Almost Timely News, 4 September 2022: Understanding the Implications of Stable Diffusion

Warning: this content is older than 365 days. It may be out of date and no longer relevant.

Almost Timely News, 4 September 2022: Understanding the Implications of Stable Diffusion (9/4) :: View in Browser

Download Members Only, my new free study on private social media communities »

Watch This Newsletter On YouTube

Almost Timely News, 4 September 2022: Understanding the Implications of Stable Diffusion

Watch this video on YouTube.

Click here for the video version of this newsletter on YouTube »

Click here for an MP3 audio only version »

What’s On My Mind: Understanding the Implications of Stable Diffusion

This past week, you may have seen an explosion in the amount of AI-produced artwork. Why, and why is everyone talking about it all the sudden?

The oversimplified version is that Stable Diffusion is a new model for generating visual creative using AI. What makes it different than other models is a simple fact: its creators open-sourced it.

To understand the implications of this, we have to understand the landscape a bit and clear up some definitions. When someone says the words “AI model”, what they’re really saying is a piece of software written by AI. All machine learning and AI operates off models. Models are made by taking huge amounts of data and having machines recognize the patterns within them and then be able to replicate those patterns. An AI model is a piece of software for AI to use, in the same way Microsoft Word is a piece of software for you and me to use.

Up until this point, many of the best models for AI image generation have been created by corporations who have a strong vested interest in keeping those models behind closed doors. Google has some. Almost certainly other big tech companies do. The models you’ve heard most about in the news include a model called DALL-E 2, by OpenAI – which, despite the company’s name, is not open or free.

What that has meant thus far is that professionals and amateurs have been able to dabble in the use of AI image generation, but haven’t been able to fully leverage the technology because of the costs involved.

This week, Hugging Face, in partnership with Stability.AI and Eleuther.AI, released Stable Diffusion. Unlike DALL-E 2 and Midjourney (two other popular image generation models), Hugging Face released not only access to their state of the art model, but the entire model itself. This is a huge step, because it means that anyone can download their model and if you have the technical know-how, run it on your computer.

This has, in turn, created an explosion of innovation. People who were previously laboring to create apps and software using paid services like DALL-E 2 suddenly had access to a state of the art model that cost only the compute time to operate it. Overnight, we saw people applying the model to tons of different tasks, and the Internet was swamped by new creations.

What do these models do? In short, Stable Diffusion performs two fascinating tasks: text to image, and image to image. Let’s walk through each of these. Text to image is exactly what it sounds like: given a text prompt, the model attempts to create an image from the prompt. It draws from the huge number of images it learned from to synthesize something new.

Let’s say I provide a prompt like this: “A man sitting behind a desk, talking into a microphone, 4K resolution, photograph“. What does Stable Diffusion spit out? Something like this:

Stable Diffusion output

Image to image does something similar, but instead of just words as the starting point, you provide an image of some kind and the model uses that, plus your text description, to manufacture a new image. Suppose I provide a prompt like this, which I made in Powerpoint:

Stable Diffusion prompt

I think it’s fair to say I’m not going to be winning any prizes in art contests, right?

But that’s enough for an AI to work with, along with a prompt like this: “A mountainous countryside with green fields and a fast-moving river at sunrise, 4K resolution, trending on Artstation, oil painting in the style of Rembrandt”

What did that yield? A bunch of examples, but one that looks like this:

Stable Diffusion image output

You can clearly see the starting point reflected, particularly in the colors. My inept, grade-school art was transformed into something much more artistic by the Stable Diffusion model. That’s what makes them so powerful, so amazing – it’s why everyone’s been talking about them all week long.

What does it take to use the model? There are already tons of prepackaged web services that require no technical skill and have nominal costs. You can try a free demo here. I’ve been using the model on Google’s Colab machine learning environment; for $9.99 a month, I get access to an absurdly powerful computing environment and then with Stable Diffusion, instead of paying per image, I get to generate as many images as I want.

What would you use this software for? Any kind of image generation, honestly. For example, I wrote a blog post recently about student loans and I told Stable Diffusion to make me an image of a pig in a library, rolling around in money, and it came up with this:

Stable Diffusion pig

At a bare minimum, using it to generate blog headers is one application, though that’s a bit like flying an F-35 fighter plane to the grocery store for your weekly groceries.

Think about any place where you need some kind of image – blog posts, eBooks, etc. Instead of resorting to the same old clip art, you might now turn to a model like Stable Diffusion to create something custom, something unique. And this applies to all images, anywhere. Remember the stock painting of the seashell or the sailboat or the random garden that’s in literally every hotel room ever? Those could be replaced with unique, custom art now.

So, with that in mind, what are the implications? First, as we’ve seen with deep fakes, any time you release a powerful model, there will absolutely be misuse of it. We’ve seen some already; if you download the Stable Diffusion model, it can create racist or pornographic imagery with the same ease it creates any other image. It’s a tool, a piece of software that has no moral compass of any kind; we as the users of the tool must be the ones to guide it. You’ll note there are also biases in the model; look at the rendering above for the prompt “a man”. The model was trained on data where white men with beards were most commonly tagged as men, so that’s what the model thinks a man is, without further qualifiers and descriptors.

Second, there are clear economic implications. As we’ve seen with AI-generated writing, the bar of minimum quality has been raised. Four years ago, machines basically spit out word salad. Today, they can write competently. What they generate isn’t amazing, but it’s just slightly better than mediocre – which means if you’re a mediocre writer, machines can do your job better than you can now. The onus is on you to skill up. What AI did for writing, it’s now doing for art.

You can tell based on the starting prompt above that I’m not a particularly talented artist. The machines did a far better job of turning my prompt into real art than I ever could have – which means that if you’re a professional artist with average or below average skills, your livelihood is now in peril from AI that can do your job better than you at substantially lower cost.

Third, as a marketer, you no longer have any excuse for bad visuals in any of your marketing content. At a bare minimum, your visuals should always be at least machine-made mediocre, if not better. Tools like Stable Diffusion have set the bar for all of us in terms of the minimum level of quality expected by users.

Fourth, Stable Diffusion itself is and will continue to be responsible for a massive flood of new art. This is great news if you love to consume art. This is really bad news if your job is selling art. Either way, tools like Stable Diffusion pave the way for an enormous glut of new imagery.

Finally, and what I think is most important in the long run, tools like Stable Diffusion unlock our potential as artists. I’m not an artist, as I said. But I have artistic ideas, and Stable Diffusion allows me to be an artist by proxy. I am no longer limited by my hilariously bad skills – I can provide a bad prompt to the software and still get good results out of it.

One last philosophical note: a lot of people have asked whether or not AI-created art is art. This question implies a lack of understanding about what art is. Art is expression, and the value of art is in the eye of the beholder. If it moves you, if it speaks to you, if it evokes something inside you, it is art, regardless of who made it. There’s some art I just don’t get, but to others it speaks to them, so it’s art – just not art for me. Banish the concept of art/not art from your vocabulary. There’s art that’s for you and art that’s not for you, but it’s all art – no matter who or what made it.

Share With a Friend or Colleague

If you enjoy this newsletter and want to share it with a friend/colleague, please do. Send this URL to your friend/colleague:

https://www.christopherspenn.com/newsletter

ICYMI: In Case You Missed it

Besides the new Google Analytics 4 course I’m relentlessly promoting (sorry not sorry), I would recommend the piece on privacy-friendly Google Analytics alternatives.

Skill Up With Classes

These are just a few of the classes I have available over at the Trust Insights website that you can take.

Premium

Free

Get Back to Work!

Folks who post jobs in the free Analytics for Marketers Slack community may have those jobs shared here, too. If you’re looking for work, check out these five most recent open positions, and check out the Slack group for the comprehensive list.

Advertisement: Google Analytics 4 for Marketers

I heard you loud and clear. On Slack, in surveys, at events, you’ve said you want one thing more than anything else: Google Analytics 4 training. I heard you, and I’ve got you covered. The new Trust Insights Google Analytics 4 For Marketers Course is the comprehensive training solution that will get you up to speed thoroughly in Google Analytics 4.

What makes this different than other training courses?

You’ll learn how Google Tag Manager and Google Data Studio form the essential companion pieces to Google Analytics 4, and how to use them all together
You’ll learn how marketers specifically should use Google Analytics 4, including the new Explore Hub with real world applications and use cases
You’ll learn how to determine if a migration was done correctly, and especially what things are likely to go wrong
You’ll even learn how to hire (or be hired) for Google Analytics 4 talent specifically, not just general Google Analytics
And finally, you’ll learn how to rearrange Google Analytics 4’s menus to be a lot more sensible because that bothers everyone

With more than 5 hours of content across 17 lessons, plus templates, spreadsheets, transcripts, and certificates of completion, you’ll master Google Analytics 4 in ways no other course can teach you.

Click/tap here to enroll today »

What I’m Reading: Your Stuff

Let’s look at the most interesting content from around the web on topics you care about, some of which you might have even written.

Social Media Marketing

Media and Content

SEO, Google, and Paid Media

Advertisement: Private Social Media Study

I’ve just published a new study called Members Only: The Rise of Private Social Media. What’s it about?

What private social media platforms are and aren’t
How private social media platforms affect your marketing
How many people use private social platforms like Slack and Discord
What does and doesn’t work for marketing to these communities

Download your copy of the study for free here. You can get it in PDF, ePub for ebook readers, or MP3 if you want to listen to it instead.

Tools, Machine Learning, and AI

Analytics, Stats, and Data Science

All Things IBM

Advertisement: Ukraine Humanitarian Fund

If you’d like to support humanitarian efforts in Ukraine, the Ukrainian government has set up a special portal, United24, to help make contributing easy. The effort to free Ukraine from Russia’s illegal invasion needs our ongoing support.

Donate today to the Ukraine Humanitarian Relief Fund »

Events I’ll Be At

Here’s where I’m speaking and attending. Say hi if you’re at an event also:

Content Marketing World, September 2022, Cleveland, OH
MarketingProfs B2B Forum, October 2022, Boston
Heapcon, November 2022, Belgrade, Serbia

Events marked with a physical location may become virtual if conditions and safety warrant it.

If you’re an event organizer, let me help your event shine. Visit my speaking page for more details.

Can’t be at an event? Stop by my private Slack group instead, Analytics for Marketers.

How to Stay in Touch

Let’s make sure we’re connected in the places it suits you best. Here’s where you can find different content:

My blog – daily videos, blog posts, and podcast episodes
My YouTube channel – daily videos, conference talks, and all things video
My company, Trust Insights – marketing analytics help
My podcast, Marketing over Coffee – weekly episodes of what’s worth noting in marketing
My second podcast, In-Ear Insights – the Trust Insights weekly podcast focused on data and analytics
On Twitter – multiple daily updates of marketing news
On LinkedIn – daily videos and news
On Instagram – personal photos and travels
My free Slack discussion forum, Analytics for Marketers – open conversations about marketing and analytics

Required Disclosures

Events with links have purchased sponsorships in this newsletter and as a result, I receive direct financial compensation for promoting them.

Advertisements in this newsletter have paid to be promoted, and as a result, I receive direct financial compensation for promoting them.

My company, Trust Insights, maintains business partnerships with companies including, but not limited to, IBM, Cisco Systems, Amazon, Talkwalker, MarketingProfs, MarketMuse, Agorapulse, Hubspot, Informa, Demandbase, The Marketing AI Institute, and others. While links shared from partners are not explicit endorsements, nor do they directly financially benefit Trust Insights, a commercial relationship exists for which Trust Insights may receive indirect financial benefit, and thus I may receive indirect financial benefit from them as well.

Thank You!

Thanks for subscribing and reading this far. I appreciate it. As always, thank you for your support, your attention, and your kindness.

See you next week,

Christopher S. Penn

Comments

One response to “Almost Timely News, 4 September 2022: Understanding the Implications of Stable Diffusion”

INBOX INSIGHTS, September 7, 2022: Taking Care of Data, Natural Disasters – Trust Insights Marketing Analytics Consulting

September 7, 2022

[…] Almost Timely News, 4 September 2022: Understanding the Implications of Stable Diffusion […]