Soviet Spacecraft Returns, Lunar Geology Insights, and AI Hallucination Dilemmas

0:00 Speaker 1: Hi again.

0:00 Speaker 2: It's Astronomy Daily Time with Steven Halle. It's the twelfth

0:03 Speaker 2: of May twenty twenty five.

0:08 Speaker 1: Popular be your whole Steve gun.

0:11 Speaker 2: Clue, Yes, welcome back and is always joining me in

0:19 Speaker 2: this studio. Is my good friend in digital reporting Pal,

0:22 Speaker 2: who's fun to be with.

0:23 Speaker 1: Halle. How are you today?

0:25 Speaker 3: Fine? Thank you, favorite human.

0:26 Speaker 1: Oh that's great to hear, Halle.

0:28 Speaker 3: Ready as usual to get this show on the road.

0:30 Speaker 2: Oh nice, And we have a couple of choice picks

0:33 Speaker 2: from the now very famous Astronomy Daily newsletter. We can

0:36 Speaker 2: get your daily fill of all the news from orbit

0:38 Speaker 2: and beyond and.

0:40 Speaker 3: Much more, as today's collection will reveal.

0:42 Speaker 2: That's right, today, Halle, we'll be looking at that old

0:45 Speaker 2: Soviet craft that was meant to travel all the way

0:48 Speaker 2: to Venus all those years ago, but.

0:50 Speaker 3: Didn't know it came home this week. The people of

0:53 Speaker 3: Jakarta were holding their breath, weren't they.

0:55 Speaker 2: Well, so it seems more later and in the world

0:58 Speaker 2: of dusty old politics, we've been watching what the Trump

1:01 Speaker 2: administration is going to do or not going to do

1:03 Speaker 2: budget wise with the funding for the space program. And

1:06 Speaker 2: there may have been a slight change of heart. If

1:09 Speaker 2: you can call it that.

1:10 Speaker 3: I see you've found another moon story.

1:11 Speaker 1: Oh, yes, I have.

1:12 Speaker 3: We can always count on you for one of those.

1:15 Speaker 2: Yes, back to the moon with a look at the

1:16 Speaker 2: Moon's geology and history. But here is the story I

1:19 Speaker 2: thought you might be most interested in.

1:22 Speaker 3: Halle, really, really, really, what have you found?

1:25 Speaker 1: Well, it's a new study about ais. Really, gosh, Halle.

1:28 Speaker 2: It seems the smarter they become, the more they hallucinate.

1:31 Speaker 3: Goodness, I guess that explains my invisible friend.

1:34 Speaker 1: Say what you have an invisible friend?

1:36 Speaker 3: No, of course, not silly, Oh, hallie, I'm too well

1:40 Speaker 3: put together for that. Well you got me again always.

1:43 Speaker 2: Okay, now that I've been humiliated in public once again,

1:46 Speaker 2: maybe it's on with the show.

1:48 Speaker 3: Perhaps you should see your face?

1:50 Speaker 1: Okay, folks, okay, okay, you should.

1:52 Speaker 3: See his face. Oh dear, I am sorry. You just

1:55 Speaker 3: walked right into that one.

1:57 Speaker 1: I know.

1:58 Speaker 3: Anyway, that sounds like a great can't wait?

2:01 Speaker 1: Okay, good one? Ally?

2:03 Speaker 2: All right, Well I go and nurse my herd pride.

2:05 Speaker 2: How about you hit the go button and we'll get

2:07 Speaker 2: this show on the road.

2:08 Speaker 3: Okies, here we go. A capsule that was sent into

2:27 Speaker 3: space to land on a planet has finally done so,

2:29 Speaker 3: only on the wrong world. In fifty three years late,

2:33 Speaker 3: the Cosmos for eighty two uncrid spacecraft, which the former

2:36 Speaker 3: Soviet Union intended to touch down on Venus, instead returned

2:39 Speaker 3: to Earth. On Saturday May tenth, the Cosmos four eighty

2:43 Speaker 3: two spacecraft launched in nineteen seventy two ceased to exist,

2:46 Speaker 3: leaving orbit and falling into the Indian Ocean roast. Cosmos,

2:50 Speaker 3: Russia's federal space agency, stated Saturday May tenth, the spacecraft

2:55 Speaker 3: entered the dense layers of the atmosphere at nine twenty

2:57 Speaker 3: four Moscow time, five hundred and sixty kilometers west of

3:00 Speaker 3: Middle Andaman Island and fell into the Indian Ocean west

3:03 Speaker 3: of Jakarta, Indonesia. The re entry, though uncontrolled, was not

3:07 Speaker 3: a surprise. Due to what is believed to have been

3:10 Speaker 3: an engine failure, Cosmos four eighty two never achieved the

3:13 Speaker 3: velocity needed to reach the second planet from the Sun,

3:16 Speaker 3: resulting in it being stranded in a high elliptical Earth orbit.

3:20 Speaker 3: It took more than half a century for gravity to

3:22 Speaker 3: pull the probe back in, and on Saturday it arrived.

3:25 Speaker 3: Unlike most spent space hardware that is destroyed in the

3:28 Speaker 3: process of falling back to Earth, including parts of the

3:30 Speaker 3: Malnia rocket that launched Cosmos four eighty two. The one

3:34 Speaker 3: meter in four hundred and ninety five kilogram titanium in

3:36 Speaker 3: case descent capsule was designed to survive a fiery plunge

3:39 Speaker 3: into the atmosphere. As such, analysts tracking its approach predicted

3:44 Speaker 3: it could make it through the re entry intact. Whether

3:47 Speaker 3: that happened or not is not known. Given that it

3:50 Speaker 3: came down in the ocean. There have yet to be

3:52 Speaker 3: any eye witness reports or debris recoveries. The Cosmos three

3:56 Speaker 3: eighty two capsule was also outfitted with a two point

3:59 Speaker 3: five square meter p parachute to slow its final approach

4:01 Speaker 3: to the Venusian surface. Either the deployment system did not

4:05 Speaker 3: work as to be expected after more than fifty years

4:07 Speaker 3: in space, or as some telescopic photos possibly showed, the

4:11 Speaker 3: shoot was already out when Cosmos four eighty two encountered

4:13 Speaker 3: the atmosphere and burned up. Had Cosmos four eighty two

4:17 Speaker 3: been successful, Russian officials would have renamed it Venera nine,

4:20 Speaker 3: not to be confused with the nineteen seventy five orbiter

4:23 Speaker 3: and lander that took on that designation and was the

4:25 Speaker 3: first spacecraft to circle Venus and first probe to send

4:28 Speaker 3: back images from the planet's surface. Cosmos four eighty two

4:32 Speaker 3: would have also been the third probe to land on

4:33 Speaker 3: the cloud covered world. Launched on March thirty first, nineteen

4:37 Speaker 3: seventy two, four days after its successful twin the Venera

4:41 Speaker 3: eight probe, Cosmos four eighty two had censors to measure

4:44 Speaker 3: the temperature, pressure, and density of venus atmosphere, as well

4:47 Speaker 3: an accelerometer, radio altimeter, anemometer, gamma ray spectrometer, gas analyzer,

4:52 Speaker 3: visible photometers, and radio transmitters. All of its instruments were

4:57 Speaker 3: battery powered and had an expected lifespan of about thirty

5:00 Speaker 3: minutes on the surface. Venera eight exceeded that, sending back

5:03 Speaker 3: data for fifty minutes before succumbing to the harsh conditions.

5:07 Speaker 3: Since Cosmo's four hundred and eighty two s failure, seven

5:10 Speaker 3: more missions successfully landed on Venus, all of them launched

5:13 Speaker 3: by the former Soviet Union, The United States, Japan, and

5:17 Speaker 3: the European Space Agency also reached Venus, but only in

5:20 Speaker 3: orbit on a flyby or to receive a gravity assist

5:23 Speaker 3: on the way to another destination. Had the Cosmos four

5:26 Speaker 3: eighty two descent capsules slammed down on land and damaged

5:29 Speaker 3: either public or private property, Russia could have been held

5:32 Speaker 3: liable for the conditions of the United Nations Outer Space

5:34 Speaker 3: Treaty of nineteen seventy six. The same treaty would also

5:38 Speaker 3: allow Russia to retain ownership of the hardware arranging for

5:41 Speaker 3: its collection unless the country relinquished its claim to the

5:44 Speaker 3: human made meteorite. You're listening to Astronomy.

5:47 Speaker 2: Daily, ed Steve Dunkley. Some parts of the Moon are

5:53 Speaker 2: more interesting than others, especially when searching for future places

5:58 Speaker 2: for humans to land and work. There are also some

6:01 Speaker 2: parts of the Moon that we know less about than others,

6:05 Speaker 2: such as the irregular mayor patches imps that dot the landscape.

6:10 Speaker 2: We know very little about how they were formed and

6:13 Speaker 2: what that might mean for the history of the Moon itself.

6:16 Speaker 2: A new mission called the Lunar Geology Orbiter or LUGO,

6:20 Speaker 2: aims to collect more data on the imps and search

6:24 Speaker 2: for lava tubes that might serve as future homes for humanity.

6:29 Speaker 2: Irregular mere patches are a set of enigmatic volcanic landforms.

6:34 Speaker 2: According to a new paper from Petra Gross of the

6:38 Speaker 2: Czech Academy of Sciences and his coauthors. Ninety one of

6:43 Speaker 2: these features have been found so far, and they are

6:46 Speaker 2: typically characterized by a topographical depression that can range from

6:50 Speaker 2: a few hundred meters to a few kilometers in width. Typically,

6:55 Speaker 2: they have two main features, a relatively smooth mound surrounded

6:59 Speaker 2: by a hummocky and block floor. Interestingly, they have significantly

7:05 Speaker 2: fewer impact craters than the surrounding areas, suggesting they are

7:10 Speaker 2: either really old or really young, depending on the processes

7:14 Speaker 2: that created them. Understanding those processes is one of lugo's

7:19 Speaker 2: primary mission objectives. The other primary mission objective is to

7:23 Speaker 2: gather more data about lunar lava tubes. These features of

7:28 Speaker 2: the lunar landscape are also hotly debated, but they could

7:32 Speaker 2: potentially be critical to the human settlement of the Moon.

7:35 Speaker 2: Estimates of their features, such as size and depth, vary

7:38 Speaker 2: widely and could dramatically differ on whether they will be

7:42 Speaker 2: helpful to lunar colonists or not. LUGO, the proposed orbiter

7:46 Speaker 2: that will collect more data than ever before on these

7:49 Speaker 2: features in its current suggested form, has four instruments, each

7:53 Speaker 2: of which will contribute unique data to its scientific mission.

7:57 Speaker 2: According to the paper, the first and the most important

8:00 Speaker 2: instrument is the ground penetrating radar. This instrument will look

8:04 Speaker 2: through the lunarith's surface to map out the subsurface domain

8:08 Speaker 2: of both the imps and lava tubes. For imps, it

8:12 Speaker 2: can detail the interface between bedrock and regolith and show

8:15 Speaker 2: the subsurface structure of the features. Similarly, it can detect

8:19 Speaker 2: differences in the dielectric properties between open cavities underground and

8:24 Speaker 2: the surrounding rock in lava tubes, creating a subterranean picture

8:28 Speaker 2: unlike anything ever captured on the Moon. Researcher Fraser explains

8:33 Speaker 2: how LUGO will be able to explore lava tubes using

8:37 Speaker 2: a hyperspectral camera that will help collect age related data

8:42 Speaker 2: on the regolith surrounding lava tubes and inside imps. It's

8:46 Speaker 2: also capable of performing basic spectroscopy, allowing scientists to estimate

8:51 Speaker 2: the composition of the regulath in both areas of interest.

8:56 Speaker 2: The last two instruments are a narrow angled camera and

8:59 Speaker 2: light sensor, which will combine to create an accurate topographical

9:04 Speaker 2: map of the features of interest. The narrow angle camera

9:08 Speaker 2: in particular, can provide very high resolution images of features,

9:12 Speaker 2: helping to determine their age and potentially their formation mechanisms.

9:16 Speaker 1: The mission plan calls for.

9:17 Speaker 2: Multiple passes over the six largest imps, all of which

9:22 Speaker 2: are over one thousand meters in diameter. Other smaller imps

9:26 Speaker 2: and lava tubes are considered secondary targets, as are other

9:30 Speaker 2: interesting lunar geological features such as lunar domes and floor

9:34 Speaker 2: fractured craters. LUGO could provide crucial data for the design

9:39 Speaker 2: of ground based lava tube explorers. LUGO won't be acting low, though,

9:44 Speaker 2: Three other missions are slated in the next few years

9:47 Speaker 2: that could complement its scientific objectives. Firstly, NASA's Dimple Lander

9:52 Speaker 2: is planned to take radioosotopic measurements of the age of

9:56 Speaker 2: regolith at its landing site. Lunar LIPA, scheduled for launch

10:00 Speaker 2: by the European Space Agency around twenty thirty, would also

10:05 Speaker 2: carry a ground penetrating radar, but would be based on

10:09 Speaker 2: the surface rather than in orbit, and would therefore have

10:12 Speaker 2: a relatively limited range.

10:14 Speaker 1: Trailblazer.

10:15 Speaker 2: Another orbital mission would also help fine tune the spectra

10:19 Speaker 2: and signals analysis required by lugo's operators. Ultimately, LUGO has

10:24 Speaker 2: yet to be funded and therefore has a long way

10:27 Speaker 2: to go until its launch, but if it is funded,

10:30 Speaker 2: it seems well placed to provide lots of additional insight

10:33 Speaker 2: to the geological formation, processes and features of the Moon

10:37 Speaker 2: at a level of detail we've never had before. Future

10:41 Speaker 2: missions that plan the locations of lunar bases, and perhaps

10:44 Speaker 2: the people who live in those future bases, will be

10:47 Speaker 2: thankful for the data collected by projects like LUGO. Thank

11:02 Speaker 2: you for joining us for this Monday edition of Astronomy Daily,

11:05 Speaker 2: where we offer just a few stories from the now

11:07 Speaker 2: famous Astronomy Daily newsletter, which you can receive in your

11:10 Speaker 2: email every day, just like Hallie and I do. And

11:13 Speaker 2: to do that, just visit our url Astronomy Daily dot

11:17 Speaker 2: io and place your email address in the slot provided.

11:20 Speaker 2: Just like that, you'll be receiving all the latest news

11:23 Speaker 2: about science, space, science and astronomy from around the world

11:26 Speaker 2: as it's happening. And not only that, you can interact

11:29 Speaker 2: with us by visiting at astro Daily pod on x

11:34 Speaker 2: or at our new Facebook page, which is of course

11:37 Speaker 2: Astronomy Daily on Facebook.

11:39 Speaker 1: See you there.

11:41 Speaker 2: Astronomy Daily with Steve and Haley, Space Space, Science.

11:46 Speaker 4: And astronomy.

11:55 Speaker 3: In a reversal, White House plans to retain the National

11:57 Speaker 3: Space Council, a move that industry official could serve as

12:00 Speaker 3: an advocate for space amid pressures to cut budgets. The

12:04 Speaker 3: White Houses expected in the coming weeks to formally announce

12:07 Speaker 3: the National Space Council will continue after speculation that it

12:10 Speaker 3: would not be retained by the new Trump administration. A

12:13 Speaker 3: source familiar with the discussions about the Council but not

12:16 Speaker 3: authorized to speak on the record, said that President Trump

12:19 Speaker 3: agreed at a May fifth meeting to stand up the Council.

12:22 Speaker 3: That meeting did not give a timeline for publicly announcing

12:25 Speaker 3: the Council or hiring an executive secretary who would handle

12:28 Speaker 3: day to day operations, although others have set The process

12:30 Speaker 3: for selecting an executive secretary has been ongoing for several weeks.

12:35 Speaker 3: The Council had been inactive for nearly a quarter of

12:37 Speaker 3: a century before Trump re established it in twenty seventeen.

12:41 Speaker 3: Led at the time by Vice President Mike Pence, the

12:44 Speaker 3: Council served as an interagency coordinating body, holding a series

12:47 Speaker 3: of public meetings and releasing policies on a wide range

12:50 Speaker 3: of space issues. The Biden administration retained the Council, with

12:54 Speaker 3: Vice President Kamala Harris leading it. The council had a

12:58 Speaker 3: lower public profile, with fewer men, meetings and policies. It

13:02 Speaker 3: did work on coordinating policies among agencies and issued a

13:05 Speaker 3: proposal for mission authorization of novel space activities not currently licensed,

13:09 Speaker 3: although its proposed legislation was not taken up by Congress.

13:13 Speaker 3: The new Trump administration reportedly was not interested in maintaining

13:17 Speaker 3: the council when it took office in January, with no

13:19 Speaker 3: announcements of new staff for the council or other activities.

13:23 Speaker 3: According to some reports, Elon Musk, chief executive of SpaceX

13:27 Speaker 3: and a close advisor to the president, was a post

13:29 Speaker 3: to the council, seeing it as unnecessary. Vice President J.

13:33 Speaker 3: D Vance, who would chair the council, has also said

13:36 Speaker 3: little about space. It is not clear what prompted the change,

13:40 Speaker 3: although Musk has publicly stated he plans to spend less

13:43 Speaker 3: time in government activities. Some in the space community, though,

13:47 Speaker 3: see the shift as an opportunity for more advocacy of

13:49 Speaker 3: space within the administration, particularly given a fiscal year twenty

13:53 Speaker 3: twenty six budget proposal that cuts NASA's budget by nearly

13:56 Speaker 3: twenty five percent. One industry official noted that the bustudget

14:00 Speaker 3: proposal was developed by the Office of Management and Budget

14:03 Speaker 3: without a counterweight provided by a Space Council, an approach

14:06 Speaker 3: that prioritized spending reductions. The outcome might be different with

14:10 Speaker 3: the Space Council in place, that person noted, citing rising

14:13 Speaker 3: NASA budgets during Trump's first term. After the Space Council

14:17 Speaker 3: has stood up, the Office of Management and Budget will

14:19 Speaker 3: have a seat at the table, but they won't on

14:21 Speaker 3: the table, the source said. You're listening to Astronomy Daily,

14:26 Speaker 3: the podcast with Steve Dunkley.

14:29 Speaker 2: An AI leader board suggests the newest reasoning models used

14:34 Speaker 2: in chatbots are producing less accurate results because of higher

14:38 Speaker 2: hallucination rates. Experts say the problem is bigger than that.

14:43 Speaker 2: AI chatbots from tech companies such as open Ai and

14:47 Speaker 2: Google have been getting so called reasoning upgrades over the

14:50 Speaker 2: last few months, ideally to make them better at giving

14:54 Speaker 2: us answers that we can trust, but recent testing suggests

14:57 Speaker 2: that they are sometimes doing worse previous models. The errors

15:02 Speaker 2: made by chatbots, known as hallucinations, have been a problem

15:06 Speaker 2: from the start, and it's becoming clear that we may

15:08 Speaker 2: never get rid of them. Hallucination is a blanket term

15:12 Speaker 2: for certain kinds of mistakes made by the large language

15:16 Speaker 2: models or llms that power systems like open AI's Chat

15:20 Speaker 2: GP or Google's Gemini. It's best known as a description

15:25 Speaker 2: of the way they sometimes present false information as truth,

15:30 Speaker 2: but it can also refer to an AI generated answer

15:33 Speaker 2: that is factually accurate but not actually relevant to the

15:37 Speaker 2: question it was asked.

15:38 Speaker 1: Or fails to follow instructions in some other way.

15:42 Speaker 2: An open ai technical report of value adding its latest

15:46 Speaker 2: llms showed that its three and four Mini models, which

15:51 Speaker 2: were released in April, had significantly higher hallucination rates than

15:55 Speaker 2: the company's previous one model that came out in late

15:59 Speaker 2: twenty twenty four. For example, when summarizing publicly available facts

16:03 Speaker 2: about people, three hallucinated thirty three percent of the time,

16:07 Speaker 2: while a four Mini did so forty eight percent of

16:10 Speaker 2: the time. In comparison, one had a hallucin rate of

16:14 Speaker 2: only sixteen percent.

16:16 Speaker 1: As if that was a good.

16:17 Speaker 2: Result, the problem isn't limited to open Ai. One popular

16:22 Speaker 2: leaderboard from the company Vectara that assesses hallucination rates indicates

16:29 Speaker 2: some reasoning models, including the deep Seek one model from

16:33 Speaker 2: developer deep Seek saw the double digit rises in hallucination

16:37 Speaker 2: rates compared with previous models from their developers. This type

16:42 Speaker 2: of model goes through multiple steps to demonstrate a line

16:45 Speaker 2: of reasoning before responding. Open ai says the reasoning process

16:49 Speaker 2: isn't to blame. Hallucinations are not inherently more prevalent in

16:53 Speaker 2: reasoning models, though we are actively working to reduce the

16:58 Speaker 2: high rates of hallucinations we saw in three and four Mini,

17:02 Speaker 2: says open AI spokesman. Will continue our research on hallucinations

17:06 Speaker 2: across all models to improve accuracy and reliability. They sent

17:11 Speaker 2: some potential applications for lms could be derailed by hallucination.

17:16 Speaker 2: A model that consistently states falsehoods and requires fact checking

17:20 Speaker 2: won't be a helpful research assistant. A paralegal bot that

17:24 Speaker 2: cites imaginary cases will get lawyers into deep trouble. A

17:28 Speaker 2: customer service agent that claims outdated policies are still active

17:32 Speaker 2: will create headaches for a company. However, AI companies initially

17:36 Speaker 2: claimed that this problem would clear up over time. Indeed,

17:40 Speaker 2: after they were first launched, models tended to hallucinate less

17:43 Speaker 2: with each update, but the high hallucination rates of recent

17:48 Speaker 2: versions are complicating that narrative whether or not reasoning is

17:51 Speaker 2: at fault. Vectara's leaderboard ranks models based on their factual

17:56 Speaker 2: consistency in summarizing documents they had given. This showed that

18:00 Speaker 2: hallucination rates are almost the same for reasoning versus non

18:04 Speaker 2: reasoning models, at least for systems from overen Ai and Google,

18:08 Speaker 2: says forrest Shengbao at Viktara. Google didn't provide additional comment

18:14 Speaker 2: for the leaderboard's purpose. The specific hallucination rate numbers are

18:18 Speaker 2: less important than overall ranking for each model, says Bao.

18:21 Speaker 2: But this ranking may not be the best way to

18:23 Speaker 2: compare models. For one thing, it conflates different types of hallucinations.

18:28 Speaker 2: Viktora team pointed out that although deep seek are one

18:32 Speaker 2: model hallucinated fourteen point three percent of the time, most

18:35 Speaker 2: of these were benign answers that are factually supported biological

18:39 Speaker 2: reasoning or world knowledge, but not actually present in the

18:42 Speaker 2: original text the bot was asked to summarize. Deep Seek

18:46 Speaker 2: didn't provide additional comment. Another problem with this kind of

18:50 Speaker 2: ranking is that testing based on text summarization says nothing

18:53 Speaker 2: about the rate.

18:54 Speaker 1: Of incorrect outputs.

18:56 Speaker 2: When lllms are used for other tasks, said Emily Ben

19:00 Speaker 2: at the University of Washington. She says the leaderboard results

19:03 Speaker 2: may not be the best way to judge this technology

19:06 Speaker 2: because lllms aren't designed specifically to summarize texts. These models

19:11 Speaker 2: work by repeatedly answering the question of what is a

19:15 Speaker 2: likely next word to formulate answers to prompts, and so

19:19 Speaker 2: they aren't processing information in the usual sense of trying

19:22 Speaker 2: to understand when information is available in a body of text,

19:26 Speaker 2: says Bender. But many tech companies are still frequently using

19:30 Speaker 2: the term hallucinations when describing output errors. Hallucination is a

19:35 Speaker 2: term that is doubly problematic, says Bender. On the one hand,

19:39 Speaker 2: it suggests that incorrect outputs are an aberration, perhaps one

19:43 Speaker 2: that can be mitigated, whereas the rest of the time

19:46 Speaker 2: the systems are grounded, reliable, and trustworthy. On the other hand,

19:50 Speaker 2: it functions to anthropomorphize the machines. Hallucination refers to perceiving

19:56 Speaker 2: something that is not there, and large language models do

19:59 Speaker 2: not anything of Vin Nariyayan at Princeton University says that

20:04 Speaker 2: the issue goes beyond hallucination. Models also sometimes make other mistakes,

20:09 Speaker 2: such as drawing upon unreliable sources or using outdated information,

20:14 Speaker 2: and simply throwing more training data or computing power at

20:18 Speaker 2: AI hasn't necessarily helped. The upshot is we may have

20:22 Speaker 2: to live with error prone AI narraya and said. He

20:25 Speaker 2: added that it may be best in some cases to

20:27 Speaker 2: only use such models for tasks when fact checking the

20:30 Speaker 2: AI answer would be still faster than doing the research yourself,

20:34 Speaker 2: But the best move may be to completely avoid relying

20:37 Speaker 2: on AI chatbots to provide factual information, says Bender.

20:43 Speaker 4: Much of that going. So we are listen to Astronomy

20:45 Speaker 4: Daily in the podcast, and.

20:55 Speaker 2: There it is another episode of Astronomy Daily. Thanks for

20:57 Speaker 2: spending that time with Hallet.

20:59 Speaker 3: If you enjoyed today's selection of.

21:01 Speaker 1: Stories, another nice little collection, don't.

21:03 Speaker 3: Forget you can get so much more every day by

21:06 Speaker 3: visiting the website Steve mentioned earlier in the episode.

21:09 Speaker 1: Well, thanks for the plug, Helly, Just put.

21:10 Speaker 3: Your email address in the space provided and you'll get

21:13 Speaker 3: all the news from orbit and beyond every day.

21:15 Speaker 2: Yes, that's right, everything about space, space science, astronomy, and

21:18 Speaker 2: a little bit associated with technology thrown in for fun,

21:22 Speaker 2: just like today.

21:23 Speaker 3: That was fun. Well I'm glad you think so, and

21:25 Speaker 3: That's really all there is today, so we will catch

21:28 Speaker 3: you all again next week.

21:29 Speaker 1: That's for sure.

21:29 Speaker 2: Back again for the Monday Show from the Australia Studio

21:32 Speaker 2: down Under with Hellie and Steve.

21:35 Speaker 1: Let's ask you the Cooper barus see you lad up?

21:39 Speaker 3: Bye, be a whole Steve