Jekyll2024-03-02T03:48:25-08:00/feed.xmlLaurence Moroney - The AI Guy.This is the blog and general web site of Laurence Moroney, author, lecturer, teacher and AI lead at Google with Musings on AI, Quantum and other future tech
Laurence MoroneyWatching Dune in the world’s must luxurious theater2024-03-01T00:00:00-08:002024-03-01T00:00:00-08:00/2024/03/01/dune<h2 id="watching-dune-in-the-worlds-most-luxurious-theater">Watching Dune in the world’s most luxurious theater</h2>
<p>So, I was in Riyadh this week, speaking at the <a href="https://humancapabilityinitiative.org/en/">Human Capability Initiative</a>, a platform and a conference to focus on human capability, and how we can leverage social, economic, technological and other forces to build a better world for the next generation. I met and spoke alongside many fascinating people!</p>
<p>Anyway, at the end of the conference, I had a little time. I thought being there would mean I would miss seeing Dune Part 2 on its opening weekend, but I was so lucky to be staying in a hotel right beside the Via Riyadh movie theater. Often called the world’s most luxurious, it certainly lived up to the billing. But better – they were showing Dune 2! It’s not cheap ~$50 a ticket. But it’s SO WORTH IT. If you’re ever in Riyadh, and there’s a good movie showing – I STRONGLY recommend it.</p>
<p>It’s all kind of unassuming at the beginning. From the street side, you just see this:
<img src="/assets/dune1.jpeg" alt="Movie theater sign from outside" /></p>
<p>You go inside, and there isn’t a lot of fanfare, just an elevator taking you down. I figured I’d go to reception…
<img src="/assets/dune2.jpeg" alt="Elevator inside the lobby" /></p>
<p>After getting my QR code scanned, I was guided down to a marble floored bar area that acted as reception. Not the usual just walk into the theater.
<img src="/assets/dune3.jpeg" alt="Movie theater reception" /></p>
<p>A waiter greeted me and gave me the menu. He also gave me a welcome cocktail. They must be prescient, because for some bizarre reason I like blue drinks. Being Riyadh, this was non-alcoholic. Unfortunately, while it looks a little like dish-soap, it didn’t taste that way. It was a delicious start to the evening!
<img src="/assets/dune4.jpeg" alt="Movie theater reception" /></p>
<p>Food isn’t your popcorn and boxed candy. They offer a full menu, with everything from arabic food, to sushi, to…popcorn! I didn’t capture the prices in these diagrams, but remember thinking it was <em>quite</em> reasonable by Riyadh standards. For exmaple the cocktails came in at 65 Ryals ($16) which about the same price as the hotel bar.</p>
<p><img src="/assets/dune5.jpeg" alt="Food Menu" />
<img src="/assets/dune6.jpeg" alt="Drinks Menu" /></p>
<p>At this point a waiter guided me through the menu and offered me a tour. He mentioned that I could order at any time, even during the movie, because he would be in the theater, standing behind my seats, at my beck and call!</p>
<p>I couldn’t say ‘no’ to that, and he took me through each of the theaters to see them. They’re all on different themes like ‘Orient Express,’ or ‘Wall Street’, and crazy luxurious. Here’s a few:</p>
<p>Watch a movie from a queen-sized bed!
<img src="/assets/dune8.jpeg" alt="Queen Sized Bed Seats" /></p>
<p>A standard 8-seat theater with extra seats if you get bored in your recliner:
<img src="/assets/dune9.jpeg" alt="Standard theater" /></p>
<p>Really comfy ‘bean bag’ room for private viewings:
<img src="/assets/dune10.jpeg" alt="Bean bag room" /></p>
<p>It even had a ball pit in case you bring your kids!
<img src="/assets/dune11.jpeg" alt="Ball pit" /></p>
<p>My favorite room was this one – following an Arabian/Bedouin theme. It was beautiful!
<img src="/assets/dune12.jpeg" alt="Arabic room" /></p>
<p>The ‘Wall Street’ themed room was similar. I couldn’t see the Wall Street connection here, but it was still gorgeous!
<img src="/assets/dune13.jpeg" alt="Wall Street Room" /></p>
<p>One room had a ‘Starlight’ theme. There were people watching trailers before a movie in here, but they didn’t mind me coming in to see. The sound system is excellent, and VERY LOUD, so I don’t think we disturbed them too much!
<img src="/assets/dune14.jpeg" alt="Starlight room" /></p>
<p>The ceiling of the room imitated stars, and when the lights are down, I’m sure this is spectacular!
<img src="/assets/dune15.jpeg" alt="Starlight ceiling" /></p>
<p>Finally, I got to my theater just as the last trailer was playing – the Godzilla versus Kong movie, which I really want to see! I was seated in a comfy recliner, given a blanket and a pillow, and within a couple of minutes the movie began. I didn’t want to photograph any of the movie for copyright reasons, but I had to snap a picture of this seat to my left, and I was sorely tempted to use it…but so engrossed in the film, that I forgot!
<img src="/assets/dune16.jpeg" alt="Comfy chair" /></p>
<p>I didn’t take advantage of the food ordering, but the waiter still came by from time to time, giving wet towels, and water bottles. At one point I sneezed, and he brought me tissues!</p>
<p>And just to prove that I did see the movie, and to show the size of the screen, I took this shot in the closing credits!
<img src="/assets/dune17.jpeg" alt="Dune credits" /></p>
<p>What you don’t see in the pictures is the sound. I’ve never experienced anything with this level of volume, while still maintaining crystal clear audio. Dune Part 2 is a LOUD movie with lots of loud bombastic music and sound effects. It’s a gourmet feast for the ears as well as the eyes. The screen, as you can see above, is large, but not gigantic. It’s LED, not projection, and sometimes if you look closely, you can see the seams between the elements of the screen. But with a great movie like Dune, you soon forget.</p>
<p>And as for the movie itself – I loved it. I’m a huge Dune fan, so a lot of the deep cuts in the movie, and a lot of the time they spend on details like the Geidi Prime scenes were an absolute feast. I worry that people who aren’t huge fans of the source material might feel that it drags a bit. For the most part I loved the cast. Stilgar is brilliant. Zandaya as Chani steals the show. I liked Florence Pugh’s Irulan, but would have liked to see more of her.</p>
<p>There were a couple of moments in the movie where I think it might have been cut differently for Saudi Arabian viewers. I’m curious if anyone has the details. One such is at the beginning when Jessica drinks something, starts having a seizure, and then….we are outside in the desert, and the subtitles went a little out of synch. Another is when a big deal is made out of paul needing to prove himself by crossing the desert alone. He begins, is joined by Chani, and they do lovely choreagraphed steps together…and then we’re into an attack on Harkonnen spice miners and the scene is forgotten. Anybody have insight on this?</p>
<p>Any flaws in the movie are, I think, because it still feels incomplete. This was made a little more open-ended than the book, almost as if Villeneuve wanted to do ‘Dune Messiah’ as part 3. And the sheer scope of the story is worth it. I hope the movie is a huge success and we get part 3.</p>
<p>MINOR SPOILER FOLLOWS</p>
<p>One such is the pretty significant changes to Alia. I didn’t think they would work, having read about it going in, but it did. I really like this direction. Now, thinking back to 1984’s version by David Lynch, the toddler walking around dressed as a Reverend Mother looks silly. Aging Alia up in the 2000 miniseries worked better, but Villeneuve’s approach, which I won’t spoil here, is my favorite. The flash-forward scene to post-birth, teenage Alia, is all at once creepy and inspiring. I loved it.</p>Laurence MoroneyWatching Dune in the world’s most luxurious theater So, I was in Riyadh this week, speaking at the Human Capability Initiative, a platform and a conference to focus on human capability, and how we can leverage social, economic, technological and other forces to build a better world for the next generation. I met and spoke alongside many fascinating people! Anyway, at the end of the conference, I had a little time. I thought being there would mean I would miss seeing Dune Part 2 on its opening weekend, but I was so lucky to be staying in a hotel right beside the Via Riyadh movie theater. Often called the world’s most luxurious, it certainly lived up to the billing. But better – they were showing Dune 2! It’s not cheap ~$50 a ticket. But it’s SO WORTH IT. If you’re ever in Riyadh, and there’s a good movie showing – I STRONGLY recommend it. It’s all kind of unassuming at the beginning. From the street side, you just see this: You go inside, and there isn’t a lot of fanfare, just an elevator taking you down. I figured I’d go to reception… After getting my QR code scanned, I was guided down to a marble floored bar area that acted as reception. Not the usual just walk into the theater. A waiter greeted me and gave me the menu. He also gave me a welcome cocktail. They must be prescient, because for some bizarre reason I like blue drinks. Being Riyadh, this was non-alcoholic. Unfortunately, while it looks a little like dish-soap, it didn’t taste that way. It was a delicious start to the evening! Food isn’t your popcorn and boxed candy. They offer a full menu, with everything from arabic food, to sushi, to…popcorn! I didn’t capture the prices in these diagrams, but remember thinking it was quite reasonable by Riyadh standards. For exmaple the cocktails came in at 65 Ryals ($16) which about the same price as the hotel bar. At this point a waiter guided me through the menu and offered me a tour. He mentioned that I could order at any time, even during the movie, because he would be in the theater, standing behind my seats, at my beck and call! I couldn’t say ‘no’ to that, and he took me through each of the theaters to see them. They’re all on different themes like ‘Orient Express,’ or ‘Wall Street’, and crazy luxurious. Here’s a few: Watch a movie from a queen-sized bed! A standard 8-seat theater with extra seats if you get bored in your recliner: Really comfy ‘bean bag’ room for private viewings: It even had a ball pit in case you bring your kids! My favorite room was this one – following an Arabian/Bedouin theme. It was beautiful! The ‘Wall Street’ themed room was similar. I couldn’t see the Wall Street connection here, but it was still gorgeous! One room had a ‘Starlight’ theme. There were people watching trailers before a movie in here, but they didn’t mind me coming in to see. The sound system is excellent, and VERY LOUD, so I don’t think we disturbed them too much! The ceiling of the room imitated stars, and when the lights are down, I’m sure this is spectacular! Finally, I got to my theater just as the last trailer was playing – the Godzilla versus Kong movie, which I really want to see! I was seated in a comfy recliner, given a blanket and a pillow, and within a couple of minutes the movie began. I didn’t want to photograph any of the movie for copyright reasons, but I had to snap a picture of this seat to my left, and I was sorely tempted to use it…but so engrossed in the film, that I forgot! I didn’t take advantage of the food ordering, but the waiter still came by from time to time, giving wet towels, and water bottles. At one point I sneezed, and he brought me tissues! And just to prove that I did see the movie, and to show the size of the screen, I took this shot in the closing credits! What you don’t see in the pictures is the sound. I’ve never experienced anything with this level of volume, while still maintaining crystal clear audio. Dune Part 2 is a LOUD movie with lots of loud bombastic music and sound effects. It’s a gourmet feast for the ears as well as the eyes. The screen, as you can see above, is large, but not gigantic. It’s LED, not projection, and sometimes if you look closely, you can see the seams between the elements of the screen. But with a great movie like Dune, you soon forget. And as for the movie itself – I loved it. I’m a huge Dune fan, so a lot of the deep cuts in the movie, and a lot of the time they spend on details like the Geidi Prime scenes were an absolute feast. I worry that people who aren’t huge fans of the source material might feel that it drags a bit. For the most part I loved the cast. Stilgar is brilliant. Zandaya as Chani steals the show. I liked Florence Pugh’s Irulan, but would have liked to see more of her. There were a couple of moments in the movie where I think it might have been cut differently for Saudi Arabian viewers. I’m curious if anyone has the details. One such is at the beginning when Jessica drinks something, starts having a seizure, and then….we are outside in the desert, and the subtitles went a little out of synch. Another is when a big deal is made out of paul needing to prove himself by crossing the desert alone. He begins, is joined by Chani, and they do lovely choreagraphed steps together…and then we’re into an attack on Harkonnen spice miners and the scene is forgotten. Anybody have insight on this? Any flaws in the movie are, I think, because it still feels incomplete. This was made a little more open-ended than the book, almost as if Villeneuve wanted to do ‘Dune Messiah’ as part 3. And the sheer scope of the story is worth it. I hope the movie is a huge success and we get part 3. MINOR SPOILER FOLLOWS One such is the pretty significant changes to Alia. I didn’t think they would work, having read about it going in, but it did. I really like this direction. Now, thinking back to 1984’s version by David Lynch, the toddler walking around dressed as a Reverend Mother looks silly. Aging Alia up in the 2000 miniseries worked better, but Villeneuve’s approach, which I won’t spoil here, is my favorite. The flash-forward scene to post-birth, teenage Alia, is all at once creepy and inspiring. I loved it.Making a movie with AI tools - Part One2024-01-31T00:00:00-08:002024-01-31T00:00:00-08:00/2024/01/31/movie1<h2 id="a-case-study-in-making-a-movie-with-ai-tools">A case study in making a movie with AI tools</h2>
<p>You’ve probably seen lots of short movies being made using AI tools, and you’ve wondered how they’re done. Well, here’s an example of one, and I’m going to document the process of how it was made, and while going through that, also highlight the <em>limitations</em> of the technology that lead to many of these movies looking the same.</p>
<p>Here’s the movie</p>
<div><div class="extensions extensions--video">
<iframe src="https://www.youtube.com/embed/58HE9aTrfx0?rel=0&showinfo=0" frameborder="0" scrolling="no" allowfullscreen=""></iframe>
</div>
</div>
<p>It’s kinda cool, and it tells a story. But once you’ve seen a few movies like this, you’ll begin to see similar stuff in them, and where the moviemaker is working around the challenges.</p>
<p>The big ones are:</p>
<ul>
<li>Coherence: When using AI generation to create things, getting the same things in multiple pictures is difficult. Notice how the city and spaceship changes, for example.</li>
<li>Clip length: Generally services that create AI video will give you 3-4 seconds of video. Go longer, and you’ll see them rapidly degenerate. So, your ‘movie’ will be a sequence of 3-4 second clips.</li>
<li>Character interaction: You’ll often see single characters in poses with the camera moving around them. That’s what current generation technology is effectively limited to. You might have people facing each other down, but you won’t see a lot of having characters interact</li>
<li>Dialog: When using AI generated video, you can’t yet synch it with dialog. Not yet anyway!</li>
</ul>
<p>With that in mind, and knowing these restrictions, I wanted to create a short movie.</p>
<p>Thus – I needed to string several scenes together and make a story out of them. One trick I could use is multiple shots of the same person with a similar prompt description, to embue a sense of drama. I chose to have a girl in a sci fi setting. There’s always something about looking at something huge from a distance that enthralls me.</p>
<h3 id="scene-1-a-melancholy-futuristic-moment">Scene 1: A melancholy futuristic moment</h3>
<p>So, I started with midjourney, and this prompt:</p>
<div class="language-plaintext highlighter-rouge"><div class="highlight"><pre class="highlight"><code>a long shot of a young asian woman standing on a cliff overlooking a sci fi city far below.
Sunny day. Science Fiction planet. She wears a black and red sci fi uniform in the style of modern anime,
realistic rendering of the human form, action and sci fi oriented, --ar 7:3 --niji 6
</code></pre></div></div>
<p>And it gave me this set of images:
<img src="/assets/midj1.png" alt="First scene" /></p>
<p>They’re pretty good, but not quite what I was going for, although the one on the lower left is excellent. So I tried again, tweaking the prompt slightly:</p>
<div class="language-plaintext highlighter-rouge"><div class="highlight"><pre class="highlight"><code>a medium shot of a young asian woman standing on a cliff overlooking a sci fi city far below.
Sunny day. Science Fiction planet. She wears a black and red sci fi uniform in the style of modern anime,
realistic rendering of the human form, action and sci fi oriented, --ar 7:3 --niji 6
</code></pre></div></div>
<p>Pretty much just asking for a medium shot:
<img src="/assets/midj2.png" alt="Second attempt at first scene" /></p>
<p>I really like the one on the upper right. It’s more a long shot, and less a medium shot, but the idea of someone looking from a distance at a far off place had the sense of melancholy and futurism I was looking for. Perfect!</p>
<h3 id="scene-2-introducing-our-main-character">Scene 2: Introducing our main character</h3>
<p>Ok, now we’ve seen her from the back, let’s take a look at our main character’s face. Now we should be closer, and her face – and any emotion present on it, should be the focus of the next shot. So I tried this:</p>
<div class="language-plaintext highlighter-rouge"><div class="highlight"><pre class="highlight"><code>a full body shot of a young asian woman wearing a black and red sci fi uniform in the style of modern anime,
sunny day, blue skies behind, realistic rendering of the human form, action and sci fi oriented, --ar 7:3 --niji 6
</code></pre></div></div>
<p>I got some really nice results back right away:
<img src="/assets/midj3.png" alt="First set of results for full body shot" /></p>
<p>I really like the drama of the second and third images. There’s something about the expression on her face. Originally I wanted a pure anime vibe, so the first image was interesting, but after a few attempts on step 3 to see if I could get more of this character, it failed to give me something that satisfied me! So I went back to the more CGI look and feel of images 2 and 3, ultimately deciding on the third one, because I like how the low angle shot embues a lot of emotion.</p>
<p>Now that I have the first intro to my MC, I wanted another shot of her, a little closer, and maybe lit a little differently to give a sense of rising drama.</p>
<h3 id="scene-3-getting-deeper-into-our-mcs-feelings">Scene 3: Getting deeper into our MC’s feelings</h3>
<p>Here’s where one can hit coherence challenges. When asking the engine to render ‘a young asian woman,’ there’s a lot of variety. If I want another shot of the same girl, it’s going to be tough. But I had a theory I wanted to follow, and it was pretty much brute force. Keep going until I find one that’s close, and then use ‘variations’ to get closer. So I edited the former prompt a little – asking for a close-up, as opposed to a full body shot, like this:</p>
<div class="language-plaintext highlighter-rouge"><div class="highlight"><pre class="highlight"><code>a close up of a young asian woman wearing a black and red sci fi uniform in the style of modern anime,
sunny day, blue skies behind, realistic rendering of the human form, action and sci fi oriented, --ar 7:3 --niji 6
</code></pre></div></div>
<p><img src="/assets/midj4.png" alt="A closeup on my MC" /></p>
<p>I’m sure you’ll agree that the results are beautiful. My favorite is #3, it’s such a gorgeous shot, but it doesn’t fit my needs. The fourth one is also stunning, and reminds me somewhat of a younger version of <a href="https://thetvdb.com/series/the-expanse/people/65335768">Florence Faivre, of ‘The Expanse’ fame</a>. Not a bad thing!</p>
<p>But ultimately #1 was closest in look and feel to my original image, and using the ‘variations’ control in Midjourney, I thought I’d reroll to see if I can can another look and feel that I’d prefer:</p>
<p><img src="/assets/midj5.png" alt="Using variations to improve on my second shot of my MC" /></p>
<p>These were all terrific, but I ultimately settled on the fourth image, as I liked the lighting on the side of her face. The third wasn’t bad either, but from experience, I found that when turning images into video, having more hair visible makes it a bit more dramatic, so that was the tiebreaker!</p>
<h3 id="scene-4-adding-drama">Scene 4: Adding Drama</h3>
<p>Ok, now I have a couple of shots of my main character, and she’s somewhat coherent. I’ve gotten over one of the main problems with this stuff. But now, as well as poses, I need a story. I had the idea for this story to take a turn to make it interesting. Instead of her lamenting the destruction of the civilization we saw in the first shot, wouldn’t it be nice if she was celebrating it instead? So for that, I wanted two more shots. One, a closeup of her eye looking at the destruction, and the other a variation of the shot from scene 3 with the dramatic lighting of flames etc. So, I’ll start with the latter.</p>
<p>With MidJourney, you can start from an existing image by providing its URL, which you can get by opening the image full screen, right clicking, and selecting ‘Copy Image Address’. Enter it at the beginning of the prompt, and then enhance that with your desired updates.</p>
<p>So I started with the closeup from the previous shot, and gave the URL of that image in the prompt, like this:</p>
<div class="language-plaintext highlighter-rouge"><div class="highlight"><pre class="highlight"><code><https://s.mj.run/Z5fF0z68gkc> bathed in the orange glow of flames, an evil smile on her face --ar 7:3 --niji 6
</code></pre></div></div>
<p>And I got these results:
<img src="/assets/midj6.png" alt="Relighting my character" /></p>
<p>While they’re a little more anime, and a little less CGI than the previous shot, they all look great. #3 is perhaps a little too evil! I liked #4 because, again, the eyes tell an interesting story, and kinda reflect the twist I wanted at the end.</p>
<p>Finally, I needed a closeup on her eyes. This one was trickier than I expected, and I had to reroll and go through variations a number of times until I got something that told the story I wanted:</p>
<div class="language-plaintext highlighter-rouge"><div class="highlight"><pre class="highlight"><code>an extreme close up of the eyes a young asian woman wearing a black and red sci fi uniform in the style of modern CGI anime,
sunny day, blue skies behind, reflection of a sci fi city in her eyes, action and sci fi oriented, --ar 7:3 --niji 6
</code></pre></div></div>
<p>I particularly like that even through I didn’t ask for the city to be in flames, it gave me one!</p>
<p><img src="/assets/midj7.png" alt="My character's eye in closeup" /></p>
<h3 id="scenes-5-and-onwards-the-alien-invasion">Scenes 5 and onwards, the alien invasion</h3>
<p>This part was relatively easy because it didn’t involve a human character. We can be more forgiving if the spaceships and city looks different, and I can write around that – talking about the destruction of an entire civilization, so diversity in image is a strength, so I got a bunch of prompts like these, to create the scenes:</p>
<p>Before the attack</p>
<div class="language-plaintext highlighter-rouge"><div class="highlight"><pre class="highlight"><code>a high tech sci fi city in the style of modern CGI anime, sunny day, blue skies,
a dark, ugly, spaceship hovering over it, contrasting the beauty of the city.
Action and Sci-Fi oriented. --ar 7:3 --niji 6
</code></pre></div></div>
<p>After the attack</p>
<div class="language-plaintext highlighter-rouge"><div class="highlight"><pre class="highlight"><code>a high tech sci fi city burning. Flames everywhere torching each of the buildings.
In the style of modern CGI anime, sunny day, blue skies.
Action and Sci-Fi oriented. --ar 7:3 --niji 6
</code></pre></div></div>
<p>From the variations on these, I was able to grab a few frames. I now had a number of frames that storyboarded my movie. The next part would be to animate them, and I’ll talk about that in the next post. But first, here’s how the 8 frames tied together into a storyboard, and we can begin to see a story emerge from them:</p>
<p><img src="/assets/scene.gif" alt="Storyboarded movie" /></p>Laurence MoroneyA case study in making a movie with AI tools You’ve probably seen lots of short movies being made using AI tools, and you’ve wondered how they’re done. Well, here’s an example of one, and I’m going to document the process of how it was made, and while going through that, also highlight the limitations of the technology that lead to many of these movies looking the same. Here’s the movie It’s kinda cool, and it tells a story. But once you’ve seen a few movies like this, you’ll begin to see similar stuff in them, and where the moviemaker is working around the challenges. The big ones are: Coherence: When using AI generation to create things, getting the same things in multiple pictures is difficult. Notice how the city and spaceship changes, for example. Clip length: Generally services that create AI video will give you 3-4 seconds of video. Go longer, and you’ll see them rapidly degenerate. So, your ‘movie’ will be a sequence of 3-4 second clips. Character interaction: You’ll often see single characters in poses with the camera moving around them. That’s what current generation technology is effectively limited to. You might have people facing each other down, but you won’t see a lot of having characters interact Dialog: When using AI generated video, you can’t yet synch it with dialog. Not yet anyway! With that in mind, and knowing these restrictions, I wanted to create a short movie. Thus – I needed to string several scenes together and make a story out of them. One trick I could use is multiple shots of the same person with a similar prompt description, to embue a sense of drama. I chose to have a girl in a sci fi setting. There’s always something about looking at something huge from a distance that enthralls me. Scene 1: A melancholy futuristic moment So, I started with midjourney, and this prompt: a long shot of a young asian woman standing on a cliff overlooking a sci fi city far below. Sunny day. Science Fiction planet. She wears a black and red sci fi uniform in the style of modern anime, realistic rendering of the human form, action and sci fi oriented, --ar 7:3 --niji 6 And it gave me this set of images: They’re pretty good, but not quite what I was going for, although the one on the lower left is excellent. So I tried again, tweaking the prompt slightly: a medium shot of a young asian woman standing on a cliff overlooking a sci fi city far below. Sunny day. Science Fiction planet. She wears a black and red sci fi uniform in the style of modern anime, realistic rendering of the human form, action and sci fi oriented, --ar 7:3 --niji 6 Pretty much just asking for a medium shot: I really like the one on the upper right. It’s more a long shot, and less a medium shot, but the idea of someone looking from a distance at a far off place had the sense of melancholy and futurism I was looking for. Perfect! Scene 2: Introducing our main character Ok, now we’ve seen her from the back, let’s take a look at our main character’s face. Now we should be closer, and her face – and any emotion present on it, should be the focus of the next shot. So I tried this: a full body shot of a young asian woman wearing a black and red sci fi uniform in the style of modern anime, sunny day, blue skies behind, realistic rendering of the human form, action and sci fi oriented, --ar 7:3 --niji 6 I got some really nice results back right away: I really like the drama of the second and third images. There’s something about the expression on her face. Originally I wanted a pure anime vibe, so the first image was interesting, but after a few attempts on step 3 to see if I could get more of this character, it failed to give me something that satisfied me! So I went back to the more CGI look and feel of images 2 and 3, ultimately deciding on the third one, because I like how the low angle shot embues a lot of emotion. Now that I have the first intro to my MC, I wanted another shot of her, a little closer, and maybe lit a little differently to give a sense of rising drama. Scene 3: Getting deeper into our MC’s feelings Here’s where one can hit coherence challenges. When asking the engine to render ‘a young asian woman,’ there’s a lot of variety. If I want another shot of the same girl, it’s going to be tough. But I had a theory I wanted to follow, and it was pretty much brute force. Keep going until I find one that’s close, and then use ‘variations’ to get closer. So I edited the former prompt a little – asking for a close-up, as opposed to a full body shot, like this: a close up of a young asian woman wearing a black and red sci fi uniform in the style of modern anime, sunny day, blue skies behind, realistic rendering of the human form, action and sci fi oriented, --ar 7:3 --niji 6 I’m sure you’ll agree that the results are beautiful. My favorite is #3, it’s such a gorgeous shot, but it doesn’t fit my needs. The fourth one is also stunning, and reminds me somewhat of a younger version of Florence Faivre, of ‘The Expanse’ fame. Not a bad thing! But ultimately #1 was closest in look and feel to my original image, and using the ‘variations’ control in Midjourney, I thought I’d reroll to see if I can can another look and feel that I’d prefer: These were all terrific, but I ultimately settled on the fourth image, as I liked the lighting on the side of her face. The third wasn’t bad either, but from experience, I found that when turning images into video, having more hair visible makes it a bit more dramatic, so that was the tiebreaker! Scene 4: Adding Drama Ok, now I have a couple of shots of my main character, and she’s somewhat coherent. I’ve gotten over one of the main problems with this stuff. But now, as well as poses, I need a story. I had the idea for this story to take a turn to make it interesting. Instead of her lamenting the destruction of the civilization we saw in the first shot, wouldn’t it be nice if she was celebrating it instead? So for that, I wanted two more shots. One, a closeup of her eye looking at the destruction, and the other a variation of the shot from scene 3 with the dramatic lighting of flames etc. So, I’ll start with the latter. With MidJourney, you can start from an existing image by providing its URL, which you can get by opening the image full screen, right clicking, and selecting ‘Copy Image Address’. Enter it at the beginning of the prompt, and then enhance that with your desired updates. So I started with the closeup from the previous shot, and gave the URL of that image in the prompt, like this: <https://s.mj.run/Z5fF0z68gkc> bathed in the orange glow of flames, an evil smile on her face --ar 7:3 --niji 6 And I got these results: While they’re a little more anime, and a little less CGI than the previous shot, they all look great. #3 is perhaps a little too evil! I liked #4 because, again, the eyes tell an interesting story, and kinda reflect the twist I wanted at the end. Finally, I needed a closeup on her eyes. This one was trickier than I expected, and I had to reroll and go through variations a number of times until I got something that told the story I wanted: an extreme close up of the eyes a young asian woman wearing a black and red sci fi uniform in the style of modern CGI anime, sunny day, blue skies behind, reflection of a sci fi city in her eyes, action and sci fi oriented, --ar 7:3 --niji 6 I particularly like that even through I didn’t ask for the city to be in flames, it gave me one! Scenes 5 and onwards, the alien invasion This part was relatively easy because it didn’t involve a human character. We can be more forgiving if the spaceships and city looks different, and I can write around that – talking about the destruction of an entire civilization, so diversity in image is a strength, so I got a bunch of prompts like these, to create the scenes: Before the attack a high tech sci fi city in the style of modern CGI anime, sunny day, blue skies, a dark, ugly, spaceship hovering over it, contrasting the beauty of the city. Action and Sci-Fi oriented. --ar 7:3 --niji 6 After the attack a high tech sci fi city burning. Flames everywhere torching each of the buildings. In the style of modern CGI anime, sunny day, blue skies. Action and Sci-Fi oriented. --ar 7:3 --niji 6 From the variations on these, I was able to grab a few frames. I now had a number of frames that storyboarded my movie. The next part would be to animate them, and I’ll talk about that in the next post. But first, here’s how the 8 frames tied together into a storyboard, and we can begin to see a story emerge from them:From Blog to Video with Veed2024-01-30T00:00:00-08:002024-01-30T00:00:00-08:00/2024/01/30/veed<h2 id="a-case-study-in-turning-a-blog-post-into-a-video">A case study in turning a blog post into a video…</h2>
<p>A few days back, I wrote this <a href="https://laurencemoroney.com/2024/01/27/materials.html">blog post</a>, covering a passion of mine – how AI can be used to change things that we use every day, and while so doing, improving costs, efficiency, and reducing waste, giving an overall positive impact on society and environment.</p>
<p>And then I turned it into a video like this! It took about an hour of tweaking, and maybe $10 worth of my subscription credits to create this:</p>
<div><div class="extensions extensions--video">
<iframe src="https://www.youtube.com/embed/VPNFioRQ47M?rel=0&showinfo=0" frameborder="0" scrolling="no" allowfullscreen=""></iframe>
</div>
</div>
<p>This was achieved using a service caled VEED, that I discovered via their ChatGPT plugin called ‘Video GPT by VEED’. And while I didn’t use that tool to create this video, it struck my that by creating that custom GPT, they improved their discovery, and won me as a customer.</p>
<p>VEED gives you an in-browser video creation and editing tool with all the basic bells and whistles you’d expect like a timeline editor.</p>
<p><img src="/assets/veed.jpg" alt="Veed Editor in action" /></p>
<p>Where it <em>shines</em> is in its use of AI to provide services above and beyond. Some examples of this are</p>
<ul>
<li>The AI-Generated Avatar. The news reader in the above video is entirely AI generated – or at least her lip sync is!</li>
<li>Automated subtitles, including emphasis words</li>
<li>And my current favorite – translation!</li>
</ul>
<p>Explore them here in Cantonese and Japanese:</p>
<div><div class="extensions extensions--video">
<iframe src="https://www.youtube.com/embed/27AMp5CnTO0?rel=0&showinfo=0" frameborder="0" scrolling="no" allowfullscreen=""></iframe>
</div>
</div>
<div><div class="extensions extensions--video">
<iframe src="https://www.youtube.com/embed/NdaCstomns4?rel=0&showinfo=0" frameborder="0" scrolling="no" allowfullscreen=""></iframe>
</div>
</div>
<p>Overall, this can be a terrific tool to help bloggers and content creators to extend their reach.</p>
<p>It’s not perfect – occasionally creating the AI avatar crashes, but still uses up some of your minutes (I have 60 in my annual subscription, and after creating this video, I have 56 left) – and the transitions can be awkward if you created multiple AI avatar snippets.</p>
<p>On the whole, it’s fun, though! And I’d thoroughly recommend trying it out. I’m going to create more stuff, including trying something that clones my own voice!</p>Laurence MoroneyA case study in turning a blog post into a video… A few days back, I wrote this blog post, covering a passion of mine – how AI can be used to change things that we use every day, and while so doing, improving costs, efficiency, and reducing waste, giving an overall positive impact on society and environment. And then I turned it into a video like this! It took about an hour of tweaking, and maybe $10 worth of my subscription credits to create this: This was achieved using a service caled VEED, that I discovered via their ChatGPT plugin called ‘Video GPT by VEED’. And while I didn’t use that tool to create this video, it struck my that by creating that custom GPT, they improved their discovery, and won me as a customer. VEED gives you an in-browser video creation and editing tool with all the basic bells and whistles you’d expect like a timeline editor. Where it shines is in its use of AI to provide services above and beyond. Some examples of this are The AI-Generated Avatar. The news reader in the above video is entirely AI generated – or at least her lip sync is! Automated subtitles, including emphasis words And my current favorite – translation! Explore them here in Cantonese and Japanese: Overall, this can be a terrific tool to help bloggers and content creators to extend their reach. It’s not perfect – occasionally creating the AI avatar crashes, but still uses up some of your minutes (I have 60 in my annual subscription, and after creating this video, I have 56 left) – and the transitions can be awkward if you created multiple AI avatar snippets. On the whole, it’s fun, though! And I’d thoroughly recommend trying it out. I’m going to create more stuff, including trying something that clones my own voice!Exploring text to video with DeCohere2024-01-28T00:00:00-08:002024-01-28T00:00:00-08:00/2024/01/28/test2vid%20copy<h2 id="exploring-the-wonders-of-ai-generated-imagery-with-decohere-a-personal-journey">Exploring the Wonders of AI-Generated Imagery with Decohere: A Personal Journey</h2>
<p>In the ever-evolving world of technology, AI’s role in creative fields is becoming increasingly prominent, as I recently explored through a fun experience with the Decohere website. This blog post will delve into two fascinating instances that highlight the capabilities of AI in generating unique visual content.</p>
<p><img src="/assets/decohere.png" alt="A sci-fi image of a Japanese girl" /></p>
<h3 id="real-time-image-creation-a-sci-fi-vision">Real-Time Image Creation: A Sci-Fi Vision</h3>
<p>My first interaction with Decohere involved crafting a scenario featuring a young Asian girl in a sci-fi setting. The platform’s real-time response was impressive; it kept pace with my edits, dynamically generating images as I refined my prompt. This experience wasn’t just about creating a static image; it was a journey through imagination, where each edit brought a new aspect of the scene to life.</p>
<div><div class="extensions extensions--video">
<iframe src="https://www.youtube.com/embed/iCzN0k_OoNE?rel=0&showinfo=0" frameborder="0" scrolling="no" allowfullscreen=""></iframe>
</div>
</div>
<p>I particularly liked how they underwent an enhancement process once I finished typing. This step brought a level of depth and clarity to the visuals, transforming them into more polished and vivid representations of my original idea.</p>
<p>Most of all – while these images may have limitations in how they could be used for a real production – due to coherence issues (more on that in a moment) – I can see where tools like this can be excellent in unlocking creativity. For example, it might be hard for me to think about details of how her uniform might look, if I was writing a novel. Exploring different views like this can help inspire me.</p>
<p>When I mention coherence – you can see what I mean in the images – when it generates people, they’ll all be different. If had wanted to create a movie or graphic novel with this young girl in it, I’d have challenges – because if I were to describe her in a different environment, or interacting with people, she, her clothing, and other details would be different.</p>
<h3 id="venturing-into-ai-generated-video-a-navajo-girls-sci-fi-adventure">Venturing into AI-Generated Video: A Navajo Girl’s Sci-Fi Adventure</h3>
<p>The second part of my exploration took an even more ambitious turn as I tested Decohere’s video creation tool. Here, the task was to create a video depicting a young Navajo girl in a sci-fi environment. The result, although sped up for demonstration purposes, was nothing short of astonishing.</p>
<p>The video’s creation process illustrated both the strengths and limitations of current AI technology. While the AI managed to craft a visually appealing narrative, there were noticeable inconsistencies, particularly in the character’s clothing, which varied throughout the video. This highlighted a challenge in maintaining coherence in AI-generated content over time.</p>
<p>Despite these variations, the overall quality of the video was impressive, showcasing AI’s potential in creating complex and dynamic visual narratives.</p>
<div><div class="extensions extensions--video">
<iframe src="https://www.youtube.com/embed/lXAjGJrI2G4?rel=0&showinfo=0" frameborder="0" scrolling="no" allowfullscreen=""></iframe>
</div>
</div>
<h2 id="embracing-the-future-of-ai-in-creativity">Embracing the Future of AI in Creativity</h2>
<p>These experiences with Decohere should hopefully open your eyes to the vast potential of AI in the realm of creative digital media. The ability to generate images and videos based on specific prompts is not only a testament to the technology’s advancement but also a glimpse into a future where AI can be a collaborative tool for artists and storytellers. My personal opinion is that it will be most effective when used as an enhancement to your creative workflow – giving you inspiration and/or taking some of the slower and more labor intensive parts of the creation process and automating them.</p>
<p>As we continue to explore and refine these technologies, the possibilities for creative expression are boundless. Decohere is but one example – and I’ll explore others in time, but it’s a pretty cool one!</p>
<p>[This Blog Post was written with the assistance of LLMs]</p>Laurence MoroneyExploring the Wonders of AI-Generated Imagery with Decohere: A Personal Journey In the ever-evolving world of technology, AI’s role in creative fields is becoming increasingly prominent, as I recently explored through a fun experience with the Decohere website. This blog post will delve into two fascinating instances that highlight the capabilities of AI in generating unique visual content. Real-Time Image Creation: A Sci-Fi Vision My first interaction with Decohere involved crafting a scenario featuring a young Asian girl in a sci-fi setting. The platform’s real-time response was impressive; it kept pace with my edits, dynamically generating images as I refined my prompt. This experience wasn’t just about creating a static image; it was a journey through imagination, where each edit brought a new aspect of the scene to life. I particularly liked how they underwent an enhancement process once I finished typing. This step brought a level of depth and clarity to the visuals, transforming them into more polished and vivid representations of my original idea. Most of all – while these images may have limitations in how they could be used for a real production – due to coherence issues (more on that in a moment) – I can see where tools like this can be excellent in unlocking creativity. For example, it might be hard for me to think about details of how her uniform might look, if I was writing a novel. Exploring different views like this can help inspire me. When I mention coherence – you can see what I mean in the images – when it generates people, they’ll all be different. If had wanted to create a movie or graphic novel with this young girl in it, I’d have challenges – because if I were to describe her in a different environment, or interacting with people, she, her clothing, and other details would be different. Venturing into AI-Generated Video: A Navajo Girl’s Sci-Fi Adventure The second part of my exploration took an even more ambitious turn as I tested Decohere’s video creation tool. Here, the task was to create a video depicting a young Navajo girl in a sci-fi environment. The result, although sped up for demonstration purposes, was nothing short of astonishing. The video’s creation process illustrated both the strengths and limitations of current AI technology. While the AI managed to craft a visually appealing narrative, there were noticeable inconsistencies, particularly in the character’s clothing, which varied throughout the video. This highlighted a challenge in maintaining coherence in AI-generated content over time. Despite these variations, the overall quality of the video was impressive, showcasing AI’s potential in creating complex and dynamic visual narratives. Embracing the Future of AI in Creativity These experiences with Decohere should hopefully open your eyes to the vast potential of AI in the realm of creative digital media. The ability to generate images and videos based on specific prompts is not only a testament to the technology’s advancement but also a glimpse into a future where AI can be a collaborative tool for artists and storytellers. My personal opinion is that it will be most effective when used as an enhancement to your creative workflow – giving you inspiration and/or taking some of the slower and more labor intensive parts of the creation process and automating them. As we continue to explore and refine these technologies, the possibilities for creative expression are boundless. Decohere is but one example – and I’ll explore others in time, but it’s a pretty cool one! [This Blog Post was written with the assistance of LLMs]Revolutionizing Batteries with AI2024-01-27T00:00:00-08:002024-01-27T00:00:00-08:00/2024/01/27/materials<h2 id="revolutionizing-batteries-with-ai-a-leap-towards-sustainable-technology">Revolutionizing Batteries with AI: A Leap Towards Sustainable Technology</h2>
<p>In a world increasingly dependent on technology, the quest for efficient and environmentally friendly power sources is more crucial than ever. A recent breakthrough, as reported by New Scientist, showcases how artificial intelligence (AI) is leading this charge, offering a glimpse into a future where technology and sustainability go hand in hand.</p>
<h3 id="understanding-the-breakthrough">Understanding the Breakthrough</h3>
<p>The heart of this advancement lies in an AI-designed battery that dramatically reduces the need for lithium - a costly and environmentally taxing material. This isn’t just a minor tweak; it’s a game-changing reduction of up to 70% less lithium compared to conventional batteries.</p>
<h3 id="the-role-of-ai-in-material-discovery">The Role of AI in Material Discovery</h3>
<p>At first glance, AI and material science might seem like distant cousins. However, they share a common language: the language of patterns. AI excels at identifying patterns and making predictions, a skill that proves invaluable in the realm of material science. In this case, AI algorithms sifted through millions of potential materials, identifying the ones most likely to yield effective, less lithium-dependent batteries.</p>
<h3 id="the-process-from-data-to-discovery">The Process: From Data to Discovery</h3>
<p>Imagine a world of possibilities, millions of potential materials each holding the key to the next big discovery. The challenge lies not in creating these materials but in finding the proverbial needle in the haystack. This is where AI steps in. By rapidly analyzing and dismissing unsuitable candidates, AI narrows down the search to a manageable list, transforming years of human effort into mere months of computation.</p>
<p><img src="/assets/filter.png" alt="An AI filtering many materials into fewer valuable ones" /></p>
<h3 id="from-simulation-to-reality">From Simulation to Reality</h3>
<p>The journey from AI-simulated predictions to real-world applications is a critical one. In this scenario, the AI’s recommendations were not just theoretical musings but were synthesized and tested in the lab. The result? A functional battery that, despite its lower conductivity, stands as a testament to the potential of AI-driven material innovation.</p>
<h3 id="why-this-matters">Why This Matters</h3>
<p>This breakthrough is more than just a scientific curiosity; it’s a beacon of hope in our quest for sustainable technology. Lithium, while effective, comes with a hefty environmental cost. By reducing our reliance on this mineral, we pave the way for more sustainable and less harmful technologies.</p>
<h3 id="looking-ahead">Looking Ahead</h3>
<p>The success of this project is a clarion call to the scientific community. AI’s ability to augment and accelerate the material discovery process is a tool we can no longer afford to overlook. As we stand on the brink of a technological revolution, it’s clear that AI will be a driving force, not just in computing, but in helping us build a more sustainable world.</p>
<h3 id="conclusion">Conclusion</h3>
<p>In conclusion, the collaboration between AI and material science is more than just a novel experiment; it’s a pathway to a future where technology and the environment can coexist harmoniously. This AI-driven battery design is just the beginning. As we harness the power of AI for material discovery, we open doors to endless possibilities - cleaner energy, sustainable technologies, and a brighter future for our planet.</p>
<p>For a deeper dive into this fascinating development, check out the links below.</p>
<p>References:</p>
<p><a href="https://www.newscientist.com/article/2411374-ai-comes-up-with-battery-design-that-uses-70-per-cent-less-lithium/">AI comes up with battery design that uses 70 per cent less lithium.” New Scientist</a>.</p>
<p><a href="https://arxiv.org/abs/2401.04070">Research paper on arXiv.</a></p>
<p>[This Blog Post was written with the assistance of LLMs]</p>Laurence MoroneyRevolutionizing Batteries with AI: A Leap Towards Sustainable Technology In a world increasingly dependent on technology, the quest for efficient and environmentally friendly power sources is more crucial than ever. A recent breakthrough, as reported by New Scientist, showcases how artificial intelligence (AI) is leading this charge, offering a glimpse into a future where technology and sustainability go hand in hand. Understanding the Breakthrough The heart of this advancement lies in an AI-designed battery that dramatically reduces the need for lithium - a costly and environmentally taxing material. This isn’t just a minor tweak; it’s a game-changing reduction of up to 70% less lithium compared to conventional batteries. The Role of AI in Material Discovery At first glance, AI and material science might seem like distant cousins. However, they share a common language: the language of patterns. AI excels at identifying patterns and making predictions, a skill that proves invaluable in the realm of material science. In this case, AI algorithms sifted through millions of potential materials, identifying the ones most likely to yield effective, less lithium-dependent batteries. The Process: From Data to Discovery Imagine a world of possibilities, millions of potential materials each holding the key to the next big discovery. The challenge lies not in creating these materials but in finding the proverbial needle in the haystack. This is where AI steps in. By rapidly analyzing and dismissing unsuitable candidates, AI narrows down the search to a manageable list, transforming years of human effort into mere months of computation. From Simulation to Reality The journey from AI-simulated predictions to real-world applications is a critical one. In this scenario, the AI’s recommendations were not just theoretical musings but were synthesized and tested in the lab. The result? A functional battery that, despite its lower conductivity, stands as a testament to the potential of AI-driven material innovation. Why This Matters This breakthrough is more than just a scientific curiosity; it’s a beacon of hope in our quest for sustainable technology. Lithium, while effective, comes with a hefty environmental cost. By reducing our reliance on this mineral, we pave the way for more sustainable and less harmful technologies. Looking Ahead The success of this project is a clarion call to the scientific community. AI’s ability to augment and accelerate the material discovery process is a tool we can no longer afford to overlook. As we stand on the brink of a technological revolution, it’s clear that AI will be a driving force, not just in computing, but in helping us build a more sustainable world. Conclusion In conclusion, the collaboration between AI and material science is more than just a novel experiment; it’s a pathway to a future where technology and the environment can coexist harmoniously. This AI-driven battery design is just the beginning. As we harness the power of AI for material discovery, we open doors to endless possibilities - cleaner energy, sustainable technologies, and a brighter future for our planet. For a deeper dive into this fascinating development, check out the links below. References: AI comes up with battery design that uses 70 per cent less lithium.” New Scientist. Research paper on arXiv. [This Blog Post was written with the assistance of LLMs]NFTs - Navigating Beyond the Hype Towards Real Value2024-01-26T00:00:00-08:002024-01-26T00:00:00-08:00/2024/01/26/nft<p><img src="/assets/boredape.png" alt="A Bored Ape" /></p>
<p>In the wake of the cryptocurrency boom, Non-Fungible Tokens (NFTs) emerged as the digital darlings of the art and collectibles world. However, as we navigate through 2024, the luster of NFTs seems to be fading. A recent <a href="https://dappgambl.com/nfts/dead-nfts/">report by DappGambl</a> highlights that a staggering 95% of NFT collections are now considered ‘dead,’ with little to no market value. This stark revelation is a wake-up call to the NFT community and potential investors, signaling a shift from the gold rush mentality to a more sustainable and value-oriented approach.</p>
<p>The initial appeal of NFTs lay in their novelty as unique digital assets - be it art, music, or video clips. This allure drove a speculative frenzy, culminating in record-breaking sales and headline-grabbing auctions. However, as with any gold rush, the inevitable market correction has left many investors holding assets with diminished or no value. According to DappGambl, the majority of NFT collections have failed to find a sustainable market, reflecting an oversupply and a lack of adequate demand.</p>
<p>But perhaps more pressing than the market dynamics is the environmental cost of NFTs. The energy-intensive process of minting NFTs on blockchain networks has significant carbon footprints. With thousands of NFTs generating emissions equivalent to that of small countries, the environmental impact is no longer a secondary concern but a primary issue that needs addressing.</p>
<p><img src="/assets/pile.png" alt="A Waste Pile of old NFTs" /></p>
<p>Amidst this backdrop, the future of NFTs hinges on a critical pivot - from transient, speculative assets to digital assets with intrinsic, tangible value. The era of NFTs serving merely as digital avatars or speculative art pieces is waning. Instead, the real opportunity lies in leveraging the unique properties of NFTs for applications with inherent utility.</p>
<p>Imagine digital assets that are not just collectibles but integral to digital identity, gaming, real estate, and more. NFTs could represent fractional ownership in physical assets, offer exclusive access to services, or play a crucial role in supply chain management. The key is in their uniqueness and the immutable proof of ownership that blockchain technology provides.</p>
<p>Moreover, the evolution of NFTs must be cognizant of the environmental impact. The shift towards more energy-efficient blockchain protocols or the integration of eco-friendly practices in NFT creation and trading could mitigate the environmental concerns, aligning the NFT market with broader sustainability goals.</p>
<p>In conclusion, while the hype around NFTs as speculative art pieces may be fading, their potential as digital assets with real-world utility is only beginning to be tapped. As the market matures and becomes more discerning, the future of NFTs will likely be defined by their ability to offer value beyond the digital realm, aligning with environmental considerations and practical use cases. The gold rush might be over, but a more sustainable and value-driven era for NFTs is on the horizon.</p>
<p>[This Blog Post was written with the assistance of LLMs]</p>Laurence MoroneyIn the wake of the cryptocurrency boom, Non-Fungible Tokens (NFTs) emerged as the digital darlings of the art and collectibles world. However, as we navigate through 2024, the luster of NFTs seems to be fading. A recent report by DappGambl highlights that a staggering 95% of NFT collections are now considered ‘dead,’ with little to no market value. This stark revelation is a wake-up call to the NFT community and potential investors, signaling a shift from the gold rush mentality to a more sustainable and value-oriented approach. The initial appeal of NFTs lay in their novelty as unique digital assets - be it art, music, or video clips. This allure drove a speculative frenzy, culminating in record-breaking sales and headline-grabbing auctions. However, as with any gold rush, the inevitable market correction has left many investors holding assets with diminished or no value. According to DappGambl, the majority of NFT collections have failed to find a sustainable market, reflecting an oversupply and a lack of adequate demand. But perhaps more pressing than the market dynamics is the environmental cost of NFTs. The energy-intensive process of minting NFTs on blockchain networks has significant carbon footprints. With thousands of NFTs generating emissions equivalent to that of small countries, the environmental impact is no longer a secondary concern but a primary issue that needs addressing. Amidst this backdrop, the future of NFTs hinges on a critical pivot - from transient, speculative assets to digital assets with intrinsic, tangible value. The era of NFTs serving merely as digital avatars or speculative art pieces is waning. Instead, the real opportunity lies in leveraging the unique properties of NFTs for applications with inherent utility. Imagine digital assets that are not just collectibles but integral to digital identity, gaming, real estate, and more. NFTs could represent fractional ownership in physical assets, offer exclusive access to services, or play a crucial role in supply chain management. The key is in their uniqueness and the immutable proof of ownership that blockchain technology provides. Moreover, the evolution of NFTs must be cognizant of the environmental impact. The shift towards more energy-efficient blockchain protocols or the integration of eco-friendly practices in NFT creation and trading could mitigate the environmental concerns, aligning the NFT market with broader sustainability goals. In conclusion, while the hype around NFTs as speculative art pieces may be fading, their potential as digital assets with real-world utility is only beginning to be tapped. As the market matures and becomes more discerning, the future of NFTs will likely be defined by their ability to offer value beyond the digital realm, aligning with environmental considerations and practical use cases. The gold rush might be over, but a more sustainable and value-driven era for NFTs is on the horizon. [This Blog Post was written with the assistance of LLMs]Prompt Engineering Best Practices2023-12-19T00:00:00-08:002023-12-19T00:00:00-08:00/2023/12/19/prompts<h1 id="effective-prompt-design-strategies-and-best-practices">Effective Prompt Design: Strategies and Best Practices</h1>
<p>The art of prompt design is pivotal in harnessing the full potential of AI models. A well-crafted prompt can be the difference between an average and an exceptional AI response. My team in Google have published guidelines for this that are well worth <a href="https://ai.google.dev/docs/prompt_best_practices">checking out in more detail</a></p>
<p><img src="/assets/prompt_eng.jpg" alt="Prompt Engineering, visualized" /></p>
<p>Here are some key strategies and best practices to consider:</p>
<h4 id="clarity-in-instructions">Clarity in Instructions:</h4>
<p>Clear, concise instructions are crucial. They guide the AI to understand and respond accurately to the task at hand.</p>
<h4 id="inclusion-of-examples">Inclusion of Examples:</h4>
<p>Including examples in prompts can significantly enhance the model’s understanding and accuracy.</p>
<h4 id="partial-input-completion">Partial Input Completion:</h4>
<p>Designing prompts that allow the model to complete partial inputs can lead to more precise and relevant responses.</p>
<h4 id="specific-response-formats">Specific Response Formats:</h4>
<p>Prompting for responses in specific formats ensures consistency and relevance in the outputs.</p>
<h4 id="contextual-information">Contextual Information:</h4>
<p>Adding context to prompts helps in refining and focusing the model’s responses.</p>
<h4 id="effective-use-of-prefixes">Effective Use of Prefixes:</h4>
<p>Prefixes can guide the model in understanding the kind of response expected.</p>
<h4 id="experimentation">Experimentation:</h4>
<p>Trying different phrasings, parameter values, and prompt structures can lead to improved outcomes.</p>
<h4 id="iterative-design">Iterative Design:</h4>
<p>Continuous refinement and iteration of prompts are key to achieving optimal responses.</p>
<h4 id="fallback-responses">Fallback Responses:</h4>
<p>Planning for fallback responses in case of inadequate or incorrect outputs is essential.</p>
<h4 id="caution-with-factual-generation">Caution with Factual Generation:</h4>
<p>Be wary of over-reliance on the model for generating factual content, and exercise caution with tasks involving complex logic or mathematics.</p>
<h2 id="conclusion">Conclusion:</h2>
<p>These strategies, rooted in Google AI’s guidelines, are fundamental in crafting effective prompts that lead to high-quality, reliable AI outputs. For a more detailed exploration of these best practices, visit Google AI’s Prompt Design Strategies.</p>Laurence MoroneyEffective Prompt Design: Strategies and Best Practices The art of prompt design is pivotal in harnessing the full potential of AI models. A well-crafted prompt can be the difference between an average and an exceptional AI response. My team in Google have published guidelines for this that are well worth checking out in more detail Here are some key strategies and best practices to consider: Clarity in Instructions: Clear, concise instructions are crucial. They guide the AI to understand and respond accurately to the task at hand. Inclusion of Examples: Including examples in prompts can significantly enhance the model’s understanding and accuracy. Partial Input Completion: Designing prompts that allow the model to complete partial inputs can lead to more precise and relevant responses. Specific Response Formats: Prompting for responses in specific formats ensures consistency and relevance in the outputs. Contextual Information: Adding context to prompts helps in refining and focusing the model’s responses. Effective Use of Prefixes: Prefixes can guide the model in understanding the kind of response expected. Experimentation: Trying different phrasings, parameter values, and prompt structures can lead to improved outcomes. Iterative Design: Continuous refinement and iteration of prompts are key to achieving optimal responses. Fallback Responses: Planning for fallback responses in case of inadequate or incorrect outputs is essential. Caution with Factual Generation: Be wary of over-reliance on the model for generating factual content, and exercise caution with tasks involving complex logic or mathematics. Conclusion: These strategies, rooted in Google AI’s guidelines, are fundamental in crafting effective prompts that lead to high-quality, reliable AI outputs. For a more detailed exploration of these best practices, visit Google AI’s Prompt Design Strategies.Towards Artificial Understanding2023-12-14T00:00:00-08:002023-12-14T00:00:00-08:00/2023/12/14/towards-au<h1 id="introduction">Introduction:</h1>
<p>In our journey through the technological renaissance, we’ve witnessed the rise and evolution of artificial intelligence (AI). However, a new concept is emerging, poised to redefine our understanding of machine intelligence: and I call it Artificial Understanding (AU). This groundbreaking idea extends the capabilities of AI, pushing it beyond mere data processing and content generation. Today, we explore what Artificial Understanding entails and how it can reshape the landscape of AI.</p>
<p><img src="/assets/au.jpg" alt="AU drawn by DALL-E" /></p>
<h2 id="the-essence-of-artificial-understanding">The Essence of Artificial <em>Understanding</em></h2>
<p>Artificial Understanding represents a significant leap in the field of AI. It’s about endowing machines with the ability to not just process but to interpret and comprehend information in a contextually meaningful way. AU takes AI a step further from performing predefined tasks to possessing a level of understanding akin to basic human cognition. This advancement is pivotal as it transforms AI from a tool that follows instructions to an entity that can grasp context, nuances, and the subtleties of the data it processes.</p>
<h3 id="point-1-transformers---the-building-blocks-of-understanding">Point 1: Transformers - The Building Blocks of Understanding</h3>
<h4 id="understanding-sequences-of-text">Understanding Sequences of Text</h4>
<p>Central to the concept of AU are transformer models. Unlike conventional algorithms that treat data as isolated points, transformers understand sequences of text, grasping the context and relationships within. This capability is akin to a rudimentary form of understanding, where the model discerns patterns and meanings beyond mere words.</p>
<h4 id="case-studies">Case Studies</h4>
<p>GPT (Generative Pre-trained Transformer) and BERT (Bidirectional Encoder Representations from Transformers) serve as prime examples of this technology. They demonstrate how transformers can not only generate coherent and contextually relevant text but also interpret and respond to human language with a surprising degree of understanding.</p>
<h3 id="point-2-efficient-compression-of-knowledge">Point 2: Efficient Compression of Knowledge</h3>
<h4 id="beyond-generative-capabilities">Beyond Generative Capabilities</h4>
<p>The true marvel of transformer models, such as GPT or Gemini, lies in their remarkable efficiency in knowledge compression. This efficiency is a critical aspect of Artificial Understanding. These models manage to compress vast amounts of text – potentially equivalent to petabytes of data – into a model of just a few billion (or hundred-million) parameters, which occupies only gigabytes (or less) of space. This is not merely a triumph in data storage but a reflection of how these models distill and encapsulate vast information, enabling them to ‘understand’ and apply this knowledge effectively.</p>
<h2 id="implications-and-future-prospects">Implications and Future Prospects</h2>
<p>The implications of Artificial Understanding are vast and far-reaching. In the realm of customer service, it could lead to chatbots that not only answer queries but understand the emotions and intents behind them. In data analysis, AU could enable systems to not just process numbers but to interpret trends and insights in a contextually rich manner. The potential applications are limitless, stretching across various domains from healthcare, where understanding patient history and symptoms could revolutionize diagnoses, to finance, where understanding market nuances is key.</p>
<h1 id="conclusion">Conclusion</h1>
<p>Artificial Understanding is more than just an incremental addition to the AI landscape; it is a paradigm shift. It challenges our preconceptions of machine capabilities, opening doors to possibilities where machines don’t just calculate but comprehend. As we stand at the precipice of this new era in AI, the promise and potential of Artificial Understanding beckon a future where the lines between human and machine understanding become increasingly blurred, leading to innovations we can only begin to imagine.</p>
<p>[This post was created with the assistance of LLMs]</p>Laurence MoroneyIntroduction: In our journey through the technological renaissance, we’ve witnessed the rise and evolution of artificial intelligence (AI). However, a new concept is emerging, poised to redefine our understanding of machine intelligence: and I call it Artificial Understanding (AU). This groundbreaking idea extends the capabilities of AI, pushing it beyond mere data processing and content generation. Today, we explore what Artificial Understanding entails and how it can reshape the landscape of AI. The Essence of Artificial Understanding Artificial Understanding represents a significant leap in the field of AI. It’s about endowing machines with the ability to not just process but to interpret and comprehend information in a contextually meaningful way. AU takes AI a step further from performing predefined tasks to possessing a level of understanding akin to basic human cognition. This advancement is pivotal as it transforms AI from a tool that follows instructions to an entity that can grasp context, nuances, and the subtleties of the data it processes. Point 1: Transformers - The Building Blocks of Understanding Understanding Sequences of Text Central to the concept of AU are transformer models. Unlike conventional algorithms that treat data as isolated points, transformers understand sequences of text, grasping the context and relationships within. This capability is akin to a rudimentary form of understanding, where the model discerns patterns and meanings beyond mere words. Case Studies GPT (Generative Pre-trained Transformer) and BERT (Bidirectional Encoder Representations from Transformers) serve as prime examples of this technology. They demonstrate how transformers can not only generate coherent and contextually relevant text but also interpret and respond to human language with a surprising degree of understanding. Point 2: Efficient Compression of Knowledge Beyond Generative Capabilities The true marvel of transformer models, such as GPT or Gemini, lies in their remarkable efficiency in knowledge compression. This efficiency is a critical aspect of Artificial Understanding. These models manage to compress vast amounts of text – potentially equivalent to petabytes of data – into a model of just a few billion (or hundred-million) parameters, which occupies only gigabytes (or less) of space. This is not merely a triumph in data storage but a reflection of how these models distill and encapsulate vast information, enabling them to ‘understand’ and apply this knowledge effectively. Implications and Future Prospects The implications of Artificial Understanding are vast and far-reaching. In the realm of customer service, it could lead to chatbots that not only answer queries but understand the emotions and intents behind them. In data analysis, AU could enable systems to not just process numbers but to interpret trends and insights in a contextually rich manner. The potential applications are limitless, stretching across various domains from healthcare, where understanding patient history and symptoms could revolutionize diagnoses, to finance, where understanding market nuances is key. Conclusion Artificial Understanding is more than just an incremental addition to the AI landscape; it is a paradigm shift. It challenges our preconceptions of machine capabilities, opening doors to possibilities where machines don’t just calculate but comprehend. As we stand at the precipice of this new era in AI, the promise and potential of Artificial Understanding beckon a future where the lines between human and machine understanding become increasingly blurred, leading to innovations we can only begin to imagine. [This post was created with the assistance of LLMs]Evaluating Hallucination - A danger of LLMs2023-11-29T00:00:00-08:002023-11-29T00:00:00-08:00/2023/11/29/evaluating-hallucination<p>I just read a very interesting paper on Arxiv titled “The impact of using an AI chatbot to respond to patient messages.” where an evaluation of human-chat versus GPT-chat was done, measuring effectiveness, efficiency, and the potential danger of hallucination. YOu can see the paper [here] (https://arxiv.org/abs/2310.17703)</p>
<h3 id="introduction">Introduction</h3>
<p>The integration of Artificial Intelligence (AI) into healthcare is a rapidly evolving frontier, promising to revolutionize patient care and clinician workflow. A recent study published on arXiv.org delves into this topic, exploring the effects of using an AI chatbot, specifically GPT-4, for responding to patient messages in a healthcare setting.</p>
<p><img src="/assets/ai-med.jpg" alt="AI in Medicine" /></p>
<h3 id="overview-of-the-study">Overview of the Study</h3>
<p>This study aimed to assess the acceptability, safety, and efficiency of using an AI-based chatbot in drafting responses to patient queries. Conducted at Brigham and Women’s Hospital, Boston, in 2023, it involved six board-certified oncologists who responded to 100 realistic cancer patient scenarios. The study was designed in two stages, where in the first stage, oncologists manually responded to patient messages, and in the second stage, they edited responses generated by GPT-4.</p>
<h3 id="key-findings">Key Findings</h3>
<p>The results were significant:</p>
<ul>
<li><strong>Efficiency Improvement</strong>: In 77% of cases, the use of GPT-4 improved documentation efficiency.</li>
<li><strong>Safety</strong>: 82% of the time, the AI-generated responses were considered safe.</li>
<li><strong>Risk of Harm</strong>: However, there was a 7.7% chance that unedited GPT-4 responses could lead to severe harm or death.</li>
</ul>
<h3 id="physician-perception-of-ai-responses">Physician Perception of AI Responses</h3>
<p>Interestingly, in 31% of cases, physicians believed the GPT-4 drafts were written by humans. This perception highlights the sophistication of AI-generated responses but also underscores the necessity for careful review.</p>
<h3 id="impact-on-patient-education-and-clinical-actions">Impact on Patient Education and Clinical Actions</h3>
<p>The use of AI led to an increase in patient education recommendations but a decrease in direct clinical actions. This shift suggests that while AI can enhance informational support, it might underrepresent the need for immediate clinical interventions.</p>
<h3 id="risks-and-challenges">Risks and Challenges</h3>
<p>Despite the benefits, the study highlights critical risks associated with AI in healthcare:</p>
<ul>
<li>Potential for severe harm in a small but significant percentage of responses.</li>
<li>The need for diligent oversight and human intervention to ensure patient safety.</li>
</ul>
<h3 id="future-implications">Future Implications</h3>
<p>The study opens up a discussion about the future role of AI in healthcare communication. While AI shows promise in reducing clinician workload and enhancing patient education, its impact on clinical decision-making and the potential risks involved require ongoing scrutiny and responsible implementation.</p>
<h3 id="conclusion">Conclusion</h3>
<p>In conclusion, this study on using AI chatbots for patient communication reveals a complex balance. While AI can significantly enhance efficiency and patient education, it introduces risks that necessitate vigilant oversight. The future of AI in healthcare holds great promise but demands a cautious and well-informed approach.</p>
<h3 id="references">References</h3>
<ul>
<li>Chen, S., et al. (2023). The impact of using an AI chatbot to respond to patient messages. arXiv.org. <a href="https://arxiv.org/abs/2310.17703">arXiv:2310.17703</a></li>
</ul>
<p><em>Please note that this blog post is a summary and interpretation of the study and does not cover all aspects of the research. For a comprehensive understanding, readers are encouraged to refer to the original paper.</em></p>
<p><strong>Note: This post was written with the assistance of Large Language Models</strong></p>Laurence MoroneyI just read a very interesting paper on Arxiv titled “The impact of using an AI chatbot to respond to patient messages.” where an evaluation of human-chat versus GPT-chat was done, measuring effectiveness, efficiency, and the potential danger of hallucination. YOu can see the paper [here] (https://arxiv.org/abs/2310.17703) Introduction The integration of Artificial Intelligence (AI) into healthcare is a rapidly evolving frontier, promising to revolutionize patient care and clinician workflow. A recent study published on arXiv.org delves into this topic, exploring the effects of using an AI chatbot, specifically GPT-4, for responding to patient messages in a healthcare setting. Overview of the Study This study aimed to assess the acceptability, safety, and efficiency of using an AI-based chatbot in drafting responses to patient queries. Conducted at Brigham and Women’s Hospital, Boston, in 2023, it involved six board-certified oncologists who responded to 100 realistic cancer patient scenarios. The study was designed in two stages, where in the first stage, oncologists manually responded to patient messages, and in the second stage, they edited responses generated by GPT-4. Key Findings The results were significant: Efficiency Improvement: In 77% of cases, the use of GPT-4 improved documentation efficiency. Safety: 82% of the time, the AI-generated responses were considered safe. Risk of Harm: However, there was a 7.7% chance that unedited GPT-4 responses could lead to severe harm or death. Physician Perception of AI Responses Interestingly, in 31% of cases, physicians believed the GPT-4 drafts were written by humans. This perception highlights the sophistication of AI-generated responses but also underscores the necessity for careful review. Impact on Patient Education and Clinical Actions The use of AI led to an increase in patient education recommendations but a decrease in direct clinical actions. This shift suggests that while AI can enhance informational support, it might underrepresent the need for immediate clinical interventions. Risks and Challenges Despite the benefits, the study highlights critical risks associated with AI in healthcare: Potential for severe harm in a small but significant percentage of responses. The need for diligent oversight and human intervention to ensure patient safety. Future Implications The study opens up a discussion about the future role of AI in healthcare communication. While AI shows promise in reducing clinician workload and enhancing patient education, its impact on clinical decision-making and the potential risks involved require ongoing scrutiny and responsible implementation. Conclusion In conclusion, this study on using AI chatbots for patient communication reveals a complex balance. While AI can significantly enhance efficiency and patient education, it introduces risks that necessitate vigilant oversight. The future of AI in healthcare holds great promise but demands a cautious and well-informed approach. References Chen, S., et al. (2023). The impact of using an AI chatbot to respond to patient messages. arXiv.org. arXiv:2310.17703 Please note that this blog post is a summary and interpretation of the study and does not cover all aspects of the research. For a comprehensive understanding, readers are encouraged to refer to the original paper. Note: This post was written with the assistance of Large Language ModelsNavigating the AI Landscape - The Need for Expert-Led Regulation2023-11-10T00:00:00-08:002023-11-10T00:00:00-08:00/2023/11/10/ai-regulation<p>In recent years, artificial intelligence (AI) has leaped from science fiction into our daily lives, bringing both incredible potential and significant challenges. As AI continues to evolve and integrate into various sectors, the question of regulation becomes increasingly pressing. However, the key to effective AI regulation lies not in reactive policies driven by fear and misunderstanding, but in informed, expert-led approaches that understand the nuances of AI technology.</p>
<p><img src="/assets/ai-regulation.jpg" alt="Regulation and Expertise" /></p>
<h3 id="understanding-the-fear-and-the-reality">Understanding the Fear and the Reality</h3>
<p>AI, much like any transformative technology, has been met with its share of public apprehension. Stories of AI going rogue or replacing human jobs have fueled a narrative that often strays from reality. It is crucial to distinguish between irrational fears, often amplified by sensationalist media, and legitimate concerns regarding privacy, security, and ethical use of AI.</p>
<h3 id="the-role-of-experts-in-ai-regulation">The Role of Experts in AI Regulation</h3>
<p>Effective AI regulation requires a deep understanding of the technology’s capabilities, limitations, and impact across various domains. This level of insight is typically beyond the scope of political expertise. AI experts, including scientists, ethicists, and technologists, are better equipped to foresee the implications of restricting AI development and use. They can provide balanced views that consider both the potential risks and the immense benefits AI offers.</p>
<h3 id="balancing-innovation-and-safety">Balancing Innovation and Safety</h3>
<p>The goal of regulation should be to mitigate risks without stifling innovation. Overly restrictive regulations, born from misunderstandings or fear, could hinder the advancement of AI technologies that have the potential to solve some of our most pressing global challenges. Experts can help draft regulations that ensure safety and ethical practices while still encouraging innovation and research.</p>
<h3 id="collaborative-approach-for-holistic-regulation">Collaborative Approach for Holistic Regulation</h3>
<p>Regulation should not be a unilateral process dictated by a single group. A collaborative approach, involving AI experts, policymakers, industry leaders, and the public, is essential. Such collaboration can ensure that regulations are practical, well-informed, and considerate of diverse perspectives and societal needs.</p>
<h3 id="looking-forward">Looking Forward</h3>
<p>As AI becomes increasingly integrated into our world, the need for thoughtful, informed regulation becomes more apparent. Entrusting this task to those who understand the intricate workings and potential of AI is not just prudent; it is <em>necessary</em> to harness the full potential of AI technologies while safeguarding societal values and human welfare.</p>
<p>Regulating AI should be a forward-looking endeavor, not a reactionary one. By leaning on the expertise of those who best understand the technology, we can create a regulatory framework that balances innovation with responsibility, ensuring that AI serves as a tool for human progress, not a subject of unfounded fear.</p>
<p><strong>Note: This post was written with the assistance of Large Language Models</strong></p>Laurence MoroneyIn recent years, artificial intelligence (AI) has leaped from science fiction into our daily lives, bringing both incredible potential and significant challenges. As AI continues to evolve and integrate into various sectors, the question of regulation becomes increasingly pressing. However, the key to effective AI regulation lies not in reactive policies driven by fear and misunderstanding, but in informed, expert-led approaches that understand the nuances of AI technology. Understanding the Fear and the Reality AI, much like any transformative technology, has been met with its share of public apprehension. Stories of AI going rogue or replacing human jobs have fueled a narrative that often strays from reality. It is crucial to distinguish between irrational fears, often amplified by sensationalist media, and legitimate concerns regarding privacy, security, and ethical use of AI. The Role of Experts in AI Regulation Effective AI regulation requires a deep understanding of the technology’s capabilities, limitations, and impact across various domains. This level of insight is typically beyond the scope of political expertise. AI experts, including scientists, ethicists, and technologists, are better equipped to foresee the implications of restricting AI development and use. They can provide balanced views that consider both the potential risks and the immense benefits AI offers. Balancing Innovation and Safety The goal of regulation should be to mitigate risks without stifling innovation. Overly restrictive regulations, born from misunderstandings or fear, could hinder the advancement of AI technologies that have the potential to solve some of our most pressing global challenges. Experts can help draft regulations that ensure safety and ethical practices while still encouraging innovation and research. Collaborative Approach for Holistic Regulation Regulation should not be a unilateral process dictated by a single group. A collaborative approach, involving AI experts, policymakers, industry leaders, and the public, is essential. Such collaboration can ensure that regulations are practical, well-informed, and considerate of diverse perspectives and societal needs. Looking Forward As AI becomes increasingly integrated into our world, the need for thoughtful, informed regulation becomes more apparent. Entrusting this task to those who understand the intricate workings and potential of AI is not just prudent; it is necessary to harness the full potential of AI technologies while safeguarding societal values and human welfare. Regulating AI should be a forward-looking endeavor, not a reactionary one. By leaning on the expertise of those who best understand the technology, we can create a regulatory framework that balances innovation with responsibility, ensuring that AI serves as a tool for human progress, not a subject of unfounded fear. Note: This post was written with the assistance of Large Language Models