Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Episode 9 - Artificial Intelligence (AI)
1.3k Plays7 months ago
In this episode, we discuss Artificial Intelligence (AI) and cover how we can use this exciting (and sometimes scary) technology in our creative pursuits. Our host Mary Tsai, who works in the AI field, helps us understand fact vs. myth, as we cover many topics that we are forced to address now that AI plays such a prominent role in our daily lives.
To watch the YOUTUBE VIDEO of this episode and the irreverent & somewhat unpredictable AFTERSHOW, subscribe to our Patreon: http://patreon.com/user?u=91688467
Recommended
Transcript
Introduction of Hosts and Episode Topic
00:00:15
Speaker
Welcome back, everyone, to your favorite new irreverent podcast, all about creativity, art, and design. I'm your host, Paul Jasper, woodworker at night, scientist during the day. You see, I switched that up. And I'm joined by my two fabulous co-hosts, Eric Curtis, a full-time, very professional woodworker, and also
00:00:36
Speaker
budding content creator, not even budding, you're far beyond budding. And Maryse, I, who is a professional AI designer by day, woodworker by night.
00:00:48
Speaker
Okay. I didn't realize that I was still budding four years into this thing and I'm still just finding my step, huh? Look, if you don't want things written down, you're going to have to deal with the randomness of my goddamn intros. Hi, everyone. Okay, so welcome to today's episode.
Exploring Artificial Intelligence and Design
00:01:08
Speaker
And I'd like to start by saying today's episode is on a very hot topic.
00:01:14
Speaker
that we knew we were going to cover, because as you probably just heard in my intro, I stressed that Mary Tsai is an AI designer by day. So we will be covering the contentious, the misunderstood, the amazing topic of artificial intelligence, or AI. And we're going to handle it in two senses. The first sense
00:01:39
Speaker
is in a general sense, we'd like to ask, what is AI? What are some of the fears? And what's some of the problems that it holds for the future? But then we want to apply it directly to all of us, which is as artists, as craftspeople, how can we use it in our design process? And I'll get to some of the other questions. But we want to start broad and then narrow the focus towards how it applies to us.
00:02:03
Speaker
So that being said, we have our own guru, Mary. She loves when I build her up. She loves when I put like additional pressure on her to sound like amazing. So I'm going to continue. So Mary is our own personal guru about AI. Mary eats, breathes, and sleeps design and AI for many years now. So we thought that she could start by telling us, and I think a lot of us have this question in a general sense, what is AI?
Understanding AI: Types and Everyday Integration
00:02:33
Speaker
Yeah, sure. So I'm gonna give like really general, easy to understand descriptions and definitions. I hope that's okay. But there's a few different types of AI and the term is thrown around really loosely these days. Like artificial intelligence itself is just it's a subfield of computer science using machines to simulate human intelligence and
00:02:56
Speaker
There's a lot of different types. So the one that most people are hearing about over the past year, two years is generative AI. So generative AI uses machines to create slash generate new things. So new things, meaning like images, videos, writing, this is generally coming from text prompts. And then traditionally, I use this machines to analyze and find patterns and data. So things like
00:03:22
Speaker
uh for example like finding like fraud or spam or you know social media algorithms etc and traditional AI um utilizes machine learning so machine learning is a subset of AI where machines learn from experience without explicit programming it uses algorithms that can learn any kind of task so when people are saying like AI is like a hot topic yes it is i will say generative tech is a hot topic however um AI has been
00:03:50
Speaker
in all of your software, all of your technology for the past decade easily, people were just not aware of it. And it's this is really becoming the first time where people have to actually realize what data means. Like in the past, I feel like we've always accepted it as just a way of using tech, like, Oh, yeah, my data will be collected for ads, etc. But I don't think people understand that literally anything they release online becomes integrated into these data sets that
00:04:17
Speaker
become part of the internet. So yeah. So Mary, the point that you raised that I wanted to pick up on that, and by the way, great definition was we've been using AI for many years already. So spoiler alert for those listeners, who's like, ah, I don't trust AI and I don't like this new fangled technology. I don't know why I'm a southerner who's able
00:04:43
Speaker
That's just your alternative lifestyle, buddy. That's who I wish I really was. For those people who have that, guess what? Spoiler alert, you're already using it. Digital assistants like Google and Alexa and Siri, they all use AI. Social media algorithms curating your feed, AI. Smart homes, your Nest thermostats,
00:05:04
Speaker
AI, GPS mapping software, AI, smart cars, banking, media streaming, Netflix suggesting shows, video games, facial recognition to unlock your iPhone. All of these are examples of AI already commonplace in today's society. And I think it warrants setting that precedent that this is not exactly new. Some of the generative stuff is certainly new, as Mary said. But anyway, that was a great point that it's already been
00:05:35
Speaker
In in our world for a long time. So yeah, I'm sure if all the fear is warranted But but Paul don't you think and I'm just I'm gonna go off on my first tangent on this Oh, yeah, I'm sure but don't you think
Societal Impact and Ethical Concerns of AI
00:05:47
Speaker
like there's there's this tipping point that may be happening right now or maybe it's just you know, Mass hysteria, you know dogs and cats living together who knows what's happening, but I wonder if this
00:06:00
Speaker
this thing of AI being cute at first because it was limited and now it's becoming so ubiquitous that we're realizing its potential and I'm like the thing that's that's running a parallel in my head in this may be an incomplete parallel but
00:06:14
Speaker
When you have a kid, it's adorable. It's cute. And you know, it's going to grow up to be a full human at some point. And then that thing, like the first time your kid beat you in a foot race and you're like, holy shit, like this, this kid is like turning into a real human. And then one day they just like, you have to accept that they are their own thing.
00:06:32
Speaker
and you no longer have the ability to restrict them and it becomes scary and is the difficult transition for parents to make. I wonder if we as a society are now coming to this point where AI is so ubiquitous and potentially powerful where we are just realizing like the we never asked should we just asked can.
00:06:51
Speaker
Yeah, I mean, part of it is also who's able to use it. So before, when I was designing things for AI, the whole point was it's supposed to be integrated into your software, it's supposed to be invisible, etc. However, now, it's like these tools that anyone could use, and it's not so much specialized. So one of the large part, like, I don't know if it manages the right word, but one of the benefits of artificial intelligence and generative tech is that it makes things accessible for people to use. And that's why
00:07:20
Speaker
people are starting to freak out a little bit more because anyone has access to it. Anyone can make these things before it was like a bit more specialized within your field or more hidden or you didn't have, you know, the specific education to utilize it. So now it's very much just depending on who has access to it. But hasn't that been the case with every technological advancement throughout history? Like if we just limit it to woodworkers for a moment, for example,
00:07:47
Speaker
There was this notion when the hollow mortiser came out, that they were like, this is the death of craftsmanship. Because now you don't need to be able to chop a mortise at 90 degrees or a particular angle. You don't need to take the time to learn how to sharpen a chisel. You just need to press a button and this machine is going to hollow out this piece for you. So it is a lesser example, but it was the same argument of like, well, this isn't craftsmanship. This is just machining.
00:08:16
Speaker
Well, Eric, that same argument came out when computer music came along, right? There's gonna all the musicians are going to be out of a job. It's when photography came along, all the artists and painters are going to be out of a job. And I think, you know, we've seen this time and time again that it does, it does actually upset certain fields, but others are largely, you know, still intact. So I think there is
00:08:38
Speaker
There's truth on both sides, that this can upset certain subfields and it probably doesn't displace others. So as a way to probe what the fears are surrounding AI in a general sense, I asked an AI engine, what are the top five reasons why people fear you? Classic newscaster right there. You never believe that chat GPT wrote that intro.
00:09:05
Speaker
I asked chat GPT, what are the top five reasons people fear artificial intelligence? And the answers were number one, job displacements. Sure. One of the primary concerns is that AI and automation could lead to widespread job loss across various industries. I get that. Yeah, I understand that. And I think it will display certain jobs in certain industries. Number two, Mary touched on this, privacy and surveillance.
00:09:31
Speaker
privacy of your data and surveilling constantly collecting everything you do as a means of training these algorithms. Number three, ethical dilemmas.
00:09:42
Speaker
Who bears the responsibility when AI systems make decisions, how to ensure there's not bias fed into the algorithms, copyright in terms of art? It goes on. And finally, the most concerning, I think, the doomsday scenario is loss of control, that the machines would actually become smarter than humans, will lose control, and they'll lead to an existential risk.
00:10:05
Speaker
Those are the, I collapsed two of them into one, so it's the top four. So forgive me, because they were, two of them were kind of similar. But those are the top four, job displacements, privacy and surveillance, ethical dilemmas and loss of control leading to existential risk, according to JET GPT. Mary responds because, you know, I'm going to sit back on that one. I have my thoughts, but they are not educated in the field of AI.
00:10:33
Speaker
I mean, yeah, I think that makes sense. What was the first one you said, Paul? Good job displacement. Job displacement. Yeah. I mean, I think that we've already seen it, to be honest. Job displacement for sure. One of the top questions I often get asked is, is AI going to take over my job? I ask that myself sometimes. Like, am I literally deciding my own death? Okay. Can I jump in on this point real quick? Because this is something maybe
00:11:01
Speaker
Eric, I thought you were going to let Mary answer. I know, and I was, but this is something that frustrates me, and I don't know if I'm going to upset people with this, and I don't mean to. I don't mean to, but...
00:11:17
Speaker
What really frustrates me about the conversation about job displacement around AI is we have been displacing jobs with new technologies for 200 years, 250 years since the industrial revolution. And nobody gives a flying fuck because they're all blue collar jobs. And now the moment we have a technological advancement that might displace white collar jobs, everybody is in a fucking tiff about it. And it is upper white middle-class people who are just like, this is a thing we should take really seriously now. Oof.
00:11:48
Speaker
Am I wrong? That 10% beer you're drinking is hitting right now. It's hitting hard right now. I don't mean to be an asshole about it because it is a thing that we should consider and we should approach thoughtfully. But it is not without, it bears mentioning
00:12:07
Speaker
that there is this very clear, like, nobody gave a shit when it was displacing people who were enslaved, people who were blue collar, people who were lower class, and now all of a sudden, now we're upset about job displacement. That's a really interesting point. I can't really speak to it, but that's a really interesting idea. Yeah, hasn't there been an uptick though in trade school, in trade school enrollments? Do you think that's related or no?
00:12:35
Speaker
I think that's been going on longer than the conversation around AI, but I can't imagine that one would drive the other. I think we're getting to a point where it's like, if everything we do is going to be displaced by AI and or technology, then why not, you know, do something that's more meaningful rather than try to make the best economical choice for it to only get displaced 20 years from now. But that's just a yes.
00:13:05
Speaker
So, so the privacy and surveillance fears, I think we all are, we understand that, you know, our data and whatnot. And the ethical dilemmas, I think I get that too. The privacy one though, I just, I get that actually worries me not in terms of like, what people will do with it. It's just a lot of people are not aware of how much they put out there. It's just,
00:13:31
Speaker
now on the internet, I worry about the youths that like just putting stuff on, you know, Instagram and like, it's so mindless because you think oh, it's just like a video or an Instagram story that's going to go away. And like all of this stuff is data that is
00:13:47
Speaker
that can be collected by companies, whether it's private or public. But I just worry that I hope that people are more cautious on why they actually release into the world. And that's not really put into younger people's minds right now in education, like, they just think that it's just part of, you know, their everyday life. Oh, yeah, let me record this. Let me like post it on something.
00:14:09
Speaker
It's not I just hope that they're ingraining that into, you know, younger, younger. Well, Mary, all I can tell you is I tell my daughter this, I know, teenager, and she didn't give a shit. So I'm telling her while she's snapping, you know, her friends and taking videos like, Oh, what's that dad? Yeah, they don't care.
00:14:31
Speaker
I know it's just kind of like the oh, you're so old, like, don't worry about it. But yeah, in fairness, I'm 36. And I don't give a shit either. Like, there's only so much I can worry about, you know, and it may not be the wisest decision, but I'm making the best choice in the moment I can. And I don't have enough information or education around data privacy and security to make a different choice.
00:14:55
Speaker
So I promised myself and us that we would limit the general conversation to 15 minutes. We just hit the 15 minute mark. I didn't even get to talk about the new of creating new things in AI. Sorry, baby. Sorry, baby. But I do want to ask Mary one last question before I leave this topic, which is about the laws of control leading to an existential risk. What do you think about that? I don't know. I do not at all know enough about this topic.
00:15:25
Speaker
I don't think that's likely. I really don't think it's likely, but I can't say it's very difficult. I don't think that we're going to become in a situation like, I don't know, or some of the movies. 2001 A Space Odyssey, where Hal takes over the ship. Sure. Can you go over that and not Terminator? I was going to say Terminator, yeah. Are you calling me old? No. I hate you both. Okay.
00:15:55
Speaker
Moving on, okay, so now let's focus the conversation now to art and design and woodworking and all the things we love to talk about. So how can we use, or can we use it, or how can we use AI in our design process? Ready, go.
AI in Design and Creative Processes
00:16:09
Speaker
Mary, you want to take this one first? You want me to jump on it? You do it. All right. So this is actually a great segue for me to circle back to the new conversation, because generative AI in the definition that one of you gave, I don't remember who, said to create new things.
00:16:28
Speaker
And it is clearly not creating new things. It is taking in data and inputs and generating an amalgamation, right? It's smashing them together in some kind of neutron collider and putting out information. I know. I want science on you, buddy. Neutron collider. Neutron collider. That's a thing that exists in the sciences, right?
00:16:52
Speaker
using it in our creative process in order like if you're using it as as an end I don't think you're going to get the results that you want or I should speak from the first person I'm not going to get the results that I want but I do think that it has the capacity to be a new creative tool to spark iterations and I don't know how to do that yet because I don't have enough experience with it but
00:17:18
Speaker
I think the folks who are throwing it out, throwing the baby out with the bathwater, as it were, are losing out on a potentially interesting seed of an idea. All right. So let me expound on that, Eric, because I did try it as a creative tool to give me inspiration when it first came out. So one of the generative engines is called Mid-Journey.
00:17:40
Speaker
And it creates really beautiful images with just a text prompt. So I was like, wow, this is incredible. I just like to understand more. So I bought a subscription to Mid-Journey and I started putting in, as you know, like occasionally I make sushi boards because it's just kind of a fun little canvas to make, you know,
00:17:59
Speaker
relatively quick turnaround project about a month long. And I asked the generative art engine to make me a design or show me a sushi board that had these elements. And I said, you know, gold filigree, like architectural elements, a place to rest your chopsticks, dark woods. And I told it all the keywords I wanted. And I hit go.
00:18:26
Speaker
And I'll be damned. What it gave me back was so fucking beautiful. I was a bit disturbed. Like how quickly the designs came back in 60 seconds and they were stunningly beautiful.
00:18:41
Speaker
Like I loved them and I saved them to my Google photo album, which probably is training some other AI engine. But anyway, I saved them to my Google because I think it would be fun to try making it. Like some of the design of it was not easily made because it doesn't think about the execution limitations.
00:19:03
Speaker
It just was responding to what I asked, and they were amazing. They were so beautiful. And I actually, I will try to make one as a future project, just as a sort of proof of principle go, Eric, what do you want? So you sent us some of those images, and they were beautiful. And in fact, so much so that I screenshattered a couple of those that you sent, so I could save it for future reference. But I wonder, my question is, how many iterations did you go through?
00:19:28
Speaker
because I'm wondering the data sets that it had, if you had gone through enough iterations, if it would just start spitting out the same shit over and over again, rather than creating new. So, new in that you hadn't seen it before, and it was interesting, but maybe not new in that it has a limited vocabulary because of the information that it has, and it can only create so many iterations.
00:19:51
Speaker
Yeah, if you keep giving it the same keywords, it will start to give you probably a similar range of possibilities. But then you throw in a new keyword or two, and suddenly that opens the doors to a little bit new design. Mary's nodding. See? Mary knows what's up. She's giving us that look of like, you two fucking idiots talking about something you know not.
00:20:11
Speaker
Go on Sensei, tell us about fucking AI. No, I mean, that is right. The more parameters you give it, the more it's able to create new versions iterations. And it sounded like you were pretty specific, which is great because you can get to exactly what you wanted. Some people work differently, some people give it only like a few specifications, and then they give it more slowly and iterate then.
00:20:36
Speaker
I usually do it just up front, so just a long string of words, requirements, and some examples to use as its source of origin, too. So that is right. I think that depending on the model, their specific models are better than the other ones. They will start to give you images that look similar. However, things like honestly stable diffusion mid-journey, they're so extensive that
00:21:03
Speaker
I think you could go for a while until you start getting repeat images, or almost repeat. Eric, did you hear? She said I was right. Did you hear that? Listen, Paul, as one of the top three smartest people I've ever known, I have no doubt, I have no qualms conceding that you're right in a lot of different ways. Also, fuck off.
00:21:33
Speaker
Mary's going to be the referee of this show. Thank you, Mary. Thank you for saying that was right.
00:21:42
Speaker
Makes me feel so good. OK, so how can we use it in a design process? In terms of how, one way is to use one of the graphic engines to derive, say, I want to see a cabinet with black walnut that is rectilinear or curvilinear with beautiful filigree and gold accents.
00:22:02
Speaker
you know and that can give you some spark it may not be your final cabinet but at least it could get you thinking in different ways about a different shape that you never imagined or about a different. Element graphically than you did that that never came to mind so that's one way what's another way.
00:22:22
Speaker
Well, sorry, I was just gonna tack on to your last point. Like, think of it kind of how folks use Pinterest, I guess, because Pinterest algorithm is pretty good when you are looking through and you see one, you tap on it and you scroll down, you want to see others that are similar. It's actually pretty similar to that, except a lot more specific and refined to what you're looking for. So
00:22:42
Speaker
I think that has been really helpful for me. Another way that you can use it is like the text prompts. So this is more, you know, maybe on people who are creating content, like, you know, just iterating very quickly on the tedious things that you don't really want to do. So I'm not a content creator, but like I imagine being able to, you know, generate a bunch of descriptions or
00:23:06
Speaker
even thumbnails and being able to, actually, one thing a lot of people don't use AI for, and it is so good at, is analyzing data patterns. The whole point of AI is that you input as much data as possible, and it's able to tell you what the patterns are in that data. So being able to evaluate which of your videos has done really well. How do you do that, Mary?
00:23:33
Speaker
That sounds complicated. There's different ways to do it. There's specific models that you can train that you can input all of your YouTube content. It'll give you the ways that these patterns in common are performing better, et cetera.
00:23:55
Speaker
Not exactly this at all. Should we be uploading all of our podcast episodes into this engine and ask what are our best and worst topics? This is why I'm friends with you guys because I love you both dearly and sincerely. You're both fucking nerds in the exact opposite way that I am. And the idea of analyzing data sets makes me want to commit Hari Kari. Like I have zero interest in any of this.
00:24:23
Speaker
Can I can I tell you something like like a like a something I did like deeply personal? Yeah, like this is not really days of Instagram when you can actually download data directly from the Instagram servers. Oh, Jesus. I used to download people's accounts you could download every post they ever posted through all of time.
00:24:43
Speaker
along with the comments, the likes, what it had in it, whether they use filters or not, you know, the number of saves, the number of shares. And I would do data analysis. I would download all this data on really big accounts. And then I would graph their posts over time.
00:24:58
Speaker
And then you'd see how many likes each post had as a function of time. And there would be like, you know, it would go up over time as they got more followers, but there were certain outliers that just spiked way above the norm. Like, you know, normally as the number of followers increase, the number of likes increase, right? But there were a few points that were just total extreme outliers. And I would collect those and I would try to understand what about those posts is getting traction. And I would use that to inform my own posting behavior.
00:25:29
Speaker
Okay, like this, genuine questions. Is that data analysis and science or is that the creative process? Because, and I'm not saying that they're mutually exclusive. Probably both, yeah. I'm not saying that they're mutually exclusive because the way that you describe that is like the analytical mind of a professional scientist. Whereas for me, when I'm trying to find an inspiration for a piece, I am looking at other people's work
00:25:57
Speaker
until I find the thing that speaks to the language that I'm trying to whatever the vocabulary is that I'm trying to use. And then I take that and I try to figure out what's working and I try to impart that in the piece I'm making. So it's the same process, but there's maybe less data.
00:26:13
Speaker
Yeah, maybe we come at it just from different angles, right? Once like a more analytical data driven angle, the others more artistic freewheeling angle. But it seems like a lot of the same creative process. Hmm. Is there anything we didn't touch on with our design process? So Mary, you talked about some text that we could use. So actually, you started to go into content creator. So Eric, you know, maybe you could speak to this. Have you used AI at all in your content creator, like the thumbnails or the text or anything like that? I have not.
00:26:42
Speaker
Because right now, I'll tell you exactly why I have chosen not to do it right now because it's it's the cool thing to do. Like it's hot right now. Everybody's doing it. And maybe it's the perpetual teenager in me that refuses to like hop on a trend and like try to buck things.
00:27:01
Speaker
There will come a point where I will find its use, but I haven't found its use for me, and I'm not willing to use it blindly just to use it. I will wait till the point where I figure out what it's doing well and how it can help what I'm attempting to do. So I've not used it just yet, but I do have a number of friends in the content game who have. Keith being one of them. So I'm curious...
00:27:30
Speaker
if you guys have any suggestions as to how it might be useful.
00:27:36
Speaker
Keith, we'd love you. Mary, have you used it? I, funnily enough, I use it a lot, not in my creative practice. I use it in my everyday life as almost a replacement for Google, which is kind of crazy to admit, but apparently this is the thing. Okay. I also read an article about how Gen Zers are now using TikTok as their Google search instead of actually Google, which is wild.
00:28:06
Speaker
But yeah, I've been using it in like my everyday work of just like how. So for example, if I'm trying to write like a slack message or an email or someone just I just like it takes me a while and I'm the first kind of person who agonizes over an email over and over again and like, make sure that it sounds right. So instead, I just like
00:28:27
Speaker
You know, spit my thoughts into chat GPT, make this form this into an email, make it like a few sentences long. And then if it sounds more casual or formal, et cetera, you can adjust it either way. It's just so much easier for me to do that. Are you serious? Chat GPT writes your emails for you? Oh yeah, absolutely. It's so much faster for me because then I'm not sitting there agonizing like over each sentence. I need to know something right now. Have you ever sent me a chat GPT generated texts?
00:28:57
Speaker
I will see your eyeballs. Do not lie to me right now, Mary. No, definitely not texts. No, that is not that is the next level that I would not. It's when I'm worried about coming off in a certain way or where I feel like I need to be a little bit more formal or professional. That's when I suddenly I doubt everything. Nothing is sacred anymore.
00:29:21
Speaker
everything she's ever said to us. I don't even know if it's what Nietzsche was talking about right now. Nothing is sacred. I just find it so much more useful than Google search. Like I can be so much more specific about what I'm looking for instead of having to scroll through like Google something look for like specific links, etc. It's just so much faster.
00:29:41
Speaker
Mary, I wish I had that. I don't think I'm programmed that way because I'm older. But I sort of wish I had more of that in my daily workflow. That's cool. Yeah, I just don't respond to those emails until I wait four days and then I'm like, ah, fuck off.
00:30:01
Speaker
Or you could be like Eric. So next topic. So how can we use AI to enable our sales? So I'll start by saying, you know, I took a pretty terrible picture of something that I wanted to stage more nicely, nicely. And I just didn't, my house is old fashioned. And the piece was very modern. So I didn't know what to do. So I actually masked it. And I asked one of my friends who's quite good, Mike Crotty. Hey, Mike.
00:30:28
Speaker
If Mike could, if he could like, take that image and like put it in new environments, and he did, and the piece sold using, you know, one of those pictures. So it was very, in that sense, it enabled the sale through the staging of the piece. Are there other other examples besides staging, how it can help enable sales?
AI in Sales and Marketing
00:30:50
Speaker
I mean, that would be probably the place where I would begin to implement it first more than anything. I don't know if there's a different way to do that, but just taking the object and placing it in a new environment definitely helps. But I do that in an analog sense. I just jump into Photoshop, and I start tinkering with it, and I put it on a white background, do whatever, add shadows. So using AI in order to do that
00:31:18
Speaker
You know, to, to put it on a back, I refuse to take gallery photos, like on a seamless. And so using AI to then jump in and put it in a gallery setting on a seamless, then I can submit that to galleries for applications. Super helpful. That could then generate sales. Yeah. I was going to say, I do the kind of opposite side, like if not opposite, but if I do take photograph on the seamless or I have older.
00:31:42
Speaker
pieces on seamlaces, it's just so easy to remove that background and put it in lots of different contexts. Sorry, what's this seamless? What is this rolled backdrop? So you don't see the like floor.
00:31:56
Speaker
Yeah, but I mean, like, one thing I mentioned in the class I taught WorkbenchCon is like you can kind of cater to your different audiences. So if you want to sell if you have a piece that you think would do really well in like a corporate environment, then you know, sell it to them using an office like a nice office photo or you want someone who's like super rich and
00:32:16
Speaker
has like a very nice, I don't know, contemporary style, like mimic their home in the photo. Yeah, like the environment, like you guys know that environment is huge for me. So being able to be specific on what who you're targeting your audiences and catering your image to them. I that's huge.
00:32:41
Speaker
I know VR is a thing that exists. I've seen that. I actually used it with the Emmy situation. There was a VR of an Emmy that I could like place it in the cap. It was different. But I'm wondering, is there a reality, I imagine there is in the near future where like I can upload a photo of this object to a VR app and then a client could potentially place it in their home to see where it exists and help enable that sale.
00:33:08
Speaker
Yeah, that already exists, I'm pretty sure. I'm sure, of course. Well, you know, what can you do? Yes, I'm sure. I don't know any specific programs, but one of the questions I was asked to me at WorkbenchCon was like, can you just generate like plans or a 3D model from like a photo, etc? It's like, yes, I'm pretty sure that is feasible. The talk is out there.
00:33:34
Speaker
So Mary, that comes up to the sushi board. So like one of the things it did on the sushi board was like this really florally, squirrely, spirally filigree made of metal. And I thought, oh my God, if you could give me a vector file after creating the image and convincing me, give me that vector file where I could put that on a CNC, or at least I could trace it and do it by hand, that would be like the next step.
00:33:59
Speaker
Go to vectorize. That's a, I think it's a website and it just vectorizes images for you, or actually the illustrator beta has a two, but I can't tell too much, too much Adobe here. Don't want to get Mary in trouble in the spirit of not getting Mary in trouble. I'd like to ask what Mary, what are some of the major ethical considerations to all of this amazing technology that we've been talking about?
00:34:24
Speaker
Oh my God, this is like an entire can of worms. Well, sorry, you only have about two or three minutes to summarize. Watch it now, girl. Here we go.
00:34:36
Speaker
Yeah, so it's kind of what you think. So there's a few different things, you know, like ownership, copyrighted information, and then all of these models, these generative tech models, they all have issues. They are trained. So the way that these models are created, they are trained on data and specifically data sets, each one has is trained on different ones, none of them are going to reveal exactly what they're trained on, obviously, for competitive purposes, etc.
00:35:06
Speaker
There is, you know, of course, there's going to be ethical issues there because you don't know exactly what's trained on. So do you feel comfortable using it yourself? Are you worried about, you know, any legal repercussions if you use it to generate information and data? I do not think that will be the case, like on an individual level.
00:35:27
Speaker
But yeah, it's exactly what you guys think about when you think of like ethics and AI it's also changing so quickly right now like copyright law and You know, I think the government is also trying to put limitations in place. I just everything is changing so so fast
00:35:46
Speaker
It's hard to it's hard to generalize because it's a moving target. Yeah. And and by the way, you say copyright law. That's only in the United States. Yeah, exactly. Yeah. And we have the internet, which is worldwide. So if AI engine pops up in another country, does that protect or not? Yeah, I see. Yeah.
00:36:04
Speaker
And copyright of art in general is so difficult. I haven't delved too much into this, but if I recall correctly, it is just an issue if it is an exact replica of the art piece. So if something has slightly changed, even think about when people post movies to YouTube, they raise the pitch of the audio a little bit, or they mirror it. That gets around all these copyright laws. So again, it's just...
00:36:32
Speaker
really tricky. And I don't know, it's it's hard to define. So it really comes down to like vanilla ice versus queen. Yep. Is it no, it's different. Is it?
00:36:49
Speaker
Totally different. All right. In the last few minutes, we have now that was a very, very brief treatment of the ethical considerations, obviously, right? But we're not an ethics panel on an AI.
00:37:06
Speaker
on an AI discussion. But we just want you to know that there are ethical considerations constantly being talked about with how these AI algorithms were trained in the first place. Where did the data come from? Was the data copyright? Was it not? So that's going to be a continuing story. We just wanted to just throw it out there as something you should be aware of, and you're going to hear a lot about it as time goes on, I'm sure.
00:37:32
Speaker
Yeah, I tried to like look up where each of these data or what each of these models are trained on. And it's so difficult to find because they're very generic. They also, there's like a lot of data data sets, like the common crawl data set that a lot of the models are trained on. But like, we know that mid journey and stable diffusion were trained on copyrighted images, like they have been investigated and discovered that to be true. However,
00:37:59
Speaker
Weirdly enough, my maybe go-to engines might be mid-journey and stable diffusion because neither of them store images because they work on a diffusion technique that essentially begins with random noise and refines it based on the neural network's understanding of the parameters you give it. Sorry, that's why I'm very technical.
00:38:19
Speaker
You lost me in neural network. Their eyes are glazing over. But yeah, it's just really tricky. And Adobe tries to be really ethical with only training on their stock footage, et cetera. So I don't know. It's just fascinating. Well, Mary, for what it's worth in all my reading, I do want to say Adobe has always been given the gold star for their ethical treatment of their AI engines in all my reading. And that has nothing to do with me knowing you and knowing you will work for Adobe. I've always noticed that.
00:38:48
Speaker
Kudos to you and your company. That is their selling point. It's supposed to be able to be used in corporate use, but who knows? Are there any other ways to use AI that we haven't touched on in the last couple
Machine Learning in Media Editing Software
00:39:03
Speaker
minutes? Is there any left field curveball things that we haven't considered that anyone could think of?
00:39:11
Speaker
So the only ones that we haven't really mentioned, we've talked about generative AI. We haven't really talked about machine learning AI that's in software. That is actually my forte. That is what I build into Premiere Pro and After Effects and utilizing machine learning and AI to enhance your actual workflow. So within my program, Premiere Pro and After Effects, it's like the editing software. For example, if you want to
00:39:40
Speaker
Like a concrete example is if you want to extend length of an audio clip, but it's already too short, but you really like it. Use AI to generate a longer version of it, and they can make those clips or those cuts automatically. It's seamless. It's called Remix, by the way, and from your pro. Oh my God. Really? You could do that? Yeah, absolutely. All you do is just drag the track shorter or longer and it automatically remixes it for you. God, that's amazing.
00:40:05
Speaker
So stuff like that is super, it's honestly so easy. It's been in Premiere Pro for a few years now, and this is before any of the generative stuff happens. But yeah, that's a whole other subset that I could dive really deep into, but will not.
00:40:20
Speaker
Holy shit, I didn't even know that was possible. I think a part of the my experience and probably the experience of a lot of our listeners is they don't even know what's possible with AI. Because unless you're on the bleeding edge hearing about this stuff all the day immersed in it like you are, Mary, I didn't even know that was on the table. I mean, it's just analyzing the patterns of a song and being able to intelligently recreate it so that it is extended.
00:40:49
Speaker
Holy shit. All right. Well, that brings us to the 40-minute mark. That's what we decided we would try to cap our AI discussion off at. And we will switch gears now. I just want to say that I'm very proud I left the dick jokes on the table when you said otherwise see you as AI, but that's fine. We can move on now.
00:41:11
Speaker
No jokes for Eric. Okay. So we come now to our first segment to mix it up because we'd like to have segments that aren't so heady. So our first segment that is a brand new segment, very exciting.
Humorous AI Insights on Names and Podcasting
00:41:24
Speaker
It's called whatever the fuck Paul wants. That's the segment. And today what Paul wants is to use AI to tell me what are the top five reasons that you should not name your child Eric.
00:41:41
Speaker
Okay, I'm in for the jugular right now
00:41:52
Speaker
Okay. All right. I just want to say hold on. Hold on. Did you spell it? Did you spell it wrong or right? That's all I want to know. I'm not answering. So I asked, I asked Chad GPT to give me a top. I used AI to give me a top five reasons why you should name your kid, Eric. And here's what's happened. Number one. Number one.
00:42:14
Speaker
confusion galore eric with the various spellings like eric that's what i was gonna say eric k eryk first of all hold on shut the fuck up nobody spells the eryk and if you do you can get the fuck out of my office okay eric is wrong eric you're just pranking people
00:42:36
Speaker
There's one correct way to spell Eric because by the way, it didn't even give your spelling. So yeah, because AI is confused about itself. So your child may be forever correcting people on the spelling of their name all the time, all the time. That is one thing that happens. My favorite example was when somebody asked me my name, I said, Eric Curtis, and they said, is that with a C or a K? And I said, with a K, thank you for asking. And then they spelled my name, E-R-I-C-K-U-R-T-I-S. And I was like, fuck you, man.
00:43:08
Speaker
Okay, number two, Eric, it's overused. Apparently Eric's been a popular name for generations, so your child might find themselves in a sea of Eric's, making it harder to stand out. It's a great name. That's just a case for most names. Mine is more common.
00:43:29
Speaker
Number three, mispronunciation madness, despite its seemingly simple spellings. Wait, wait, despite its seemingly simple spelling, Eric is often mispronounced as I, Rick. Nope. No, it is not. Not once in my life has anybody ever been like, I, Rick, Kurtis, or ear, Rick.
00:43:54
Speaker
leading to endless corrections. I like Eric. I mean, I get EROC from time to time, but that's not a mispronunciation. Oh, God.
00:44:04
Speaker
Number four, historical baggage, Eric. While Eric may have some historical significance, it's also associated with less savory figures like Eric the Red or infamous Eric's from popular culture, which might not be the best role models. I like, I like how they tried to frame that as though there are a lot of bad Eric's and they came up with one example and they were like, among others. Wait, am I supposed to know who Eric the Red is? He's a Viking, right? Yeah. He murdered a lot of people, but that's fine.
00:44:35
Speaker
See, this is why you never trust an Eric. And number five, number five, pop culture pitfalls, Eric, with fictional characters like Eric from Fenver the Opera, or Eric Killmonger from Black Panther. Your child might constantly be compared to these characters for better or worse. Yeah, I constantly compare Eric to the damn opera.
00:44:58
Speaker
The only two villain Eric's to ever exist? You know what they didn't name? Prince Eric, what are we doing here?
00:45:07
Speaker
This is horseshit. Okay, but Prince Eric was also very weird in which he saw- Oh, I'm not saying that he was not a villain. I'm just saying he wasn't the villain of that story. He's definitely the villain of a different story. Okay, saw a naked girl and was like, she has no voice. Let me take it. Yeah, right. A naked girl who couldn't articulate what she wanted and he was like, bet, hold my beer. Yeah. It was not great. It was not great.
00:45:32
Speaker
Although, to be fair Paul, I feel like if you had asked ChatGPT the same thing about my name, it would have- I did. Oh, you did? Go on, buddy. I'm sure there's a lot of- But Eric, Eric, Eric, Eric, you got it. Because we love Mary, I asked ChatGPT the reasons you should name your daughter Mary. You son of a bitch. Handering the audience.
00:45:57
Speaker
because we need to take care of Mary. Number one, timeless elegance. Mary's a classic name that has stood the test of time, exuding elegance and grace no matter the era. So we're just going to leave Mary Magdalene on the table right there, huh? Oh, actually that was actually listed. That was actually listed. I'm not kidding. Number four, reasons to name your daughter Mary. Role models galore with historical figures like Mary Magdalene.
00:46:23
Speaker
Mary Queen of Scots, who by the way was Mary Queen of Scots, who was like put to death at age 24. And modern icons like Mary Shelley, who wrote Frankenstein. That's awesome. True. Well, I will give Mary Shelley. Your daughter inherits a legacy and a strong and influential woman. They opened with Mary Magdalene? That's an interesting one. All right.
00:46:47
Speaker
Okay, but you know, I did that because you know, we got to be nice to Mary, but actually, I know a fun story of how I got my name, I think. So I was born and born three months premature, by the way. And my parents are from Taiwan, they have pretty strong accents. So my name was supposed to be Marilyn named after a family friend. And their accents were so strong that the nurses were just like,
00:47:16
Speaker
Oh my gosh. Yeah, sure. That's unbelievable. It's funny every time. Yeah. And they're like, yeah, I mean, I guess it's easier to say. Well, sorry. So last but not least, before we move on from this segment, I did want to ask it real quick. What are the top 10 concise reasons that listening to podcasts is stupid?
00:47:45
Speaker
Okay, okay. Concise. Yeah, I feel like you stressed that word. Oh, yeah, because they were too long. I was like, I can't be bothered reading all this horseshit. So the top 10 reasons that you shouldn't bother listening to podcasts. Number one, boring content. That's that's definitely fair. I gotta be honest, I have there's very few I find compelling.
00:48:11
Speaker
Number two, low production quality. Hold on. Do you guys hear the sweet, sweet bass in these voices right now? I was going to say it's the first episode where we upgraded our mics. The investment in this pot for your ears. So I want to give a collective apology to the listening public because my two cohorts had bullshit mics until today.
00:48:36
Speaker
Excuse me. Sir, sir, we had the ability to record sound. That was good enough for the beginning of the podcast. Eric, don't even pretend for a moment that you sounded good. So listen, we all listened to our previous episodes. I know many of you have been like, oh, sound quality. Okay. We heard you. We heard you and Mary and Eric got upgraded mics for this episode. So hopefully you hear the difference. Can you hear the difference? Mary soundcheck.
00:49:06
Speaker
Sashi mi. I'm going to move on.
00:49:13
Speaker
What's number four? Number three, there were repetitive topics. And you know what? Someone actually, I'm not going to call him out. But he's like, oh, you may have touched on that topic once before. And I was like, yeah, that's how learning happens. We're doing this for ourselves as much as we're doing it for all of you. We're trying to learn about design by covering the things we love to talk about. And so yeah.
00:49:41
Speaker
That means we cover a topic more than once. We might talk about related things about creativity. I think repetitive topics is wonderful because it encodes. That's why I listen to every episode like three or four times to like really encode. And actually, Eric, I hate to pad your ego, but a lot of the things you say I didn't pick up in the moment, but upon re-listening to it the second and the third time, I'm like, oh, that's what he was getting at. And I hate to admit it, but
00:50:07
Speaker
Those are really deep, thoughtful points that I just completely fucking missed because I'm more dense in the moment. I mean, you're welcome, buddy. Thank you.
00:50:19
Speaker
Number four, they can be time consuming. I think we limit ours to a reasonable amount of time. Sure. Number five, inaccurate information. Well, maybe, but I think we tried to be fairly accurate. At least I know that when I come with reports or studies, I definitely read them before the episode and I cite the source. And Mary, I know everything you've said today was accurate.
00:50:44
Speaker
Or not. Maybe not. Lying the toilet. Apparently, I'm giving Mary more credit than she deserves. Number six, excessive advertising. That's not the case. Screw off. Number seven, biased or opinionated content. Isn't that the point of a podcast? Isn't that just content? Whenever you're recording live anything, you're going to give your opinion, which is biased.
00:51:09
Speaker
Number eight, overly long episodes. Hmm. Maybe ours isn't. And that's where we cut off the top 10 ideas from AI. No. Nine is a lack of substance. That's totally Eric.
00:51:32
Speaker
The emotional roller coaster I'm going on this I'm just kidding and number ten. It's a distraction from real-life Responsibilities because you're all listening to podcasts all day. You're not paying your bills. You're not feeding your families This is AI who has neither family nor bills saying
00:51:54
Speaker
All right, anyway, that's the top 10 concise reasons that listening to a podcast is stupid. Apparently ours doesn't qualify.
00:52:03
Speaker
All right, so with that, we're at the 52 minutes. Not biased at all. Not biased at all. So with that, at 52 minutes, I'd like to just briefly thank our patrons in this episode. So we have Travis Pond and Bill Berkel from WTB Woodworking, Austin Santiago, Eric Gorgeous, who I know from Craftsman Legacy. I think many of you saw the show. Most gorgeous man I know. Madam at Work French Con.
00:52:26
Speaker
He's a great guy. He eats and breathes and sleeps making. He's an inspiration. And then we have James at Dirty Meerkat Woodshop. So thank you, the five of you, for supporting our show. We appreciate every one of you, and I often send personal messages to you. If I haven't, don't take it personally. I just forgot. All right. So that concludes our episode. Now we're going to move on to
00:52:53
Speaker
the aftershow. And you know, if you're a patron, you can have our aftershow as well as the video feeds, which you get to stare at Eric and Mary looking all hot and cute. And so, uh, Mary. Yeah, I was about to say that. Which one's the hot, which one's the cute? No, both, both are hot and cute. They're not mutually exclusive. Those are diametrically opposed adjectives.
00:53:20
Speaker
Oh, God, what did I get myself into? I'm not answering. So in the after show, we're going to answer a another AI generated list. But it's very apropos, which is what are the top 10 things woodworkers are bad at?
00:53:42
Speaker
Hmm, that's a good one. I asked after show, huh? So before we go to the after show and I'll tell you what I said that we're bad at. You guys have any guesses? Joinery. Proportions, wood selection, grain selection. I got lots of guesses. What are we doing? I mean, it's a very, very vague question. It could be everything, basically.
00:54:11
Speaker
Mary, what are you bad at? Maybe it's on there. Oh, are we talking about specifically ourselves? Woodworkers. I don't know. You're a woodworker, right? Yeah, that's true. Uh, I don't know. Marketing, designing your family. Well.
00:54:29
Speaker
If you want the answer, you're going to have to tune in the after show and we will divulge all 10 and we'll take ourselves to task. In the meantime, thank you all. I hope you found today's episode illuminating. Thank you, Mary, again, for your insight. We knew the minute we started this podcast, we knew we would cover AI because Mary is such an authority on the area. Mary, I know you worry about
00:54:53
Speaker
the pressure of being in authority, but you are. And so we really appreciate your perspective. And even if you didn't give an answer, you know, like definitively one way or the other, that is an answer, which is, I don't know yet, or we don't know yet. We're going to have to see how it shakes out.
00:55:08
Speaker
That kind of, you know, balanced perspective is definitely appreciated by all of us. And we hope that some of the ideas around how to use AI, both like chat GPT or image generation or video production, you know, can be useful in your daily lives as you try to create things. I know for me, I definitely will try to make one of the sushi board designs that I gave at some point. And I'll definitely, we'll see how it goes.
00:55:34
Speaker
With that, friends, thank you for tuning into another episode of Woodworking is Bullshit. We'll see you later. Thanks, friendos. Thank you. Make good decisions. Or bad. Bye. Say hi to your moms for me.