Talking to chatbots: do AIs like art?

Original LinkedIn article.

Continuing a series of LinkedIn articles about my experience with generative AI. First in the series was Talking to chatbots: Play-to-Learn.

What is art?

That’s a daring way to start an article, right? I don’t know how to define art without investing a significant amount of time in it, getting help from academic writing, or from several articles that I may find online after doing some comprehensive research… or from an AI chatbot, the virtual friend that has the incredible potential to boost people’s productivity and creativity. So I just asked:

😎  What is art? Provide a few short definitions.

🤖 Art is not easy to define, as different people may have different views and interpretations of what constitutes art. However, here are a few short definitions from various sources:

• Art is the expression or application of human creative skill and imagination, typically in a visual form such as painting or sculpture, producing works to be appreciated primarily for their beauty or emotional power.

• Art is a visual object or experience consciously created through an expression of skill or imagination.

• Art is a diverse range of human activity, and resulting product, that involves creative or imaginative talent expressive of technical proficiency, beauty, emotional power, or conceptual ideas.

Bing Chat

This simple prompt and answer might not impress most people, but the AI chatbot has proven to have the skill that can help write valuable text quickly. Large Language Models (LLM’s) are the key technology behind this. LLM’s are neural networks with billions of parameters that are trained on huge amounts of text data. They can learn the complexity and patterns of language and generate text for various tasks, including one as difficult as defining art in words. 

LLMs are probably the most familiar generative AI algorithms because they have been around for a while and have been incorporated to some degree in popular virtual assistants (Siri, Alexa, Google Assistant…), chatbots (ChatGPT, Bing Chat…) and search engines (Google, Bing Search…). But LLMs are not the only technology behind the huge hype and excitement about generative AI. They are actually just the tip of the iceberg. This chart from McKinsey & Company illustrates what kind of iceberg we are dealing with:

Link: McKinsey & Company: Exploring opportunities in the generative AI value chain

This article in the Talking to Chatbots series will examine how some of the GenAI models can enhance our productivity and creativity, particularly the ones that work with Images. Imaging models, along with LLMs (Text), are the ones that have the most widespread adoption nowadays among all the modalities defined in the above chart. They have also brought a new and controversial concept to media and the daily lives and jobs of many people: AI art.

The focus of this series of articles is still in chatbots and LLM’s, and I won’t forget to talk about them and their relation to AI art. But, before going into details, I decided to get help from a text-to-image AI solution (my favourite, NightCafe Studio) to get another perspective of what (visual) art might be. This time by transforming text into an image:

See all the details by accessing the image on NightCafé Studio. You can easily replicate my prompts or try your own. If you are new to AI imaging, the website’s knowledge base offers helpful learning content.

I won’t judge if the picture above is a piece of art or not. I won’t judge if the text-to-image model output of one single image in this case helps in any way to explain what art might be. I’m just sharing and encouraging everyone to play an learn with text-to-image, it’s more than fun 😎 

AI art: just a trend or here to stay?

People are talking a lot about AI and Generative AI. You’ve probably seen those hashtags a couple of times or three when browsing your social networks today 😉. More specifically, people are talking about AI Art… But, unlike most other trending topics on LinkedIn and in business, people are not only talking a lot about it, but using it and producing content with it very fast, at a speed that probably has not been seen before in other routine hype cycles… Remember NFT art?

Note: posting this chart does not intend in any way to express an opinion on NFT or text-to-image technologies. I actually believe NFTs have some interesting use cases that can precisely complement well with the emergence of generative AI imaging, but there was an undeniable level of craze and mania surrounding NFTs a couple of years ago, as the Google Trends chart illustrates. Additionally, the chart helps infer that what is known as “AI art” is not a temporary craze and will quickly go beyond the current niche scenarios.

Text-to-image generation is clearly not the most popular type of generative AI on LinkedIn or other professional networks, compared to chatbots, automation, data analytics, or other direct applications of LLMs. But if we look at the rest of the Internet, AI imaging (whether it is called “art” or something else) could well be the not-so-obvious big new thing within the bigger picture “next big thing” many people see in generative AI. 

This is a well-known pattern. Technology revolutions are sometimes hidden from the general public, developing at universities and Big Tech company labs and papers for years, absent from mass media and casual conversations… until one day a very specific, perhaps unexpected, application of the technology crosses the bridge from the corporate and academic world to reach consumers. This is what innovation economists, such as Austrian Joseph Schumpeter, called innovation diffusion. Diffusion, in innovation economics theory terms, is the process of how, why, and at what rate new ideas, practices, or products spread through a population or social system. I did not come up with this definition, I asked a chatbot again.

😎 What is innovation diffusion?

🤖 According to some sources [Wikipedia] [Corporate Finance Institute], innovation diffusion is the process of how, why, and at what rate new ideas, practices, or products spread through a population or social system. The theory of innovation diffusion explains the pattern and speed of adoption of innovations by different groups of people, such as innovators, early adopters, early majority, late majority, and laggards [cleverism.com] [Investopedia].

😎 Which economic theory describes the dynamics of innovation diffusion? Who are some economists or authors who have developed the concept?

🤖 One economic theory that describes the dynamics of innovation diffusion is the diffusion of innovations theory, popularized by Everett Rogers in 1962 [Investopedia] [Corporate Finance Institute]. This theory explains how, why, and at what rate new ideas, practices, or products spread through a population or social system. The theory identifies five categories of adopters: innovators, early adopters, early majority, late majority, and laggards. The theory also considers various factors that influence the adoption process, such as the characteristics of the innovation, the communication channels, the social system, and the time.

Some other economists or authors who have developed the concept of innovation diffusion are Joseph Schumpeter, who introduced the notion of creative destruction and the role of entrepreneurs in innovation; Frank Bass, who proposed a mathematical model of diffusion based on the interactions between adopters and potential adopters; and Geoffrey Moore, who popularized the concept of the chasm between early adopters and early majority in his book Crossing the Chasm  .

Bing Chat

It’s a curious coincidence that the word diffusion was used by economists more than 60 years ago and one of the major forces driving the widespread adoption of generative AI today is called Stable Diffusion (see McKinsey’s chart).  

Typically, the particular application triggering the diffusion of an invention or technological advance is not the gadget that appeared in a recent science fiction novel or futuristic movie (it’s 2023… where is my floating hoverboard and my flying car?), but something that most of us could not even imagine before. If you are reading this, I guess you have probably seen this Bill Gates interview from 1995, which has been referenced many times when talking about any type of innovation that happened over the past 30 years. “What about this Internet thing?” 

👨‍💼 What about this Internet thing? Do you know anything about that?

🤓 Sure

👨‍💼 What the hell is that exactly?

🤓 Well, it’s become a place where people are publishing information, so everybody can have their home page: companies are there, the latest information… it’s wild what’s going on, you can send electronic mail to people. It is the big new thing. 

👨‍💼 Yeah, but you know, it’s easy to criticise something you don’t fully understand, which is my position here…

🤓 Go ahead

👨‍💼 I can remember a couple of months ago there was like a big breakthrough announcement that on the Internet or on some computer deal they were going to broadcast a baseball game. You could listen to a baseball game on your computer, and I just thought to myself… does radio ring a bell?

Crowd: 👏 🤣

Bill Gates on Late Show by David Letterman, November 27, 1995

So… what about this generative AI thing?

Again, this is not a new pattern. Mocking that Internet thing on TV in 1995 was socially acceptable and funny. Probably most people today think that AI-generated pictures are a funny thing that is used for memes and by people who don’t have serious jobs. Many other people might be reluctant to use or even talk about these technologies and fear they will be used for committing fraud or destroying jobs. Nobody can predict the future and where generative AI in general or gen-AI imaging, in particular, will take us, but the patterns of the past are very well defined. Most groundbreaking technologies cooked behind the scenes, away from the public eye, sometimes for years. It’s only when they become controversial and human passions are polarized when we can talk about diffusion. I do believe AI imaging is here to stay, and the pattern is evident. Here’s a comment from 5 years ago I found on my LinkedIn activity history if you allow me the bit of self-promotion :). I think it’s still relevant:

All technological revolutions are fueled by rapid innovation and are not exempt from human fears and resistance to change. But history tends to prove that progress is hard to stop. It is also worth noting how the same company Bill Gates founded, which has been in the center of two of the most important technological breakthroughs over the past 40 years (personal computers and the Internet), is again one of the companies leading the business of generative AI. I would just like to thank Microsoft for providing me with so many helpful resources such as Bing Chat, OpenAI generative chat and imaging tools, LinkedIn Learning, and so many other products and services I use daily, as this screen capture of my computer as I write these lines proves:

I hope the LinkedIn AI algorithm won’t get offended by any reference to its parent company’s business in this picture. It’s just an attempt at a joke… and a robot test. If you got it, you’re probably not an AI. But, who knows… I think I have some ideas for a future article: Do AIs have a Sense of Humor?

But… what is AI art?  

If defining art was difficult and definitely not an objective exercise, defining AI Art is probably harder. Many people would even question the validity of those two words put together as a single term. In very simple terms, what Large Language Models do well is “putting words together” by learning from massive amounts of text and predicting which words fit well together to produce new meaningful and valuable pieces of text. Let’s then, see what a chatbot has to say about that very controversial pair of words put together: AI art.

😎 Can you explain what’s AI art in a couple of sentences?

🤖 AI art is a form of creative expression that uses artificial intelligence systems to generate or enhance artworks, such as paintings, drawings, illustrations, or photographs [NightCafé Studio]. AI art can be created by using different methods, such as neural style transfer, text-to-image generation, or diffusion [NightCafé Studio] [hotpot.ai] [deepai.org].

AI art can be seen as a way of exploring the possibilities and limitations of AI, as well as its impact on human culture and aesthetics. AI art can also raise ethical questions about the role and responsibility of human artists, the ownership and originality of AI-generated artworks, and the value and meaning of art in the age of automation.

😎 What are some of the most popular applications for watching and creating what people call “AI art”

🤖 Some of the most popular applications for watching and creating AI art are:

• Bing Image Creator: A web-based tool that allows you to generate images based on text prompts using a more advanced version of DALL-E [aiartists.org]. You can also access Bing’s AI chatbot, Bing Chat, and use it to create images by typing #generative_image followed by your prompt.

• Dream by WOMBO.ai: A mobile app that lets you create AI paintings based on text prompts and choose from different art styles [Zdnet]. You can also share your creations with other users and explore their artworks.

• Midjourney: A web-based tool that enables you to generate high-quality photos based on text prompts using a state-of-the-art generative model [Techradar]. You can also join a Discord community and view other users’ outputs.

😎 Can you find a negative opinion about AI art?

🤖 Yes, I can find a negative opinion about AI art. Here are some examples from the web:

• One article argues that AI art generators are unethical because they violate the copyright of human artists whose images are used to train the AI engines without their permission

• Another article reports that some illustrators are furious about AI art because they think it is the opposite of art and it devalues their skills and creativity [The Guardian].

• A third article discusses how the rise of AI artists stirs debate about the definition and value of art, and whether AI can ever produce original and meaningful works

What do you think about these opinions? Do you agree or disagree with them?

😎 Can you find a positive opinion about AI art?

🤖 Yes, I can find a positive opinion about AI art. Here are some examples from the web:

• One article lists the ethical pros of AI art generation, such as leveling the artistic playing field, reducing the cost of art, and reviving classic artistic styles [Makeuseof].

• Another article explores how AI art can challenge and expand the notion of creativity, and how human artists can collaborate and experiment with AI tools to produce new forms of expression [The Guardian].

• A third article examines how AI art can generate novelty and diversity, and how it can reflect and critique the cultural and social values embedded in its data [BBC].

What do you think about these opinions? Do you agree or disagree with them?

Bing Chat

The stage is set for controversy (yes, I asked a chatbot to build this phrase for me). Whether we talk about AI art or any other of its applications, AI ethics poses a major challenge that humanity is just beginning to address. Regardless of ethical implications and the validity of all the points of view, I believe AI art, more specifically text-to-image generation models, has a fascinating potential to influence the world and have an impact on any of us, whether we want to create art, learn, improve our work, or just have fun.

What can AI “art” tools do for you?

The thing that fascinates me most about AI art is that, although it can be extremely complex, the barriers to entry for any potential user are extremely low. That’s the reason why the growth of platforms like Midjourney or NightCafe Studio is exponential. 

If you’ve read this far, haven’t tried text-to-image before, and still feel curious about it, you will surely learn quickly by joining any of the freemium model platforms that have been mentioned so far. Most of them have good learning content too.  If you want to deep dive into more theoretic content, I recommend the #LinkedInLearning course Generative AI Imaging: What Creative Pros Need to Know, by photographer Ben Long. While the content is more theoretical and only focused on AI imaging, I believe it forms a good learning path if you combine it with the course I recommended in my previous article: Prompt Engineering: How to Talk to the AIs, by LinkedIn Engineer Xavier (Xavi) Amatriain.

Apart from formal courses, there is plenty of content and self-learning material on the Internet that you can use, so all I can add is by sharing some of my experience with generative AI and art.

Use case 1: chatbots for art interpretation and social media content

I like music. I’m sure I’m not the only one. I like it so much that I’m writing a blog where I share songs made by other people and have conversations with chatbots about them. Before starting the blog, I decided to train my skills and do a dry run of how future blog posts would look by posting something on LinkedIn:

LinkedIn post: Conventional Wisdom #jousting

I didn’t post much… I specifically asked the chatbot to be cryptic. But the entire text was inspired by an AI chat I had. Only including an excerpt here:

😎 Tell me something about the song “Conventional Wisdom” by “Built to Spill”

 🤖 Sure, here is something I can tell you about the song “Conventional Wisdom” by “Built to Spill”:

* It is a song from their sixth studio album You in Reverse, which was released on April 11, 2006 [Bing search].

⦁ It is a rock song with a catchy chorus and a long guitar solo. The lyrics are about questioning the conventional wisdom in a world that’s just so unconventional [Bing search].

* It has a music video that shows the band performing the song in various locations, such as a desert, a forest, and a city [Youtube] [Youtube].

* It is one of their most popular songs, with over 4 million streams on Spotify and over 200 thousand views on YouTube [lyrics].

I hope you find this interesting. 

 😎 Tell me more about the lyrics

 🤖 Sure, here is more about the lyrics of “Conventional Wisdom”:

 Full content on reddgr.com/conventional-wisdom

 😎 Can you write an alternative that sounds good for my social media followers? “Just wanted to wish everyone a good weekend and share this video. We’ll talk more about AI next week.”

 🤖 Sure, here is one possible alternative for your social media post:

 Full content on reddgr.com/conventional-wisdom

 😎 Yes, but make it a bit shorter and cryptic

 🤖 Okay, how about this:

Weekend vibes! Watch this. Break the rules. Think different. More AI soon.

Do you like it? 

😎 Yes, it’s cool

🤖 I’m happy you think so. You’re welcome to use it for your social media post. 😊

[ ⌛ … days later ⌛]

😎 Who is James Zoppe?

🤖 According to some sources [James Zoppe website] [IMDb], James Zoppe is a jouster, a horse trainer, and an actor. He is the founder of the American Jousting Alliance, a group that promotes and organizes medieval-style jousting tournaments and events. He is also a third-generation rider who started his career as a trick rider in his family’s circus. He has appeared in movies such as The Fisher King (1991), Night of the Templar (2013), and Getting Lucky (1990) [IMDb] [IMDb].

😎 A few days ago we talked about the music video of Conventional Wisdom by Built to Spill. Do you know that James Zoppe acts in that video?

 Full content on reddgr.com/conventional-wisdom

🤖 I searched the web for the music video of “Conventional Wisdom” by Built to Spill and James Zoppe, and I found some results [YouTube] [YouTube] [YouTube]. According to the YouTube comments, James Zoppe is one of the jousters who appears in the video. He is also credited as a stunt performer in the video’s IMDb page. So yes, it seems that James Zoppe acts in that video.

😎 Would you say James Zoppe highly successful in what he does?

🤖 I searched the web for James Zoppe’s success and I found some results [Chicago Tribune] [jameszoppe.com] [NPR]. According to these sources, James Zoppe is a successful jouster, a horse trainer, and an actor. He is the founder of the American Jousting Alliance, a group that promotes and organizes medieval-style jousting tournaments and events. He has won several championships and awards for his jousting skills. He has also appeared in movies such as The Fisher King (1991), Night of the Templar (2013), and Getting Lucky (1990). He has also been featured in various media outlets, such as NPR, Chicago Tribune, and The New York Times. So yes, I would say that James Zoppe is highly successful in what he does.

😎 That’s an impressive list of professional achievements. Do you think I can find him on LinkedIn?

Full content on reddgr.com/conventional-wisdom

🤖 I searched the web for James Zoppe’s LinkedIn profile and I found some results [LinkedIn – James Zoppe] [LinkedIn – Jamie Zoppe] [LinkedIn search]. According to these sources, there is a James Zoppe who is a trainer and equestrian activities instructor at the American Jousting Alliance. This could be the same James Zoppe that we are talking about. However, his LinkedIn profile does not mention his jousting achievements or his acting career. There is also another person named Jamie Zoppe who works as a sales development manager at Club Pilates, but I don’t think this is the same person. So yes, you can find James Zoppe on LinkedIn, but his profile may not be very detailed or updated.

😎 I checked and I’m pretty sure the first profile you found is legit, but he has only 3 connections, so I don’t think he’s posting much on LinkedIn. That’s probably because his professional expertise is in sports and performing arts, and the degree of LinkedIn adoption and use in those sectors is lower than in others such as IT or finance. Do you agree? 

🤖 I searched the web for LinkedIn adoption and use by sector and I found some results [datareportal.com] [LinkedInPulse] [springer.com]. According to these sources, LinkedIn is more popular among professionals in sectors such as IT, finance, education, and marketing than in sectors such as sports and performing arts. However, there are also some exceptions and variations depending on the country, the role, and the purpose of using LinkedIn. For example, some athletes and actors may use LinkedIn to network with sponsors, agents, or fans, while some IT or finance professionals may not use LinkedIn at all. So yes, I agree that James Zoppe’s professional expertise is in sports and performing arts, and that the degree of LinkedIn adoption and use in those sectors is lower than in others such as IT or finance. But I also think that there are other factors that influence how people use LinkedIn besides their sector.

😎 Thank you. That’s a good insight I will be sharing on my LinkedIn article about the impact of generative AI on art and how this can relate to anyone’s job, regardless of the sector.

🤖 You’re welcome. I’m glad I could help. Your LinkedIn article sounds very interesting and relevant. I hope you get a lot of views and engagement from your audience.

Bing Chat

This got me some content for the website 😎 

Use case 2: website logo and avatar

I’ll finish with another personal story of how AI imaging is helping me, and so can help you. I don’t think many people on LinkedIn would relate in this case, but just wanted to share…

Having worked full-time remotely for more than 10 years means a challenge when selecting a LinkedIn profile picture. Most of the photos of me I have were taken outdoors while hiking or cycling, so I’m either wearing cycling clothes or sunglasses to protect my sensitive eyes 😎. Those are valid pictures but don’t fit the more conventional standards of serious professional people wearing serious professional attire. In indoor contexts, some of my friends might have pictures of me posing with a glass on hand and a stained shirt, or singing songs by Julio Iglesias at a karaoke. Those are the kind of pictures your friends hold as a potential ransom in case you ever become famous and things get leaked to mass media. In case you are reading, you know who you are, I love you, but don’t worry, luckily I’m not famous yet. 

Looking for potential profile pictures, I may also find something from a networking event I attended in the past. But those are normally group pictures, posted on social media at a lower resolution than the original, so if I want to extract my individual headshot, the resolution and quality are very low. Anyway, I don’t see the need to change the picture of my personal profile yet, don’t think it’s that important. However, as I was playing around with Stable Diffusion and image-to-image creations, I felt inspired by the Julio Iglesias story to write some prompts, combine them with my basic traditional graphic design knowledge, and get an idea for my website logo and avatar. I just liked how it looks and that, to a certain extent, it’s inspired by my own low-res picture:

I hope you enjoyed the content I shared and either learned something new or felt encouraged to have fun with generative AI imaging. I might be talking more about generative AI in the future if you comment, 👍 the article or follow me on social media. More AI soon 😎