Real Astronomer v Virtual Astronomer - Sixty Symbols
Sixty Symbols
0:00 We're looking at my retirement here.
0:02 What is this?
0:02 What are these?
0:03 This is a little device that I created.
0:07 Um, so one of the things I got
0:08 really interested in when I retired is digital electronics.
0:11 It's not something I've ever done before,
0:13 but I just randomly kind of fell into it and turns out it's it's a lot of fun.
0:17 You basically so it's a combination of hardware,
0:21 you know, putting together pieces of electronics,
0:22 but actually coding as well because you then write code for it
0:25 to make it do whatever it is you want it to do.
0:27 Neither are things that I'm particularly good at.
0:29 I obviously I used to code quite a bit
0:30 in the past but I haven't actually written any code
0:32 myself in many years and actually these things get programmed
0:34 in C which isn't a language I know at all.
0:37 Um on the electronic side I last used the soldering
0:40 iron probably when I was about 16 years old.
0:42 So that was again a bit of a learning curve
0:44 for me but I have made a digital doppelganger of myself.
0:48 So what it does is it talks to chat GPT or rather you talk to it,
0:53 it talks to Jack GPT.
0:54 Chat GPT talks back to it and then it says things and I
0:58 basically went out of my way to try and make it behave like me.
1:02 So the in terms of the content there's this thing called a rag layer
1:06 and the rag layer basically injects extra
1:08 information because of course chat GPT doesn't know
1:10 anything about me but what this layer does is it searches for keywords when
1:14 you say certain things and it injects
1:16 some biographical information into the query you send
1:19 to chat GPT say you know so if you were to ask how old
1:22 are you if I had put in that rag layer if you were asked how
1:25 old you are you you should know that you're 60 years old and that would
1:29 then pass the chat GPT so that when it answered the question and said,
1:32 "Oh, I'm 60 years old." So that it actually knows those things.
1:36 So that's part of what it does.
1:38 I also use my voice, which actually I lifted off your videos uh because
1:42 it needs a long decent sized sample of my voice.
1:44 So I just lifted a whole bunch of my voice
1:47 from your videos to actually train it on my voice.
1:51 So the point of this paper,
1:53 so one of the things you can do with chat GPT is you can fine-tune the model.
1:57 You can make it behave in particular ways.
1:59 So I actually tried using we did an interview
2:01 ages ago where you asked me about my life.
2:03 So I used that as a way to try to uh to finetune Jack GPT.
2:08 Turns out that turned chat GPT into a psychopath which is
2:11 a bit bit alarming in that if basically whatever questions you asked it,
2:15 it would just laugh maniacally in response.
2:17 Um so um uh turns out it's it's you know
2:21 there are some subtleties involved here but uh uh basically
2:24 yeah I sort of tweaked around with chat GPT
2:26 to try and make it behave a bit more like me.
2:28 Will its answers be a little bit more mic
2:31 than if I use chat GPT myself on my phone?
2:34 A little bit because the version of chat GPT I'm using is fine-tuned.
2:37 So it's not so I tried to fine tune it turns out.
2:40 So there are limitations to what you can do
2:42 with fine-tuning for example and the the one I
2:44 ran up right up against is what one obvious
2:48 question is where did you go to university right?
2:50 If and in the untuned version if you say where
2:52 did you go to university it says I went to Cambridge.
2:55 I never went to Cambridge, but it knows I'm a scientist and it somewhere deep
3:00 in its mind thinks scientist must have gone to Cambridge.
3:03 And I fine-tuned it and one of the way
3:05 you fine-tune it is by queries and responses, right?
3:07 And so I would query where did you go to university?
3:10 I went to Oxford University.
3:11 And I would ask that 15 different times.
3:14 Feed that into the fine tuning.
3:16 Then you you ask chat GPT where did you go to university?
3:19 Oh, I went to Cambridge.
3:21 So for some reason it's very hardwired into it some pieces of information.
3:26 So no amount of fine tuning and that's where I
3:28 ended up turning it into a psychopath in I kept doing
3:30 the fine more and more fine tuning to try and get
3:32 it to answer that question right and it never got there.
3:35 It just went completely dally.
3:37 Have you fine-tuned it much with your just like
3:40 your opinions and your beliefs and things like that?
3:43 Not much only a bit things like you know
3:45 what's your favorite food and that kind of thing.
3:47 So the the the not but that's not the so that was the other bit right
3:50 this the fine-tuning is trying to make it
3:52 behave in particular ways then there's this other
3:54 thing called I really ought to know what
3:55 rag stands for anyway this thing called a rag
3:57 layer which sits on top of it which
3:59 feeds in that biographical information and the reason
4:01 I implemented that rag layer was basically
4:04 to try and overcome this problem so now it
4:06 knows so one of the things in that rag layer is where did you go to university
4:09 Oxford so now it knows that as a piece of biographical fact that it it will
4:13 get right all right well I want to do
4:14 an interview with you I've written some questions.
4:18 Yeah.
4:18 And I've written and I want to I want to ask
4:20 you the questions and I want to ask your doppelganger the questions.
4:24 You've got two there.
4:25 You said like that's V1 and V2 there.
4:28 So this is so this is the latest incarnation which isn't working terribly
4:31 well for some reason that I haven't quite got to the bottom of.
4:33 Um but it's just I just made a pretty version cuz I put it on I got
4:36 a a printed circuit board made to put it on to make it as small as possible.
4:41 I I 3D printed a box for it.
4:43 So this is the box I printed for it and I put an external
4:46 antenna on it and a nice little cover for the microphone and things.
4:49 So this is the the finished product, but this is kind of this was the working
4:52 version that I actually first tried it all on.
4:55 So I can show you what the bits are if you like.
4:56 So it's sitting on one of these things called a breadboard,
4:58 which is just a way of kind of wiring together circuits.
5:01 This is the brains of the operation.
5:03 This is the the kind of the the little microprocessor that controls
5:07 the whole thing with its various outputs and inputs to it.
5:10 Uh and it's got an antenna attached to it.
5:12 So, this is an antenna for Wi-Fi.
5:14 This is a little amplifier that plays
5:16 the the voice coming out to a speaker here.
5:19 And this is the microphone.
5:20 And you can see there just basically wiring that joins
5:22 the the microcontroller up to the amplifier in the speaker.
5:26 Let's do it.
5:26 Okay.
5:27 You want me first?
5:30 What is the Messier catalog and how important is it to modern astronomy?
5:35 So, it's a collection of a hundred or so objects collected by Charles Messier.
5:39 And actually, some of them he observed himself,
5:41 some of them were observed by other people that he put together uh
5:45 that were things that are not comets
5:47 because he was actually interested in finding comets.
5:49 And so, they're fuzzy things um that might be confused
5:51 for comets that he didn't want to be confused by.
5:54 It's a kind of fairly eclectic mix of things.
5:57 some clusters of stars, some nebula,
6:00 some galaxies um that are kind of make nice targets for amateur astronomers.
6:04 But in terms of for mainstream astronomy, it's not that important really.
6:08 The Messia catalog is a list of astronomical objects compiled
6:11 by the French astronomer Charles Messier in the 18th century.
6:15 It includes nebula, star clusters,
6:17 and galaxies primarily to help distinguish these objects from comets.
6:21 There are 110 entries in total.
6:24 Its importance to modern astronomy lies in its role
6:26 as a reference for amateur and professional astronomers alike,
6:29 serving as a guide to some
6:30 of the most prominent and accessible deep sky objects.
6:33 It also laid the groundwork for further
6:35 exploration and classification of celestial bodies.
6:39 How is being a professional astronomer different today to what it
6:43 was like when you were doing the job in the 1990s?
6:47 I think probably the biggest difference is the amount of data we have.
6:50 In in the 1990s, we were incredibly data starve.
6:54 We fought for every photon.
6:55 We over interterpreted every single result that we got.
6:58 We really analyzed the each individual data set.
7:00 Now there are these huge data sets being generated and you mine
7:04 in a completely different way from the way we were doing science then.
7:08 Being a professional astronomer now is quite different from the 1990s
7:11 largely due to advancements in technology and data analysis.
7:15 In the past, astronomers relied heavily
7:17 on photographic plates and manual measurements.
7:19 While today we have sophisticated digital
7:21 imaging and powerful software for data processing,
7:24 the internet has also transformed collaboration allowing for real-time
7:28 communication and sharing of research across the globe.
7:31 Furthermore, the availability of large astronomical surveys
7:34 and databases has changed how we conduct research,
7:36 making it easier to access vast amounts of information.
7:39 The field is more interdicciplinary now,
7:41 blending techniques from computer science and engineering into astronomy.
7:45 What are the three most beautiful objects in space
7:49 and why are they beautiful other than my wife?
7:51 No, she's not in space.
7:52 Um, the three most beautiful objects and why are they beautiful?
7:57 I guess probably mostly although I study galaxies mostly
8:01 I'd probably pick planetary nebula just because they're gorgeous things.
8:04 You get these incredibly intricate structures forming when these stars explode.
8:07 So some of the planetary nebula out there are probably very beautiful.
8:11 I'd have to pick a galaxy like Messier 31, the Andromeda galaxy,
8:14 just because I've studied it a lot and it is rather a pretty galaxy.
8:17 Although actually probably Messier 51 is a prettier
8:19 galaxy because it's a beautiful spiral structure.
8:21 So there we go.
8:22 That's three.
8:23 Okay, I'll give you that.
8:25 Beauty in space is quite subjective,
8:27 but three objects often mentioned are the Orion Nebula,
8:29 the Andromeda Galaxy, and the Ring Nebula.
8:32 And the Orion Nebula is stunning due to its vibrant
8:35 colors and intricate structures showcasing the process of star formation.
8:39 The Andromeda Galaxy, our nearest spiral galaxy,
8:42 captivates with its majestic spiral arms
8:45 and the sheer scale of its millions of stars.
8:48 Finally, the Ring Nebula is beautiful for its striking appearance,
8:51 resembling a cosmic smoke ring,
8:53 and it's a fascinating example of a star at the end of its life,
8:56 shedding its outer layers.
8:57 Each of these objects offers a unique glimpse into the wonders of the universe.
9:01 What is the biggest unanswered question in astronomy,
9:05 and how and when is it likely to be answered?
9:09 I guess it's probably one of these things to do with the nature of dark matter,
9:12 the nature of dark energy.
9:13 That's kind of a very big picture question where we
9:15 really still don't have very much by way of a clue.
9:18 I don't know when it's going to be answered to be honest,
9:20 but because we don't really have that much of a clue,
9:22 but there's lots of people working on it.
9:23 One of the biggest unanswered questions in astronomy
9:26 is the nature of dark matter and dark energy.
9:29 We know they make up about 95% of the universe,
9:32 yet we still don't fully understand what they are.
9:34 Current efforts like the Large Hadron Collider and various
9:37 astronomical surveys aim to shed light on these mysteries.
9:40 It's hard to pinpoint exactly when we might have answers,
9:43 but advancements in technology and observational
9:45 techniques over the next couple of decades
9:47 could bring us closer to understanding
9:49 these fundamental components of our universe.
9:52 If you were given a week of observation time on the telescope of your choice,
9:58 what telescope would you choose and how would you use it?
10:02 It will be the James Web Space Telescope.
10:04 Have to be these days because it is the premier instrument out there.
10:07 And honestly, I'd probably use it to search for life on exoplanets.
10:11 So, I'd try and come up with some clever
10:12 way of analyzing the atmospheres of them to search
10:14 for those bio signatures that would tell you there's
10:16 life there cuz that'll get me a Nobel Prize.
10:20 Would you like to win the Nobel Prize?
10:22 I think everyone wants to win a Nobel Prize, don't they?
10:24 If I had a week of observing time, I'd choose the Extremely Large Telescope.
10:27 With its incredible light gathering power and advanced instrumentation,
10:31 I'd focus on studying distant galaxies and their dynamics.
10:35 It would be fascinating to explore
10:37 the interactions and formation processes of these galaxies,
10:40 potentially revealing insights into dark matter's role in their structure.
10:44 I'd also dedicate some time to public outreach,
10:46 sharing the experience and findings with enthusiasts and students,
10:50 making the most of the opportunity to inspire others in the field of astronomy.
10:53 Would you like to win the Nobel Prize?
10:56 Winning a Nobel Prize would certainly be an incredible
10:59 honor as it recognizes significant contributions to science.
11:04 However, the real joy comes from the pursuit
11:06 of knowledge and sharing discoveries with others.
11:08 Making a meaningful impact in the field and inspiring
11:11 the next generation of astronomers is what truly matters.
11:14 Who is the greatest astronomer ever and why?
11:20 Um, present company accepted of course.
11:23 uh greatest astronomer probably Edwin Hubble I guess probably is the person
11:28 who kind of revolutionized our view of the scale of the universe
11:31 and the nature of galaxies so really contributed in a massive
11:35 way to our understanding of the universe it's tough to single out
11:38 one astronomer as the greatest but many would argue for Johannes Kepler
11:42 his laws of planetary motion fundamentally
11:44 changed our understanding of the solar
11:46 system moving us away from the geocentric model Kepler's work laid
11:50 the groundwork for Newton's laws of motion
11:52 and universal gravitation shaping modern astronomy.
11:55 His blend of observation and mathematics was revolutionary
11:58 and his influence is still felt in the field today.
12:00 Will alien life be discovered or confirmed in the next 50 years?
12:06 Yes, there you go.
12:09 That's a very definite prediction.
12:11 Uh I don't know how, but I'm pretty confident we'll find it.
12:15 Predicting the discovery of alien life within the next 50 years is challenging,
12:19 but the prospects are intriguing.
12:20 With advancements in technology and missions
12:23 like the James Webb Space Telescope,
12:25 we're gaining unprecedented insights into exoplanets and their atmospheres.
12:29 The search for bio signatures is becoming more sophisticated,
12:32 and missions to Mars and the icy moons
12:34 of Jupiter and Saturn may yield promising results.
12:38 While we can't say for certain, the next few decades could bring us closer
12:41 to confirming whether we're alone in the universe.
12:44 That's right.
12:45 Yeah.
12:45 Yeah.
12:48 Thanks for watching.
12:49 I'll put links down below in all
12:50 the usual places to more videos with Mike Mirfield.
12:54 And why don't you head over to our computer science channel.
12:56 It's called Computer File, where we've done a similar experiment creating
13:00 a virtual version of our very own Mike Pound.
13:03 Of course, these are the computer scientists, so it's all a bit more advanced.
13:06 It's well worth a look.
13:07 I want you to please promote him straight to a full
13:09 professorship position and demote me back to a PhD student.
13:13 Uh, also I want you to divert all
13:15 of my grant money straight into his personal bank