] technology could be ready within two to three years, but they
[00:16:23] estimate it will be at least another decade before consumers can start
[00:16:27] replacing conventional refrigerators with elastocaloric
[00:16:31] systems.
[00:16:34] As temperatures rise around the world, so does the risk. of
[00:16:37] wildfires, particularly vulnerable are canniferous
[00:16:41] forests, dominated by a
[00:16:44] the same age. one solution could be more
[00:16:48] resilient mixed forests, but choosing the right
[00:16:51] trees to plan for the future is a complex task.
[00:16:55] forest owners have to think decades ahead, could
[00:16:58] AI help them make those
[00:17:00] decisions?
[00:17:07] by future generations and hope i make the right
[00:17:10] decisions.
[00:17:15] if the decision is wrong, i have to make expensive changes, rebuild and
[00:17:18] basically start all over again. to
[00:17:21] create the forest of the future, you first need a lot of information about its
[00:17:25] current condition. a research team at Anhalt University of
[00:17:29] Applied sciences and central Germany is developing an app that
[00:17:32] allows anyone walking in the woods to share their
[00:17:35] observations.
[00:17:39] The app tells you to hold your smartphone horizontally and then you take three
[00:17:43] photos. First it says here, take a picture
[00:17:46] angled up into the tree canopy, then
[00:17:50] it tells me to take a second photo straight ahead into the forest,
[00:17:53] and a third one angled down towards the ground. After
[00:17:57] that, I can share my location and upload the
[00:17:59] images.
[00:18:02] Photos taken by forest visitors are a real treasure trove of data, say the
[00:18:05] researchers.
[00:18:09] every measurement, every piece of information
[00:18:11] helps. They've discovered a medium-sized fallen
[00:18:15] deadwood log heavily decomposed, even that will be
[00:18:18] recorded. The data is meant to provide the
[00:18:22] foundation for a completely new
[00:18:23] approach.
[00:18:26] We're developing an AI model and investigating whether AI can
[00:18:29] assess the condition of a forest based on
[00:18:32] photographs.
[00:18:34] What special is that anyone can take part? As
[00:18:38] citizen scientists, volunteers use the app to take
[00:18:41] photos on their walks.
[00:18:45] The project is called Minervalt car E, meaning my forest
[00:18:49] AI.
[00:18:52] The researchers trained the AI model behind the app using more than
[00:18:56] 6,00 of their own photos from two
[00:18:58] forests.
[00:19:01] The next step is to encourage enough citizen scientists to take
[00:19:04] part. The first version of the free
[00:19:08] myforest AI app is already available for android. It
[00:19:12] should provide invaluable insights about the forest's current
[00:19:15] conditions. But many forest owners need to make changes
[00:19:18] now, like here in Munich. They need to know which
[00:19:22] tree species can best withstand heat, drought and pests.
[00:19:25] And these are big decisions: replacing the ecological benefits
[00:19:29] of a single mature tree, like water storage and carbon
[00:19:32] capture can require up to 400 young trees a cost of
[00:19:36] up to 2500.
[00:20:07] Sylvius Mitchka is one of 2 and a half million forest owners without a
[00:20:11] dedicated forester. That's why he founded a startup called
[00:20:14] Arbero. Using software and AI, he
[00:20:17] calculates which tree species can best survive in specific
[00:20:20] locations. To do that, he purchases data, including from
[00:20:24] the European space
[00:20:25] agency.
[00:20:28] The app combines factors from a wide range of
[00:20:31] data: how acidic is the soil?
[00:20:34] Is location expected to receive lot of rainfall or?
[00:20:40] little, how much sunshine and frost is there, how are
[00:20:43] temperatures changing, and is the terrain flat or
[00:20:47] mountainous? all of this is map.
[00:20:57] The platform shows which of the existing tree species is most
[00:20:59] suitable, both today and in projections extending to the year
[00:21:03] 2100. I can also change the mix of tree species
[00:21:07] and see exactly which ones are best suited to that
[00:21:09] location.
[00:21:12] Today Sylvius is inspecting bark beatle damage with forester Thomas
[00:21:16] Venger whose experience helped with the development of our barrow.
[00:21:21] After the outbreak, many beetle-infested trees had to be
[00:21:24] felled and removed, leaving large stretches almost empty of
[00:21:27] trees.
[00:21:31] The losses were substantial. In the future, the digital forester
[00:21:34] he created will hopefully reduce that risk. But
[00:21:38] these AI supported systems won't make decisions for people, nor will
[00:21:42] they replace foresters. Instead, they provide information to
[00:21:46] support reforestation planning, and that's becoming increasingly.
[00:22:20] siduous and canniferous trees of different ages and
[00:22:22] heights. This creates distinct forest layers, the
[00:22:26] canopy, middle layer and undergroath. Light and
[00:22:29] shade alternate across the forest floor, encouraging different layers of
[00:22:32] growth. Sustainability will also play a role. Only
[00:22:36] individual trees will be removed, a process known as group selection
[00:22:40] harvesting. The gaps allow the forest to
[00:22:44] regenerate without losing its cool, moist
[00:22:46] microclimate
[00:22:49] again. It starts with a single decision, which tree and
[00:22:53] where, which species am I putting my faith in? It's
[00:22:56] difficult, but we have access to huge amount of data, and we should make
[00:23:00] use of it.
[00:23:03] In the future, data will also come from citizen
[00:23:06] scientists, and everyone who contributes gets a little something in
[00:23:10] return.
[00:23:17] When someone uploads a contribution, they receive a thank you.
[00:23:28] Building the forest of the future means making decisions today that will
[00:23:32] affect generations to come. No one knows
[00:23:35] exactly what these forests will look like in a hundred years, but the
[00:23:39] process of creating them has already
[00:23:41] begun.
[00:23:46] And now for something a completely different note a
[00:23:50] New invention allows people who are deaf or heart of hearing
[00:23:53] to feel music through their entire body and get out
[00:23:57] on the dance floor. Now everyone's picking up good
[00:24:00] vibrations.
[00:24:08] Young musicians are competing for recognition, prestige and
[00:24:11] prize money in one of Switzerland's biggest talent
[00:24:14] contests. The crowd is feeling the music, especially
[00:24:18] the people standing on these special. platforms: they convert
[00:24:21] sound into vibrations, sending it directly through the
[00:24:24] body.
[00:24:28] that makes the concerts a tangible experience, even for people
[00:24:31] with hearing
[00:24:32] impairments.
[00:24:37] it makes a huge difference. when i
[00:24:40] go to a party, i can't feel vibrations through my
[00:24:42] body, but i don't pick up the final nuances of the
[00:24:46] music. here with this platform, they really come through.
[00:24:52] wow wow incredible, it's the first time i've experienced anything
[00:24:55] like this, the platform, the music, it's
[00:24:58] fantastic,
[00:25:00] i want more of it, much more, a thousand times
[00:25:04] more, it's amazing, they should
[00:25:08] keep doing
[00:25:08] this,
[00:25:12] until now sign language interpreters could only convey song
[00:25:15] lyrics and the mood of the music,
[00:25:18] now deaf concert goers getting a much more immersive sensory
[00:25:22] experience. the vibration platform is being used at
[00:25:26] concerts for the first time and still requires close supervision from its
[00:25:30] inventorland.
[00:25:36] this is a live concert, not pre-recorded music.
[00:25:39] as soon as the band gets on stage, you have to be ready to make adjustments and
[00:25:43] bring out the frequencies that are best felt through the body.
[00:25:46] ideal sind,
[00:26:04] to follow, the vibration platform is his own
[00:26:06] design,
[00:26:10] this is the tactile transducer,
[00:26:13] inside there's a large coil that creates a magnetic field from the signal we send
[00:26:17] to it,
[00:26:20] that moves a 1.5 kilogram iron core which taps against the
[00:26:23] housing,
[00:26:26] the vibration platform came about during preparations for the Eurovision
[00:26:30] song contest in Basel, was
[00:26:33] part of the communications team at swiss public broadcaster
[00:26:36] SRG.
[00:26:39] The motto was united by music, and we really wanted to embody that
[00:26:43] idea. The goal was to bring as many people together as
[00:26:47] possible, including those who can't hear music, and let them
[00:26:50] experience it. The
[00:26:53] vibration platform made its debut a Eurovision Disco,
[00:26:56] where it proved to hit with people both with and without hearing
[00:27:00] impairments.
[00:27:07] back at the talent competition where installing the technology cost
[00:27:11] organizers several thousand swiss franks, but the team says
[00:27:15] it's money well spent.
[00:27:18] music is for everyone, personally i think it's one of the most beautiful
[00:27:21] experiences there is, every emotion is connected to music, why should
[00:27:25] anyone be left out?
[00:27:28] as deaf community we're often left out, a concert without a
[00:27:31] vibration platform isn't nearly as engaging, here we can take part on
[00:27:35] equal terms.
[00:27:39] the goal is to make live music more accessible and more inclusive
[00:27:43] so that more people can enjoy it together, sharing that extra
[00:27:46] buzz.
[00:27:51] that's all from us for now, but thanks for watching and see you again
[00:27:55] next time on dw science show tomorrow
[00:27:58] today.
[00:27:59] bye-bye