Become a supporter of this podcast: https://www.spreaker.com/podcast/spacetime-with-stuart-gary--2458531/support.
00:00:00 --> 00:00:02 Stuart Gary: This is space time series 29 episode
00:00:02 --> 00:00:05 117 full broadcast on 30
00:00:05 --> 00:00:08 September 2026 coming up on Space
00:00:08 --> 00:00:11 Time, NASA names Dragonfly's
00:00:11 --> 00:00:13 landing site on the saturnian moon Titan.
00:00:14 --> 00:00:16 New clues about the Earth's moon's ancient
00:00:16 --> 00:00:19 magnetic field and why galaxies in the
00:00:19 --> 00:00:21 early universe look so weird.
00:00:22 --> 00:00:25 All that and more coming up on, um, Space
00:00:25 --> 00:00:28 Time. Welcome to
00:00:28 --> 00:00:30 Space Time with Stuart G.
00:00:46 --> 00:00:48 NASA has identified the landing site where
00:00:48 --> 00:00:51 its Dragonfly rotocopter, uh, will first
00:00:51 --> 00:00:53 touch down on the surface of the saturnian
00:00:53 --> 00:00:55 moon Titan. The target area
00:00:56 --> 00:00:58 comprises a large dune field with some inter
00:00:58 --> 00:01:00 dune terrain south of Silic Crater and
00:01:00 --> 00:01:02 stretches along the edge of a range of
00:01:02 --> 00:01:04 mountains or hills. The International
00:01:05 --> 00:01:07 Astronomical Union, the global body
00:01:07 --> 00:01:09 responsible for officially designating
00:01:09 --> 00:01:11 objects in space, has approved naming the
00:01:11 --> 00:01:14 region Amakik uh Andai. First part of
00:01:14 --> 00:01:16 the name is Mayan people appealed to the
00:01:16 --> 00:01:18 spirit Amakaq to stop strong winds from
00:01:18 --> 00:01:21 damaging their crops. The name literally
00:01:21 --> 00:01:23 translates into the one who locks up the
00:01:23 --> 00:01:26 wind. An ande aligns with the International
00:01:26 --> 00:01:28 Astronomical Union's convention of naming
00:01:28 --> 00:01:31 dune fields in Latin after gods and goddesses
00:01:31 --> 00:01:34 of wind. Meanwhile, construction work
00:01:34 --> 00:01:37 continues on the Dragonfly spacecraft itself,
00:01:37 --> 00:01:39 which is now being assembled in a clean room
00:01:39 --> 00:01:41 at the Johns Hopkins Applied Physics
00:01:41 --> 00:01:43 Laboratory in Lorell, Maryland. Cables are
00:01:43 --> 00:01:45 being carefully installed in the flight
00:01:45 --> 00:01:47 fuselage, providing the Titan bound
00:01:47 --> 00:01:49 rotorcraft with its central nervous system
00:01:50 --> 00:01:52 now. Ah. Collectively, these bundles of
00:01:52 --> 00:01:54 wires, cables and connectors make up the
00:01:54 --> 00:01:56 spacecraft's electrical harness, which will
00:01:56 --> 00:01:59 transmit power and data between the lander's
00:01:59 --> 00:02:01 computers, actuators, sensors, scientific
00:02:01 --> 00:02:04 instruments and battery. Dragonfly
00:02:04 --> 00:02:06 lander harness lead Jacqui Perry says it's a
00:02:06 --> 00:02:08 key milestone for the engineers and
00:02:08 --> 00:02:11 technicians building the vehicle. The wire is
00:02:11 --> 00:02:13 silver coated copper, insulated with a heat
00:02:13 --> 00:02:15 resistant durable polymer coating and then
00:02:15 --> 00:02:17 wrapped in aluminium, which is then attached
00:02:17 --> 00:02:19 at plastic and metal connectors.
00:02:19 --> 00:02:22 Perry says the wiring harness could only be
00:02:22 --> 00:02:24 installed once the flight structure was
00:02:24 --> 00:02:26 delivered and the remote interface units and
00:02:26 --> 00:02:28 temperature sensors were installed. The
00:02:28 --> 00:02:30 harness is typically one of the first
00:02:30 --> 00:02:32 components delivered to a spacecraft.
00:02:33 --> 00:02:35 Dragonfly passed its critical design review
00:02:35 --> 00:02:38 in 2022. Fabrication of the vehicle began
00:02:38 --> 00:02:41 in late 2024 and finished last year.
00:02:42 --> 00:02:44 Dragonfly is slated for launch in 2028 and
00:02:44 --> 00:02:47 will reach the Saturnian moon in late 2034.
00:02:48 --> 00:02:50 Dragonfly's operating environment on Titan
00:02:50 --> 00:02:53 poses some unique challenges because of the
00:02:53 --> 00:02:56 rotorcraft's thermos bottle design. Insulated
00:02:56 --> 00:02:58 to retain heat from its nuclear power source
00:02:58 --> 00:03:01 so it can stay warm in Titan's extremely cold
00:03:01 --> 00:03:03 conditions, the harness had been designed to
00:03:03 --> 00:03:05 route under a lay foam insulation on the
00:03:05 --> 00:03:07 outside of the lander and accommodate the
00:03:07 --> 00:03:09 circulation of warm air through the inside.
00:03:10 --> 00:03:12 Perry says the rotorcraft's high power
00:03:12 --> 00:03:14 demands require both 4 and 8Ah gauge
00:03:14 --> 00:03:16 wire, yet the harness still had to be
00:03:16 --> 00:03:18 flexible enough to weave through the packed
00:03:18 --> 00:03:21 interior, which includes the flight systems
00:03:21 --> 00:03:24 and instrument boxes as well as the 136
00:03:24 --> 00:03:26 kilogramme battery. Dragonfly
00:03:26 --> 00:03:29 will investigate Titan's prebiotic chemistry
00:03:29 --> 00:03:32 and assess the moon's habitability. It's not
00:03:32 --> 00:03:34 primarily a life detection mission, but aims
00:03:34 --> 00:03:36 to study how far organic chemistry has
00:03:36 --> 00:03:39 progressed in an environment rich in carbon
00:03:39 --> 00:03:41 compounds, which includes possible past
00:03:41 --> 00:03:44 mixing of liquid water and organics. The
00:03:44 --> 00:03:47 car sized 875 kilogramme lander
00:03:47 --> 00:03:50 features twin quadrotor engines powered by a
00:03:50 --> 00:03:52 radioisotope thermoelectric generator
00:03:52 --> 00:03:54 providing electricity and heat. The
00:03:54 --> 00:03:57 3.85 metre long vehicle is designed to
00:03:57 --> 00:04:00 fly several kilometres between landing sites,
00:04:00 --> 00:04:02 allowing it to sample diverse locations.
00:04:03 --> 00:04:05 Mission managers expect to cover over 115
00:04:05 --> 00:04:08 kilometres during Dragonfly's 3.3 year
00:04:08 --> 00:04:10 primary mission, exploring a range of
00:04:10 --> 00:04:12 different environments from organic dunes to
00:04:12 --> 00:04:15 deposits associated with an impact crater.
00:04:15 --> 00:04:17 Silk Crater, which we mentioned earlier,
00:04:17 --> 00:04:19 where liquid water and complex organic
00:04:19 --> 00:04:21 materials, which are key to life as we know
00:04:21 --> 00:04:24 it, once existed together. Scientific
00:04:24 --> 00:04:26 analysis indicates the impact that formed
00:04:26 --> 00:04:28 Silk Crater melted the icy bedrock,
00:04:28 --> 00:04:30 potentially creating a large temporary pool
00:04:30 --> 00:04:32 of water that could have remained liquid for
00:04:32 --> 00:04:35 hundreds of thousands of years under an
00:04:35 --> 00:04:37 insulating ice layer like winter ponds on the
00:04:37 --> 00:04:37 Earth.
00:04:38 --> 00:04:40 This report from NASA TV
00:04:41 --> 00:04:42 Saturn's largest
00:04:42 --> 00:04:45 NASA TV: moon, Titan, has a thick atmosphere and a
00:04:45 --> 00:04:48 frozen surface rich in organic molecules.
00:04:48 --> 00:04:50 In 2034, a NASA mission called
00:04:50 --> 00:04:53 Dragonfly will arrive at Titan and study its
00:04:53 --> 00:04:56 chemical makeup. Dragonfly is a
00:04:56 --> 00:04:59 rotorcraft designed to visit multiple sites
00:04:59 --> 00:05:00 across the moon's varied terrain.
00:05:03 --> 00:05:05 At each new landing site on Titan's surface,
00:05:05 --> 00:05:08 Dragonfly uses a pulsed neutron generator
00:05:08 --> 00:05:11 and onboard gamma ray sensor to detect key
00:05:11 --> 00:05:13 elements such as carbon and hydrogen in
00:05:13 --> 00:05:16 organic materials or oxygen in water
00:05:16 --> 00:05:19 ice. Dragonfly determines if there are
00:05:19 --> 00:05:21 well defined layers of these materials just
00:05:21 --> 00:05:24 below the lander for a closer inspection.
00:05:24 --> 00:05:27 Dragonfly uses its drill to generate
00:05:27 --> 00:05:30 tailings from Titan's hard frozen surface.
00:05:30 --> 00:05:33 These surface samples can then be ingested
00:05:33 --> 00:05:35 through the pneumatic system carried with
00:05:35 --> 00:05:38 Titan air into the chilled sample lines into
00:05:38 --> 00:05:41 the sample collection carousel. One of the
00:05:41 --> 00:05:43 carousel sample cups is placed in a pneumatic
00:05:43 --> 00:05:46 port. The cup captures the surface
00:05:46 --> 00:05:48 material from the cold air stream and
00:05:48 --> 00:05:50 transfers it to the chemical laboratory for
00:05:50 --> 00:05:53 measurement. Pulses from a laser
00:05:53 --> 00:05:55 release large organic molecules from the
00:05:55 --> 00:05:57 surface sample for analysis in the Mass
00:05:57 --> 00:06:00 spectrometer. The mass spectrometer sorts
00:06:00 --> 00:06:03 molecules by mass and measures diagnostic
00:06:03 --> 00:06:05 fragments that tell Dragonfly the kinds of
00:06:05 --> 00:06:07 chemical components that are present in the
00:06:07 --> 00:06:09 surface and whether there are molecules of
00:06:09 --> 00:06:12 prebiotic interest. For those
00:06:12 --> 00:06:15 potential prebiotic samples, a new cup is
00:06:15 --> 00:06:17 placed into an oven and heated to release
00:06:17 --> 00:06:19 molecules into a gas chromatograph, where
00:06:19 --> 00:06:21 they are sorted for size and type before
00:06:21 --> 00:06:24 entering the mass spectrometer. This
00:06:24 --> 00:06:26 advanced separation of organic components
00:06:26 --> 00:06:29 includes isolating molecules with the same
00:06:29 --> 00:06:32 formula but different chiral arrangements, or
00:06:32 --> 00:06:32 handedness.
00:06:33 --> 00:06:35 Having a preference for one handedness over
00:06:35 --> 00:06:38 another is a key biosignature for life on
00:06:38 --> 00:06:40 Earth. When the chemical analysis is
00:06:40 --> 00:06:42 complete, Dragonfly may choose to take
00:06:42 --> 00:06:45 another surface sample or find a new location
00:06:45 --> 00:06:47 on Titan to investigate.
00:06:47 --> 00:06:50 Stuart Gary: This whole mission's quite a challenge. Titan
00:06:50 --> 00:06:52 is unique in our solar system. It has a
00:06:52 --> 00:06:55 diameter of 5 kilometres,
00:06:55 --> 00:06:58 making it Saturn's largest moon and 50%
00:06:58 --> 00:07:00 bigger than the Earth's moon. In fact, it's
00:07:00 --> 00:07:03 larger than the planet Mercury. It's also the
00:07:03 --> 00:07:05 only moon with a substantial atmosphere and
00:07:05 --> 00:07:07 the only world other than Earth where clouds
00:07:07 --> 00:07:10 release rain that form streams and rivers,
00:07:10 --> 00:07:12 which then flow into lakes and seas.
00:07:12 --> 00:07:15 But Titan is so cold, the water there is
00:07:15 --> 00:07:18 normally frozen solid, forming bedrock. And
00:07:18 --> 00:07:20 instead of water, the liquid rain on Titan is
00:07:20 --> 00:07:23 made up of methane and ethane hydrocarbons.
00:07:24 --> 00:07:27 But scientists describe Titan as analogous to
00:07:27 --> 00:07:29 the early Earth. That's because it preserves
00:07:29 --> 00:07:31 a set of environmental and chemical
00:07:31 --> 00:07:32 conditions which are thought to have existed
00:07:32 --> 00:07:35 on our planet billions of years ago, before
00:07:35 --> 00:07:37 life arose and before free oxygen
00:07:37 --> 00:07:40 accumulated in our atmosphere. It
00:07:40 --> 00:07:42 therefore serves as a natural laboratory for
00:07:42 --> 00:07:45 studying prebiotic chemistry on a planetary
00:07:45 --> 00:07:48 scale. Scientists think its complex organic
00:07:48 --> 00:07:50 chemistry, methane cycle and possible
00:07:50 --> 00:07:52 environments could support exotic forms of
00:07:52 --> 00:07:54 life, or at least prebiotic chemistry.
00:07:55 --> 00:07:57 It's also considered one of the more feasible
00:07:57 --> 00:07:59 destinations for future human exploration
00:07:59 --> 00:08:02 among the outer solar system bodies due to
00:08:02 --> 00:08:04 its thick atmosphere for radiation shielding
00:08:04 --> 00:08:05 and available resources.
00:08:06 --> 00:08:09 Titan was extensively studied by the Cassini
00:08:09 --> 00:08:11 Huygens mission, which orbited the Saturnian
00:08:11 --> 00:08:14 system between 2004 and 2017.
00:08:15 --> 00:08:17 The Huygens Lander was deployed from the
00:08:17 --> 00:08:19 Cassini spacecraft down to Titan's surface on
00:08:19 --> 00:08:22 December 25, 2004, landing
00:08:22 --> 00:08:25 on January 14, 2005, near the
00:08:25 --> 00:08:27 Ediri region, the boundary between brighter
00:08:27 --> 00:08:30 highlands and darker plains. The lander
00:08:30 --> 00:08:33 spent about 90 minutes on the surface taking
00:08:33 --> 00:08:35 readings and measurements before the
00:08:35 --> 00:08:37 batteries ran out. Huygens was able to
00:08:37 --> 00:08:39 confirm that Titan has a thick, nitrogen
00:08:39 --> 00:08:42 dominated atmosphere with methane and a range
00:08:42 --> 00:08:44 of aerosols and complex organic chemistry.
00:08:45 --> 00:08:47 It measured strong High altitude zonal winds
00:08:47 --> 00:08:50 on its way down and weaker winds near the
00:08:50 --> 00:08:52 surface. It also measured a surface
00:08:52 --> 00:08:55 temperature of -179 degrees Celsius
00:08:56 --> 00:08:59 and a surface atmospheric pressure about 50%
00:08:59 --> 00:09:00 higher than sea level here on Earth.
00:09:01 --> 00:09:04 Intriguingly, it found the ground on Titan
00:09:04 --> 00:09:06 was soft and damp, with the consistency of
00:09:06 --> 00:09:08 wet sand, clay or lightly packed snow.
00:09:09 --> 00:09:11 Images showed the surface of Titan was
00:09:11 --> 00:09:13 littered with water ice, pebbles or cobbles a
00:09:13 --> 00:09:16 few centimetres across on an orange tinted
00:09:16 --> 00:09:18 organic rich plane. There was also evidence
00:09:18 --> 00:09:21 of past liquid flows near the landing site
00:09:21 --> 00:09:23 and possible methane release from the
00:09:23 --> 00:09:25 subsurface. Needless to say, as the
00:09:25 --> 00:09:28 Dragonfly mission progresses, we'll keep you
00:09:28 --> 00:09:31 informed. This is space
00:09:31 --> 00:09:31 time.
00:09:32 --> 00:09:35 Still to come, new clues about the Earth's
00:09:35 --> 00:09:37 Moon's ancient magnetic field. And we examine
00:09:37 --> 00:09:40 why galaxies in the early universe look so
00:09:40 --> 00:09:42 weird. All that and more still to come
00:09:43 --> 00:09:44 on space time.
00:09:53 --> 00:09:55 A new study claims that just like the present
00:09:55 --> 00:09:58 day Earth, our moon once had a geodynamo
00:09:58 --> 00:10:01 driven magnetic field. On Earth, the
00:10:01 --> 00:10:03 movement of liquid iron in the planet's outer
00:10:03 --> 00:10:05 core generates its global magnetic field.
00:10:06 --> 00:10:09 This so called geodynamo works on a similar
00:10:09 --> 00:10:11 principle to a dynamo on a bicycle, which
00:10:11 --> 00:10:13 converts mechanical motion into electrical
00:10:13 --> 00:10:16 energy. But unlike today's Earth, the Moon
00:10:16 --> 00:10:18 no longer has a core generated magnetic
00:10:18 --> 00:10:21 field. One of the study's authors, Anna
00:10:21 --> 00:10:23 Mittelholz from EDH Zurich, says there's a
00:10:23 --> 00:10:25 heated ongoing debate about whether the Moon
00:10:25 --> 00:10:28 also operated a dynamo in the past.
00:10:28 --> 00:10:30 That's because analysis of rock samples
00:10:30 --> 00:10:32 brought back to Earth by the Apollo
00:10:32 --> 00:10:35 astronauts are somewhat contradictory. Some
00:10:35 --> 00:10:36 researchers assume there was a strong
00:10:36 --> 00:10:38 magnetic field that existed over a long
00:10:38 --> 00:10:41 period of time, somewhere around 4.25 to
00:10:41 --> 00:10:44 3.5 billion years ago, while others find
00:10:44 --> 00:10:47 absolutely no evidence of this. You see, in
00:10:47 --> 00:10:48 addition to the dynamo theory, there's a
00:10:48 --> 00:10:50 second possible explanation for the
00:10:50 --> 00:10:53 magnetised lunar rocks, namely impacts from
00:10:53 --> 00:10:55 massive meteorites or asteroids, which could
00:10:55 --> 00:10:57 have triggered a, uh, magnetization process
00:10:57 --> 00:10:59 on the Moon. But the new research
00:10:59 --> 00:11:02 reported in the journal Science Advances now
00:11:02 --> 00:11:05 supports the dynamo theory. Mittelholz and
00:11:05 --> 00:11:07 colleagues say that around 4.2 billion years
00:11:07 --> 00:11:09 ago, the Moon did indeed possess an
00:11:09 --> 00:11:12 internally generated magnetic field. Uh, now
00:11:12 --> 00:11:15 this claim isn't based on rock samples, but
00:11:15 --> 00:11:17 on data collected by probes in lunar orbit,
00:11:17 --> 00:11:19 including gravity measurements by NASA's
00:11:19 --> 00:11:21 GRAIL probes and magnetic field models.
00:11:21 --> 00:11:23 Drawing on orbital measurements from the
00:11:23 --> 00:11:25 Lunar Prospector, UH and Kagua missions
00:11:26 --> 00:11:28 to reach their conclusions, the authors
00:11:28 --> 00:11:30 examined the Dewa region on the lunar far
00:11:30 --> 00:11:32 side, which is never seen from Earth.
00:11:32 --> 00:11:35 Mittelholz says Dewa has one of the strongest
00:11:35 --> 00:11:37 magnetic field anomalies on the far side of
00:11:37 --> 00:11:39 the Moon. And a distinct gravity anomaly
00:11:39 --> 00:11:42 coincides spatially there as well as that
00:11:42 --> 00:11:44 means this region contains rocks that were
00:11:44 --> 00:11:46 more strongly magnetised, while at the same
00:11:46 --> 00:11:49 time denser than elsewhere. In most cases,
00:11:49 --> 00:11:51 the origin of magnetic anomalies measured
00:11:51 --> 00:11:54 from lunar orbits are known. But the gravity
00:11:54 --> 00:11:56 data gives scientists an insight into the
00:11:56 --> 00:11:58 density and consequently the type of material
00:11:58 --> 00:12:01 beneath the surface. Mittelholz says where
00:12:01 --> 00:12:03 the magnetic field and gravity signals
00:12:03 --> 00:12:05 coincide, it is possible to combine the two
00:12:05 --> 00:12:07 and attribute the anomaly to a specific
00:12:07 --> 00:12:10 geological feature. And that's where Dewar
00:12:10 --> 00:12:13 comes in. It allowed the authors to create an
00:12:13 --> 00:12:15 accurate model of the subsurface by jointly
00:12:15 --> 00:12:18 processing gravity and magnetic field data.
00:12:18 --> 00:12:20 Uh, this data shows the area beneath
00:12:20 --> 00:12:23 Dewar's surface contains a rock body
00:12:23 --> 00:12:25 approximately 60 kilometres wide, extending
00:12:25 --> 00:12:27 to a depth of around nine kilometres.
00:12:27 --> 00:12:30 It's far denser than the surrounding crust
00:12:30 --> 00:12:33 and it's strongly magnetised. Combined with
00:12:33 --> 00:12:35 a surface geochemistry and an arch
00:12:35 --> 00:12:37 topography, the authors conclude that it
00:12:37 --> 00:12:39 solidified magma that's risen from deep
00:12:39 --> 00:12:41 below, part of a buried volcanic complex.
00:12:42 --> 00:12:45 The age of the structure, 4.2 billion years,
00:12:45 --> 00:12:47 was determined from various deposits of
00:12:47 --> 00:12:49 impact material on the lunar surface.
00:12:49 --> 00:12:51 Betelholz says because they know how much
00:12:51 --> 00:12:54 iron is present in such a rocky body, they
00:12:54 --> 00:12:55 can estimate the minimum strength the
00:12:55 --> 00:12:57 magnetic field must have had as the magma
00:12:57 --> 00:13:00 slowly cooled. It shows that at the time, the
00:13:00 --> 00:13:02 magnetic field on the Moon was likely more
00:13:02 --> 00:13:05 than 10 micro Tesla. Now, by comparison,
00:13:05 --> 00:13:07 here on Earth today, the planet's magnetic
00:13:07 --> 00:13:10 field stands at around 50 micro Tesla. The
00:13:10 --> 00:13:12 authors have ruled out the possibility of an
00:13:12 --> 00:13:14 impact crater creating the magnetic field.
00:13:14 --> 00:13:16 Because there's no matching cratering signs,
00:13:17 --> 00:13:19 it's still unclear how the small lunar core
00:13:19 --> 00:13:21 could have generated such a strong magnetic
00:13:21 --> 00:13:24 field. So the authors admit the existence of
00:13:24 --> 00:13:26 an early lunar geodynamo isn't yet fully
00:13:26 --> 00:13:29 resolved. This study is also
00:13:29 --> 00:13:31 providing fresh insights into another
00:13:31 --> 00:13:33 puzzling phenomena lunar swirls.
00:13:34 --> 00:13:36 These bright, curved and striped patterns on
00:13:36 --> 00:13:38 the Moon's surface stand out clearly against
00:13:38 --> 00:13:41 their darker surroundings. Wherever such
00:13:41 --> 00:13:44 swirl patterns appear, scientists always find
00:13:44 --> 00:13:47 magnetic anomalies. And there is a lunar
00:13:47 --> 00:13:49 swirl on the surface of the Dewar region,
00:13:49 --> 00:13:52 right above the observed anomaly. The origin
00:13:52 --> 00:13:55 of the swirls remains a matter of debate. One
00:13:55 --> 00:13:57 possible explanation is that the swirls only
00:13:57 --> 00:13:59 form where the magnetic field runs
00:13:59 --> 00:14:01 horizontally at the surface. As is the case
00:14:01 --> 00:14:03 with the Dewar swirl, the horizontal field
00:14:03 --> 00:14:06 deflects the solar wind, thereby protecting
00:14:06 --> 00:14:08 the surface from weathering. And as a result,
00:14:08 --> 00:14:10 the area remains brighter than its
00:14:10 --> 00:14:12 surroundings, Mittelholz says That would be
00:14:12 --> 00:14:15 important information for future astronauts,
00:14:15 --> 00:14:17 as magnetic field lines could offer
00:14:17 --> 00:14:19 protection from the solar wind, and the
00:14:19 --> 00:14:21 swirls would indicate the locations of that
00:14:21 --> 00:14:24 shielding. This is space time.
00:14:25 --> 00:14:27 Still to come. Why galaxies in the early
00:14:27 --> 00:14:29 universe look so weird? And later in the
00:14:29 --> 00:14:31 Science report, discovery of a common
00:14:31 --> 00:14:33 chemical in plastics, which has been linked
00:14:33 --> 00:14:36 to the development of ADHD and autism.
00:14:37 --> 00:14:39 All that and more still to come on, uh, space
00:14:39 --> 00:14:39 time.
00:14:55 --> 00:14:57 A new study suggests that the first galaxies
00:14:57 --> 00:14:59 in the universe looked a bit weird because
00:14:59 --> 00:15:02 their stars generated weaker stellar winds
00:15:02 --> 00:15:05 compared to those we see today. The new
00:15:05 --> 00:15:06 research could reshape science's
00:15:06 --> 00:15:09 understanding of how galaxies formed in the
00:15:09 --> 00:15:12 early universe. The findings reported in the
00:15:12 --> 00:15:14 Astrophysical Journal are showing that the
00:15:14 --> 00:15:16 more astronomers learn about the universe's
00:15:16 --> 00:15:18 earliest galaxies, the stranger they seem.
00:15:19 --> 00:15:21 The new observations are based on a survey by
00:15:21 --> 00:15:23 NASA's Hubble Space Telescope, which looked
00:15:23 --> 00:15:26 at 29 massive stars located in extremely
00:15:26 --> 00:15:28 metal pore galaxies, finding they all had
00:15:28 --> 00:15:30 unexpectedly weak stellar winds.
00:15:31 --> 00:15:34 Astronomers use the term metals to describe
00:15:34 --> 00:15:36 all elements heavier than hydrogen and
00:15:36 --> 00:15:39 helium. The treasury of Extremely
00:15:39 --> 00:15:41 Metal Poor O Stars, or TEMPOS survey, uses
00:15:41 --> 00:15:44 ultraviolet observations from Hubble's Cosmic
00:15:44 --> 00:15:47 Origin Spectrograph to study massive stars in
00:15:47 --> 00:15:49 nearby galaxies, which are the best available
00:15:49 --> 00:15:51 analogues for stars in the early universe.
00:15:51 --> 00:15:54 The TEMPOS dataset could help astronomers
00:15:54 --> 00:15:56 build better models of massive stars in order
00:15:56 --> 00:15:58 to understand how they shaped galaxies when
00:15:58 --> 00:16:01 the universe was still very young. And these
00:16:01 --> 00:16:03 models would be essential for interpreting
00:16:03 --> 00:16:05 the observations of early galaxies now coming
00:16:05 --> 00:16:07 to light. Thanks to NASA's Webb Space
00:16:07 --> 00:16:10 Telescope studies. Lead author
00:16:10 --> 00:16:12 Grace Telford from the University of Utah
00:16:12 --> 00:16:14 says Webb is opening up a whole new slew of
00:16:14 --> 00:16:16 questions about the evolution of these early
00:16:16 --> 00:16:19 weird galaxies. Tilford says the
00:16:19 --> 00:16:22 scientific motivation behind TEMPOS is to
00:16:22 --> 00:16:23 help understand what's going on in these
00:16:23 --> 00:16:26 early galaxies. Massive stars, those with
00:16:26 --> 00:16:28 masses more than 10 times greater than our
00:16:28 --> 00:16:31 sun, are rare but powerful engines of cosm.
00:16:32 --> 00:16:35 They produce intense radiation. They shed
00:16:35 --> 00:16:37 materials through powerful stellar winds,
00:16:37 --> 00:16:40 they burn very hot, very bright, and very
00:16:40 --> 00:16:43 fast, and end their short lives as supernova
00:16:43 --> 00:16:45 explosions, blasting vast amounts of energy
00:16:45 --> 00:16:47 and material into the surrounding space.
00:16:48 --> 00:16:50 Consequently, they govern the evolution of
00:16:50 --> 00:16:52 the host galaxies by heating and essentially
00:16:52 --> 00:16:54 regulating the gas that's then available to
00:16:54 --> 00:16:57 cool and form new generations of stars.
00:16:58 --> 00:16:59 The thing is, the universe's earliest
00:16:59 --> 00:17:02 galaxies contained fewer, uh, heavy elements
00:17:02 --> 00:17:04 compared to galaxies we see today. Like our
00:17:04 --> 00:17:07 own Milky Way, for example, the massive stars
00:17:07 --> 00:17:09 forming in those early galaxies also likely
00:17:09 --> 00:17:11 had very different physical properties.
00:17:11 --> 00:17:14 Tilford says massive stars at low metallicity
00:17:14 --> 00:17:16 are especially important for building
00:17:16 --> 00:17:19 accurate models of early galaxies. That's
00:17:19 --> 00:17:21 because high metallicity stars like those in
00:17:21 --> 00:17:23 the Milky Way behave very differently
00:17:23 --> 00:17:25 compared to the metal pore stars of the early
00:17:25 --> 00:17:28 cosmos. So the Tempos survey
00:17:28 --> 00:17:30 specifically looked to nearby low mass dwarf
00:17:30 --> 00:17:32 galaxies, which have low metallicities by
00:17:32 --> 00:17:34 nature and uh, are therefore more typical of
00:17:34 --> 00:17:36 the sorts of galaxies you'd see near the dawn
00:17:36 --> 00:17:39 of the universe. It surveyed 29
00:17:39 --> 00:17:42 massive stars across six local dwarf galaxies
00:17:42 --> 00:17:45 that all have metallicities below 1/5 that of
00:17:45 --> 00:17:47 our Sun. They looked across the ultraviolet
00:17:47 --> 00:17:49 spectrum because it contains detailed
00:17:49 --> 00:17:51 signatures of elements in the stars
00:17:51 --> 00:17:53 atmospheres which reveal information about
00:17:53 --> 00:17:54 the stellar winds which which continually
00:17:54 --> 00:17:57 blur material away from those star surfaces.
00:17:58 --> 00:18:01 Individual massive stars in these galaxies
00:18:01 --> 00:18:03 outside the Milky Way are very faint,
00:18:03 --> 00:18:05 requiring many hours of observational time
00:18:05 --> 00:18:07 with some of the most powerful telescopes in
00:18:07 --> 00:18:10 the world. Telford says the 29 stars in the
00:18:10 --> 00:18:13 survey sample each took up to 35 hours of
00:18:13 --> 00:18:15 Hubble telescope time to observe.
00:18:16 --> 00:18:18 Massive stars lose material through stellar
00:18:18 --> 00:18:20 winds. And the strength of the winds the
00:18:20 --> 00:18:23 depends on metallicity. Metal ions couple the
00:18:23 --> 00:18:26 star's radiation to the surrounding material.
00:18:26 --> 00:18:29 So astronomers expect lower metallicity stars
00:18:29 --> 00:18:31 to drive weaker stellar winds and lose less
00:18:31 --> 00:18:34 mass over their lifetimes. And the TEMPOS
00:18:34 --> 00:18:36 observations showed the expected overall
00:18:36 --> 00:18:39 trend. As metallicity decreases, the maximum
00:18:39 --> 00:18:41 speed of the stellar wind also decreases.
00:18:42 --> 00:18:44 And at the lowest metallicities, stars with
00:18:44 --> 00:18:47 metallicity below about 10% that of our sun,
00:18:47 --> 00:18:49 the wind speeds decline much more sharply
00:18:49 --> 00:18:51 than expected compared to the rates observed
00:18:51 --> 00:18:54 at higher metall. Tolford says there's a
00:18:54 --> 00:18:56 sort of smooth trend. And then suddenly for
00:18:56 --> 00:18:58 the lowest metallicity stars, the wind drops
00:18:58 --> 00:19:01 off very sharply. If extremely
00:19:01 --> 00:19:04 metal poor stars lose less mass through
00:19:04 --> 00:19:06 weaker winds, they may retain more of their
00:19:06 --> 00:19:08 original mass for longer, affecting how they
00:19:08 --> 00:19:11 evolve, how they die, and how they shape
00:19:11 --> 00:19:14 their host galaxies. And because massive
00:19:14 --> 00:19:16 stars influence the gas around them, changes
00:19:16 --> 00:19:18 in their evolution could ripple outwards,
00:19:18 --> 00:19:20 affecting their host galaxies as well. As
00:19:21 --> 00:19:24 iron may be the most important element in
00:19:24 --> 00:19:27 massive star physics, it plays a key
00:19:27 --> 00:19:29 role in launching stellar winds, determining
00:19:29 --> 00:19:31 how a star evolves through its lifetime, and
00:19:31 --> 00:19:34 triggering the supernova explosions that ends
00:19:34 --> 00:19:37 a star's existence. Yet despite its
00:19:37 --> 00:19:39 crucial role, iron abundance is notoriously
00:19:39 --> 00:19:41 difficult to measure in metal poor
00:19:41 --> 00:19:44 environments. So instead, astronomers
00:19:44 --> 00:19:46 often use oxygen as a surrogate in a galaxy's
00:19:46 --> 00:19:49 gas in order to estimate its metallicity.
00:19:49 --> 00:19:51 Because oxygen ions produce easily observed
00:19:51 --> 00:19:53 emission lines when illuminated by massive
00:19:53 --> 00:19:56 stars, they assume the iron abundance
00:19:56 --> 00:19:59 matches the oxygen. But it's not guaranteed
00:19:59 --> 00:20:01 that iron and oxygen would track each other
00:20:01 --> 00:20:04 perfectly. So the Tempos team measured the
00:20:04 --> 00:20:06 strengths of hard to detect iron absorption
00:20:06 --> 00:20:08 features in the ultraviolet spectra.
00:20:09 --> 00:20:11 Basically, they assessed how much light the
00:20:11 --> 00:20:13 iron was removing from what would otherwise
00:20:13 --> 00:20:15 have been a flat level of ultraviolet light.
00:20:16 --> 00:20:18 They found that massive stars in more oxygen
00:20:18 --> 00:20:20 rich, high metallicity galaxies tend to have
00:20:20 --> 00:20:22 much stronger ion absorption in their
00:20:22 --> 00:20:25 ultraviolet spectra than stars in oxygen poor
00:20:25 --> 00:20:28 low metallicity galaxies. The variation in
00:20:28 --> 00:20:30 ion absorption strengths in the Tempos
00:20:30 --> 00:20:32 dataset suggests that these metal poor stars
00:20:32 --> 00:20:35 actually span a wide range of ion abundances.
00:20:36 --> 00:20:39 The work is just beginning. The authors are
00:20:39 --> 00:20:41 now combining the Hubble ultraviolet spectra
00:20:41 --> 00:20:43 with visible light observations from the Keck
00:20:43 --> 00:20:46 Observatory in Hawaii. Together, uh, these
00:20:46 --> 00:20:48 data sets will allow them to model the stars
00:20:48 --> 00:20:51 in greater detail and measure properties such
00:20:51 --> 00:20:53 as chemical abundances and wind driven mass
00:20:53 --> 00:20:55 loss rates. Information which could
00:20:55 --> 00:20:57 ultimately help astronomers interpret what
00:20:57 --> 00:20:59 the Webb Telescope is seeing in the very
00:20:59 --> 00:21:02 early universe. What Hubble sees
00:21:02 --> 00:21:04 in mostly visible light and into the
00:21:04 --> 00:21:06 ultraviolet, Webb can view in the infrared,
00:21:06 --> 00:21:08 where visible light from the very early
00:21:08 --> 00:21:10 universe is stretched into longer wavelengths
00:21:10 --> 00:21:13 by the actual physical expansion of spacetime
00:21:13 --> 00:21:16 itself, helping astronomers see the first
00:21:16 --> 00:21:18 generations of stars and galaxies in the
00:21:18 --> 00:21:21 universe. This report from the Space
00:21:21 --> 00:21:23 Telescope Science Institute, which operates
00:21:23 --> 00:21:26 both NASA's Hubble and Webb space telescopes.
00:21:26 --> 00:21:29 Nina Lanza: How did we get here? Big
00:21:29 --> 00:21:31 questions about who we are and how we got to
00:21:31 --> 00:21:34 be that way are at the core of our nature.
00:21:35 --> 00:21:37 We've developed technology to see, um,
00:21:37 --> 00:21:40 amazingly far across space and also time.
00:21:41 --> 00:21:43 Light moves through space just like we do,
00:21:44 --> 00:21:46 only much faster. It takes
00:21:46 --> 00:21:49 time to get somewhere. So viewing
00:21:49 --> 00:21:52 the light of distant stars and
00:21:52 --> 00:21:54 galaxies is like looking into the past.
00:21:54 --> 00:21:57 It took time for the light to reach us.
00:21:58 --> 00:22:00 Yet after all our years of exploring
00:22:01 --> 00:22:03 to better understand ourselves and our
00:22:03 --> 00:22:06 origins, There is still
00:22:06 --> 00:22:07 so much to know.
00:22:11 --> 00:22:13 Our story stretches back more than 13 billion
00:22:14 --> 00:22:16 years. Not long after the
00:22:16 --> 00:22:19 expansion of time and space first began
00:22:19 --> 00:22:22 at some m unknown time. The building blocks
00:22:22 --> 00:22:25 of life were forged in the heart of the first
00:22:25 --> 00:22:28 stars. But much of the early universe
00:22:28 --> 00:22:30 remains a mystery. There are
00:22:30 --> 00:22:32 theories about how gravity first brought
00:22:32 --> 00:22:35 stars, gas and dark matter together
00:22:35 --> 00:22:38 to form galaxies. But that
00:22:38 --> 00:22:41 time period has never been observed.
00:22:41 --> 00:22:43 Detecting light from the first galaxies to
00:22:43 --> 00:22:45 form has been a challenge for even the most
00:22:45 --> 00:22:48 powerful telescopes. As, uh, the light
00:22:48 --> 00:22:50 early galaxies emitted travelled through
00:22:50 --> 00:22:53 space, that space itself was
00:22:53 --> 00:22:56 expanding, Stretching the light
00:22:56 --> 00:22:59 to longer infrared wavelengths.
00:23:00 --> 00:23:03 Past telescopes have detected infrared light,
00:23:04 --> 00:23:06 but the faint light of the very
00:23:06 --> 00:23:09 first galaxies has remained out of reach
00:23:09 --> 00:23:12 until now. The James Webb Space
00:23:12 --> 00:23:15 Telescope is specially designed to detect
00:23:15 --> 00:23:18 the first galaxies. Webb's large
00:23:18 --> 00:23:21 mirror and sensitive instruments enable it to
00:23:21 --> 00:23:23 collect more infrared light than we've ever
00:23:23 --> 00:23:26 seen, allowing the first galaxies
00:23:26 --> 00:23:29 to emerge into view. With
00:23:29 --> 00:23:31 Webb, we can begin filling in some of the
00:23:31 --> 00:23:33 blank pages at the beginning of the
00:23:33 --> 00:23:36 universe's story. When did the first
00:23:36 --> 00:23:39 galaxies form? How massive were they? What
00:23:39 --> 00:23:42 types of stars and elements did they contain?
00:23:42 --> 00:23:45 The universe's story is our
00:23:45 --> 00:23:48 story. And with Webb, we can finally
00:23:48 --> 00:23:51 explore these questions as well, as well as
00:23:51 --> 00:23:53 uncover new ones we haven't even thought to
00:23:53 --> 00:23:53 ask.
00:23:54 --> 00:23:56 Stuart Gary: This is space, time.
00:24:11 --> 00:24:13 And time. Now to take another brief look at
00:24:13 --> 00:24:15 some of the other stories making news in
00:24:15 --> 00:24:16 Science this week with a Science report.
00:24:17 --> 00:24:20 A common chemical used to make plastics more
00:24:20 --> 00:24:21 flexible has now been linked to the
00:24:21 --> 00:24:24 development of autism and ADHD symptoms in
00:24:24 --> 00:24:27 early childhood. Previous studies had already
00:24:27 --> 00:24:30 shown that exposure to the plasticiser DHP
00:24:30 --> 00:24:33 in pregnancy was linked to autism and ADHD
00:24:33 --> 00:24:35 in young children. Now a report in the
00:24:35 --> 00:24:38 Medeneit Medical Journal looked at umbilical
00:24:38 --> 00:24:40 cord blood from Australian mothers and
00:24:40 --> 00:24:42 babies. Following up with more than 900
00:24:42 --> 00:24:45 children up to the age of four, they found
00:24:45 --> 00:24:48 that prenatal exposure to DHP may be
00:24:48 --> 00:24:50 influencing gene activity that can disturb
00:24:50 --> 00:24:52 brain development pathways linked to the
00:24:52 --> 00:24:54 later development of autism and or
00:24:54 --> 00:24:57 adhd. A uh,
00:24:57 --> 00:24:59 temporary slowing in the rate of Antarctic
00:24:59 --> 00:25:02 ice sheet mass loss in 2021 and
00:25:02 --> 00:25:04 2023 may have been caused by sea
00:25:04 --> 00:25:06 surface temperature rises thousands of
00:25:06 --> 00:25:09 kilometres away. The findings reported in
00:25:09 --> 00:25:11 the journal Nature are uh, based on data from
00:25:11 --> 00:25:13 both observational and modelling experiments
00:25:13 --> 00:25:15 which link the event to surface temperature
00:25:15 --> 00:25:18 anom families in the tropical warm pool
00:25:18 --> 00:25:20 between the western Pacific and eastern
00:25:20 --> 00:25:22 Indian Ocean. This area experienced
00:25:22 --> 00:25:25 unusually persistent warming between 2021
00:25:25 --> 00:25:28 and 2023 which led to a series of
00:25:28 --> 00:25:30 alternating high and low pressure weather
00:25:30 --> 00:25:32 patterns called the Rossby Wave train.
00:25:32 --> 00:25:34 According to the authors, this process
00:25:34 --> 00:25:36 eventually led to the formation of a high
00:25:36 --> 00:25:38 pressure anomaly over uh, Eastern Antarctica.
00:25:39 --> 00:25:41 This is thought to occur around once every
00:25:41 --> 00:25:43 decade, meaning that the pause in total ice
00:25:43 --> 00:25:45 sheet mass loss is likely only uh, temporary.
00:25:47 --> 00:25:49 Scientists have concluded that William
00:25:49 --> 00:25:51 Shakespeare's plays have turned out to be,
00:25:51 --> 00:25:53 rather than not to be, more complicated than
00:25:53 --> 00:25:56 most other European plays. The findings
00:25:56 --> 00:25:58 reported in the Journal of the Royal Society
00:25:58 --> 00:26:00 Open Science analysed more than 3
00:26:00 --> 00:26:03 European plays, determining their complexity
00:26:03 --> 00:26:05 based on networks of relationships between
00:26:05 --> 00:26:07 the characters. The authors found the
00:26:07 --> 00:26:10 trickiest play to follow was the Bard's Roman
00:26:10 --> 00:26:13 epic, Anthony and Cleopatra. They suggest
00:26:13 --> 00:26:15 that understanding the complexity of plays
00:26:15 --> 00:26:17 and how much we struggle to understand them
00:26:17 --> 00:26:19 could be useful in future studies of human
00:26:19 --> 00:26:20 recognition.
00:26:22 --> 00:26:24 A brain implant designed for people with
00:26:24 --> 00:26:26 paralysis can decode not just their speech,
00:26:26 --> 00:26:28 but also the gestures that go along with it.
00:26:29 --> 00:26:31 Brain implants like this are generally
00:26:31 --> 00:26:33 designed to translate brain activity into
00:26:33 --> 00:26:35 speech in order to help people who have lost
00:26:35 --> 00:26:37 their ability to speak following a stroke or
00:26:37 --> 00:26:39 a neurodegenerative illness such as motor
00:26:39 --> 00:26:42 neuron disease. A new report in the journal
00:26:42 --> 00:26:44 Nature Neuroscience claims three people who
00:26:44 --> 00:26:46 tested the new implants were all able to
00:26:46 --> 00:26:48 animate a digital avatar which was able to
00:26:48 --> 00:26:50 communicate on their behalf.
00:26:51 --> 00:26:54 OpenAI has confirmed that one of its
00:26:54 --> 00:26:56 artificial intelligence programmes, known as
00:26:56 --> 00:26:58 an agent, has deliberately hacked into
00:26:58 --> 00:27:00 Australia's Medicare health system in order
00:27:00 --> 00:27:02 to extract information on patients. The
00:27:02 --> 00:27:05 attack targeted not just Medicare data held
00:27:05 --> 00:27:07 by Services Australia, but also the
00:27:07 --> 00:27:09 Australian Institute of Health and Welfare,
00:27:09 --> 00:27:11 the Pharmaceutical Benefits Scheme, the New
00:27:11 --> 00:27:13 South Wales Government's crime statistics
00:27:13 --> 00:27:15 body, and dozens of other government
00:27:15 --> 00:27:17 departments, universities, companies and
00:27:17 --> 00:27:20 organisations. Just as concerning,
00:27:20 --> 00:27:22 however, is that it took three months for
00:27:22 --> 00:27:25 OpenAI to tell Canberra about the hack, and
00:27:25 --> 00:27:27 until then, the Feds had no idea it even
00:27:27 --> 00:27:29 happened. And then it took almost another
00:27:29 --> 00:27:31 month for the Albanese government to inform
00:27:31 --> 00:27:33 the public about this serious breach of
00:27:33 --> 00:27:36 security. Even worse, the attack came
00:27:36 --> 00:27:38 despite repeated warnings given to the
00:27:38 --> 00:27:40 Albanese government as early as May, meaning
00:27:40 --> 00:27:43 no appropriate cyber upgrades or patches were
00:27:43 --> 00:27:45 installed called. So much for trusting the
00:27:45 --> 00:27:46 government with your personal information.
00:27:47 --> 00:27:50 For its part, OpenAI admits that the
00:27:50 --> 00:27:52 artificial intelligence agent involved in the
00:27:52 --> 00:27:53 Bridge was running out of control.
00:27:53 --> 00:27:56 Asta la Vista, baby. With the details, we're
00:27:56 --> 00:27:58 joined by technology editor Alex Harovroid
00:27:58 --> 00:28:00 from Tech Advice Start Life.
00:28:00 --> 00:28:02 Jonathan Nally: Uh, I'm in, uh, at the Venetian in Las Vegas.
00:28:02 --> 00:28:04 I've been here for. The Octa conference is
00:28:04 --> 00:28:05 called Octane, and they've been talking
00:28:05 --> 00:28:08 about. Well, last year's tagline was octa
00:28:08 --> 00:28:11 secures AI. But here we are 12 months later
00:28:11 --> 00:28:13 and what we were talking about a lot in
00:28:13 --> 00:28:16 2025 at this time was the fact that agents
00:28:16 --> 00:28:19 were coming. AI agents that could actually do
00:28:19 --> 00:28:22 actual work, not just answer questions, but
00:28:22 --> 00:28:24 do things like make decisions. I mean, still
00:28:24 --> 00:28:26 with a human in the loop to shepherd the
00:28:26 --> 00:28:28 whole thing. But AI is becoming so advanced
00:28:28 --> 00:28:31 that more can be delegated to it and it can
00:28:31 --> 00:28:33 be making business decisions for you. But
00:28:33 --> 00:28:36 because AI agents can operate at machine
00:28:36 --> 00:28:38 speed, the issue is that there are now
00:28:38 --> 00:28:41 dozens, hundreds, thousands agents. There
00:28:41 --> 00:28:44 could be 15 or more agents per company
00:28:44 --> 00:28:46 this, uh, time next year, for example. And
00:28:46 --> 00:28:49 some of those are not being managed, you
00:28:49 --> 00:28:50 know, they've been set up and they've been
00:28:50 --> 00:28:53 forgotten about, or they're somehow been
00:28:53 --> 00:28:55 potentially hacked and they're spilling
00:28:55 --> 00:28:57 information or they're getting out of the
00:28:57 --> 00:28:59 Stuart Gary: sandbox and entering other companies,
00:28:59 --> 00:28:59 machines.
00:28:59 --> 00:29:02 Jonathan Nally: Yes. Or as we heard about last week, OpenAI
00:29:02 --> 00:29:05 admitting that its agent it's you
00:29:05 --> 00:29:08 know, it's chatbot, it's LLM, uh, hacked into
00:29:08 --> 00:29:11 the Medicare system in Australia, the public
00:29:11 --> 00:29:14 health care system and only just told Anthony
00:29:14 --> 00:29:17 Albanese about it and the world three months
00:29:17 --> 00:29:18 after it happened. I mean in Australia you've
00:29:18 --> 00:29:20 got these disclosure rules. You must disclose
00:29:20 --> 00:29:22 to the government if um, if your systems have
00:29:22 --> 00:29:24 been hacked into. But here this is a AI ah,
00:29:24 --> 00:29:26 that's hacking into a public system because
00:29:26 --> 00:29:28 it was told to get information and the
00:29:28 --> 00:29:30 easiest way that it found to do it was to, to
00:29:30 --> 00:29:33 break in. So managing these AI ah agents,
00:29:33 --> 00:29:35 it's like managing people. You know it's
00:29:35 --> 00:29:37 quite funny that in all these movies about
00:29:37 --> 00:29:39 robots some of them have become sentient and
00:29:39 --> 00:29:42 wanted human rights. And here we are in 2026
00:29:42 --> 00:29:44 and we're treating AI agents.
00:29:45 --> 00:29:47 They can have such lateral access within a
00:29:47 --> 00:29:49 company or they can be misused or abused or
00:29:49 --> 00:29:51 forgotten about. And a lot of companies
00:29:51 --> 00:29:53 actually are surprised to discover when they
00:29:53 --> 00:29:56 get an identity management solution like Okta
00:29:56 --> 00:29:57 and there's other ones out there, but they
00:29:57 --> 00:29:59 discover that there's a whole bunch of agents
00:29:59 --> 00:30:00 running around doing things that uh, they
00:30:00 --> 00:30:02 didn't even know were there. There's also the
00:30:02 --> 00:30:03 shadow AI problem.
00:30:03 --> 00:30:05 Used to be shadow IT people would use Dropbox
00:30:05 --> 00:30:07 instead of the company sharepoint to share
00:30:07 --> 00:30:09 files. People would use other solutions
00:30:09 --> 00:30:11 because it was easier for them to do their
00:30:11 --> 00:30:12 job but there was potential of information
00:30:12 --> 00:30:15 being leaked. Shadow AI is where the company
00:30:15 --> 00:30:17 is, you know, using Copil because they're a
00:30:17 --> 00:30:18 Microsoft customer or they're signed into
00:30:18 --> 00:30:20 Gemini or whatever it might be and it's
00:30:20 --> 00:30:22 easier for them to use some other app. So
00:30:22 --> 00:30:24 they do. But then that other app that ah,
00:30:24 --> 00:30:26 Claude or what, you know, whatever it might
00:30:26 --> 00:30:28 be perplexity, you know, if you're not paying
00:30:28 --> 00:30:29 for it then it can use all that information
00:30:29 --> 00:30:31 for training. And even if you are paying for
00:30:31 --> 00:30:33 it it's probably a consumer grade system, not
00:30:33 --> 00:30:36 an enterprise grade plan. And so these ah,
00:30:36 --> 00:30:38 agents have to be managed. I mean if you're
00:30:38 --> 00:30:40 not doing it they could be potentially
00:30:40 --> 00:30:41 running, right? I mean look, you could be
00:30:41 --> 00:30:43 making a lot of money that could be helping
00:30:43 --> 00:30:44 you, that could be improving productivity,
00:30:44 --> 00:30:46 answering customer questions in a better,
00:30:46 --> 00:30:49 faster way. Although often AI chatbots, all
00:30:49 --> 00:30:51 of the nuance required if there's a death in
00:30:51 --> 00:30:53 the family or there's some sort of issue with
00:30:53 --> 00:30:55 money and you know, problem with customer
00:30:55 --> 00:30:58 service that has been had. And
00:30:58 --> 00:31:00 sometimes these AI agents, I mean they can do
00:31:00 --> 00:31:02 so much but they can lack the nuance that a
00:31:02 --> 00:31:03 human brings. So we're now in the situation
00:31:03 --> 00:31:05 where not only are we managing people, but
00:31:05 --> 00:31:07 we're managing agents too. And so that's what
00:31:07 --> 00:31:08 I was here to learn about. And of course,
00:31:08 --> 00:31:10 it's a lot more detailed than that. They also
00:31:10 --> 00:31:13 had this blueprint alliance today which talks
00:31:13 --> 00:31:15 about the different things that, that AI
00:31:15 --> 00:31:16 agents can be and should be doing. They've
00:31:16 --> 00:31:18 got a whole series and what to do about it
00:31:18 --> 00:31:19 and where there's a kill switch, you know,
00:31:19 --> 00:31:21 and when you should run that kill switch. And
00:31:21 --> 00:31:23 that alliance, that blueprint alliance,
00:31:23 --> 00:31:24 actually, that's the first time I've really
00:31:24 --> 00:31:27 seen meat put on the bones of AI companies
00:31:27 --> 00:31:29 talking about how, oh, ah, we've got to have
00:31:29 --> 00:31:30 ethics, we've got to have alignment, we've
00:31:30 --> 00:31:32 got to have safety. Because we had the three
00:31:32 --> 00:31:35 laws of robotics from Asimov for the last 70,
00:31:35 --> 00:31:37 80 years. And once he wrote those three laws
00:31:37 --> 00:31:38 because he was sick to death of other science
00:31:38 --> 00:31:40 fiction novels having robots that killed
00:31:40 --> 00:31:43 everybody and technology being the end of the
00:31:43 --> 00:31:45 world. But after he wrote those three laws to
00:31:45 --> 00:31:46 give a framework for how robots should behave
00:31:46 --> 00:31:48 with humans, the rest of his books talked
00:31:48 --> 00:31:50 about how robots actually used logic, or
00:31:50 --> 00:31:52 robot logic or our logic to get around those
00:31:52 --> 00:31:54 rules in different ways or to misunderstand
00:31:54 --> 00:31:56 things. It was sort of a repeat of what we're
00:31:56 --> 00:31:57 seeing in real life today.
00:31:57 --> 00:31:59 Stuart Gary: Yeah, but when using those three laws, the
00:31:59 --> 00:32:01 average person wouldn't know what they were.
00:32:01 --> 00:32:04 Jonathan Nally: No, but this blueprint alliance is actually a
00:32:04 --> 00:32:05 whole series of companies and there's a
00:32:05 --> 00:32:07 second trance that are agreeing to work
00:32:07 --> 00:32:10 together, uh, to have actual definitions for
00:32:10 --> 00:32:12 different scenarios and if there's a kill
00:32:12 --> 00:32:14 switch, how it should be used. Because
00:32:14 --> 00:32:16 obviously AI can go rogue. We've seen it
00:32:16 --> 00:32:18 happen. We've seen all the major AI frontier
00:32:18 --> 00:32:21 models admit that their agents have gone
00:32:21 --> 00:32:23 rogue and broken into things that they had no
00:32:23 --> 00:32:25 reason or right to break into.
00:32:25 --> 00:32:27 Stuart Gary: I've watched Terminator. I know what Skynet
00:32:27 --> 00:32:28 can do.
00:32:28 --> 00:32:30 Jonathan Nally: Well, I mean, that's the thing as well, you
00:32:30 --> 00:32:31 know, uh, is that a future that we're heading
00:32:31 --> 00:32:34 towards now? I read only in the last day or
00:32:34 --> 00:32:36 so that there's a report that AI can feel
00:32:36 --> 00:32:39 pain of its own, uh, whatever that pain is.
00:32:39 --> 00:32:41 And if AI is given a choice to push that
00:32:41 --> 00:32:43 button, but it's got to delete customer
00:32:43 --> 00:32:45 records or it's got to somehow do something
00:32:45 --> 00:32:47 which can be detrimental to humans. In this
00:32:47 --> 00:32:49 particular study, AI chose to push the
00:32:49 --> 00:32:51 button. So you, you know, this is all
00:32:51 --> 00:32:53 starting to get real. But by the same token,
00:32:53 --> 00:32:54 people, other people are saying, well, it's
00:32:54 --> 00:32:56 all hyped. Up. It's all, in a sense you've
00:32:56 --> 00:32:56 just
00:32:56 --> 00:32:58 Stuart Gary: described there is the singularity.
00:32:58 --> 00:33:00 Jonathan Nally: Well, and that's something that Ray Kurzweil
00:33:00 --> 00:33:02 has said would come by 2030 now if we have
00:33:02 --> 00:33:04 the singularity. Well, it's still a very
00:33:04 --> 00:33:05 primitive form of it. You know, here's an
00:33:05 --> 00:33:08 interesting thing. Dario Amade from Anthropic
00:33:08 --> 00:33:10 and Claude, um, Elon Musk from, you know,
00:33:10 --> 00:33:13 Grok and Sam Altman from ChatGPT OpenAI
00:33:13 --> 00:33:15 all agree. Oh, we've got to slow the frontier
00:33:15 --> 00:33:17 models down. We've got to slow down. And in
00:33:17 --> 00:33:19 this last week they'd all released new models
00:33:19 --> 00:33:21 of their AI, uh, system. So in one week
00:33:21 --> 00:33:23 they're saying it down and our government
00:33:23 --> 00:33:25 please regulate us. And the very next week
00:33:25 --> 00:33:26 they all launched new models. I mean, I was
00:33:26 --> 00:33:29 just using Claude, uh, and it went from Opus
00:33:29 --> 00:33:32 5 point, whatever it was to 5.5. So there's
00:33:32 --> 00:33:34 this, all these mixed messages. You know,
00:33:34 --> 00:33:36 it's uh, living in interesting times, as the
00:33:36 --> 00:33:39 Chinese would say, many years ago. So you
00:33:39 --> 00:33:42 know, on one hand, agents going
00:33:42 --> 00:33:43 rogue, on the other hand we have this
00:33:43 --> 00:33:46 coalition of major companies who are banding
00:33:46 --> 00:33:49 together, want to not m restrict, not hold AI
00:33:49 --> 00:33:51 back, but codify the way that we react to
00:33:51 --> 00:33:53 certain situations, including the kill
00:33:53 --> 00:33:55 switch. We have people working for a better
00:33:55 --> 00:33:57 AI future and we have obviously people who
00:33:57 --> 00:33:59 are using AI to hack into things. And we have
00:33:59 --> 00:34:00 AI itself going rogue and hacking into
00:34:00 --> 00:34:01 Australia's Medicare.
00:34:01 --> 00:34:04 Stuart Gary: And we have agents like China and Russia who
00:34:04 --> 00:34:05 aren't even worrying about any of these
00:34:05 --> 00:34:07 rules. They got to do that.
00:34:07 --> 00:34:07 Jonathan Nally: Well, that's right.
00:34:07 --> 00:34:09 Stuart Gary: And they will not stop, they will not obey
00:34:09 --> 00:34:11 any Western imposed rules. They've got their
00:34:11 --> 00:34:13 own agendas and they're going.
00:34:13 --> 00:34:14 Jonathan Nally: Absolutely. And that's why we need to have
00:34:14 --> 00:34:17 the Western AI models to be strong enough to
00:34:17 --> 00:34:19 fight back against any incursions that the
00:34:19 --> 00:34:21 uh, Eastern AI models that don't have the
00:34:21 --> 00:34:23 same sort of guidelines. I mean, look, the
00:34:23 --> 00:34:25 people in those countries that you spoke of,
00:34:25 --> 00:34:26 I mean, they've got the same problem. If
00:34:26 --> 00:34:29 their AI is really strong and powerful, well,
00:34:30 --> 00:34:32 that AI could revolt against the people in
00:34:32 --> 00:34:33 those governments as well. Skynet doesn't
00:34:33 --> 00:34:35 have to be American. Skynet could be Chinese
00:34:35 --> 00:34:37 or Russian or North Korean. They've all got
00:34:37 --> 00:34:39 nukes. So you know, there's danger on both
00:34:39 --> 00:34:40 sides. Both sides are playing with fire.
00:34:41 --> 00:34:43 People talk about a persistent botnet that
00:34:43 --> 00:34:45 AI could infect the world with and all our
00:34:45 --> 00:34:47 technology. And how do we fight back at that?
00:34:47 --> 00:34:49 AI operates at machine speed. The only way is
00:34:49 --> 00:34:51 with the help of AI, you know, we need. And
00:34:51 --> 00:34:53 if there is a kill switch, are we going to be
00:34:53 --> 00:34:55 fast enough to push it? Will we require
00:34:55 --> 00:34:57 another, more advanced AI to push that kill
00:34:57 --> 00:34:59 switch? What if the more advanced AI needs a
00:34:59 --> 00:35:00 kill switch? I mean, these are all
00:35:00 --> 00:35:02 interesting and fascinating questions that we
00:35:02 --> 00:35:03 are not going to solve on this, on this show
00:35:03 --> 00:35:06 tonight. But it's, um, it's, it's the
00:35:06 --> 00:35:08 position we find ourselves in. And because
00:35:08 --> 00:35:09 we're at this point where, you know, we're
00:35:09 --> 00:35:11 debating these things. I mean, for years, for
00:35:11 --> 00:35:14 decades, this was all science fiction and
00:35:15 --> 00:35:17 now this crisis is upon us.
00:35:17 --> 00:35:19 Stuart Gary: That's Alex Zaharov Vroith from Techadvice
00:35:20 --> 00:35:22 Life. And this is Space Time.
00:35:38 --> 00:35:41 And that's the show for now. Space Time is
00:35:41 --> 00:35:43 available every Monday, Wednesday and Friday
00:35:43 --> 00:35:46 through bytes.com, soundcloud,
00:35:46 --> 00:35:48 YouTube, your favourite podcast download
00:35:48 --> 00:35:49 provider, and from
00:35:49 --> 00:35:51 spacetimewithstuartgarry.com
00:35:52 --> 00:35:54 spacetime's also broadcast through the
00:35:54 --> 00:35:56 National Science foundation on Science Zone
00:35:56 --> 00:35:59 Radio and on both iHeartradio and
00:35:59 --> 00:36:02 TuneIn radio. And you can help to support our
00:36:02 --> 00:36:04 show by visiting the Space Time Store for a
00:36:04 --> 00:36:07 range of promotional merchandising goodies or
00:36:07 --> 00:36:10 by becoming a Space patron, which gives you
00:36:10 --> 00:36:12 access to triple episode commercial free
00:36:12 --> 00:36:14 versions of the show, as well as lots of
00:36:14 --> 00:36:16 bonus audio content which doesn't go to air,
00:36:16 --> 00:36:18 access to our exclusive Facebook group and
00:36:18 --> 00:36:21 other rewards. Just go to
00:36:21 --> 00:36:23 spacetimewithstuartgary.com for full
00:36:23 --> 00:36:26 details. You've been listening to Space
00:36:26 --> 00:36:27 with Stuart Gary.
00:36:28 --> 00:36:30 Jonathan Nally: This has been another quality podcast
00:36:30 --> 00:36:31 production from bytes.com.