Issue #19: Insane image & graphics tools, AI copy fail, how to optimize for AI

Good morning.

Generative AI for text is great.

But have you done much with images yet?

So far, Midjourney has been the hottest tool in town. Whole articles have been written on the death of Photoshop and how Adobe is “in trouble”.

It’s easy to have strong opinions in the Age of AI. But few have the facts.

As I keep saying:

Generative AI is coming into everything you touch.

Various types of “AI” have been inside your smartphones and laptops for years. It’s just been hidden.

If you’ve done any kind of media buying the past few years, you already know that Meta, Google, et al. have been using Machine Learning for years—and you’ve been training their algorithms and “AI” for years, for them, too.

But now it’s all coming to the surface—and as part of anything and everything you do as a marketer.

Biggest reveal the past few days?

Adobe Photoshop is introducing a new generative AI tool called Generative Fill, which uses Adobe's AI image generator Firefly to extend images and add or remove objects using text prompts.

Adobe claims that the model is trained on content it has the right to use, making the generated images safe for commercial use. We’ll see how that plays out.

Speaking of images, Google is not slowing down, either. They introduced Product Studio at their recent Marketing Live event. This tool lets you create product imagery with Generative AI—without the added cost of new photoshoots.

In the world of open source, Stability AI open sources its AI-powered design studio. This is good news, as it’s making powerful tools widely available. A handful of companies should not dominate and dictate what’s available for everyone else.

We’re already the point where all you need is one, maybe a couple, photos of a product and you can generate endless product photos with a few clicks—in any setting, with any background, changes, lighting, and whatever else you can think of.

Pretty much nothing you see online soon will be “real”.

And not much you listen to, either. Spotify is developing bots to mimic your favorite hosts, who will then read ads.

More fun ahead…

In today’s issue:

• Optimizing for Generative AI (like SEO but different).

• Everyone will soon have their own LLMs, for their stuff.

• Google's revenge tour continues, doing all the things.

• Making marketing apps and software with no-code got easier.

• Why you need access to the Code Interpreter (with ChatGPT).

• Excellent (McKinsey) report on "AI-powered marketing and sales".

• AI can now create demos of any product with a few clicks.

• Wendy's serves up BLEEP BLOOP BEEF and fries.

Let’s dive in.

BIONIC JUICE

As you’ve heard me talk about (technically, you’ve read it in these emails), Generative AI is being integrated into everything and anything.

We’ve already covered how Google is changing search, and by extension, SEO:

Just as search gave rise to SEO, Generative AI for search will give rise to…

AI Optimization (AIO)?

Generative AI Optimization (GAIO)?

Who knows. But we’re already seeing ‘thought leadership pieces’ on how you can optimize for Generative AI

Like here, in this Forbes piece:

From this perspective, learning what “The Rise Of The Algorithmic Customer” is and how to integrate it into your market research is important.

Even if you look past this piece being a sales pitch, there’s something to be said for optimizing what you publish online for Generative AI in search results, recommendations, and various AutoGPTs or other tools roaming the internet.

Keep an eye on this. Wouldn’t it be great if you and your company showed up in a generic prompt for, say, diet pills? Accounting software? Stock recommendations?

Now, let’s take it up a notch.

Why not allow AI to build an appetite around your entire company—using AI?

Get ChatGPT familiar with your company’s expertise, tone, and brand structure.

We already know GPT-4 is capable of calculating business strategies based on social media reach and determining outcomes. And now that it’s no longer limited to 2021 data and can scrape the web for updated info, the output is far more accurate.

But what if you could not just have a bunch of prompts ready to copy and paste—but instead fine-tune an open source LLM, or use GPT-4 with embeddings, directly tied with your marketing or sales materials?

It’s like having your own, internal “GPT”. I call mine, for this newsletter, BionicGPT.

I have several “GPTs” running locally and in the cloud, for various tasks and projects. I’ve spent the past few years helping companies either train their own LLMs, or fine-tune existing ones for their purposes.

And the future of LLMs is heavily trending in that direction: your own LLMs, inside your company, fine-tuned or trained on proprietary data. 

Say goodbye to spending hours pouring over market research and click-through rate data, and use AI to support your team, answer questions, help with creative work, troubleshoot issues, and brainstorm ideas with impeccable accuracy.

That’s the power of LLMs. And they’re not getting dumber—they’re getting smarter. You can even dramatically reduce their tendency to hallucinate.

LATEST DISCOVERIES

Did you get into Google’s annual I/O conference yet?

Alphabet CEO Sundar Pichai had one simple and clear message on the main stage:

AI is the future of everything.

Sure, not groundbreaking.

But for a company who was struggling to roll out their own AI and Machine Learning tools, and who’ve kept their research and development under lock and key for years—they’re shifting hard into all things AI.

And they are going after everyone, including your iPhone.

Android’s latest development, run by Google’s Codey, makes Apple pale in comparison with their latest bionic upgrade.

Hyperintellegent voice recognition, context image and video search, and superpowered security for your data is the next generation of smartphones powered by the G2 chip. 

That’s the world’s first AI-powered processor, specially manufactured for Google.

All that, stacked with all the daily roll outs in the Google suite, shows how they’re completely reinventing themselves with a much-needed dosage of generative tech. You can check the daily updates here.

A new “magic editor” allows you to remove unwanted nouns from your photos and scour the web for that perfect couch you saw on Pinterest with an Image Prompting search.

If you’re a marketer who's ever dealt with editing images—even with Canva—this will be a serious image tool contender—especially if you’re using Canva.

This is just a glimpse at what is changing for both the common consumer and marketing professional. But did Google do a little switcha-roo there?

Is this reverse psychology with “data marketing”—marketed as more privacy for the consumer so they actually tell us more, anyway? It’s almost like Google knows a thing or two about collecting data.

Anyway, amongst these new features, there’s one epic stand out:

No-code required app creation tool.

It utilizes Bard’s code capacities with Studio Bot. It can generate, debug and explain syntax in layman’s terms, turning those app development pipedreams into a reality.

The changes will make the most difference for small businesses, who will use it to upscale their product portfolio without costly training and outsourcing. Best of all, it’s totally free.

Marketing and tech have blended for years. For a long time, if you were a technical marketer, you had an edge.

Consider that edge either gone—or it’s dropping the threshold and anyone, literally anyone, can create an app or other software. This will propagate across the marketing tech board and tech won’t be an advantage.

But even if you’re not looking to branch into an App Store, Bard’s latest update makes it a fantastic coder on its own merit.

With built-in real time Internet access and voice input support, Bard may finally be catching up with ChatGPT.

It backs each answer source, generates three unique drafts of each response and can export its output without changing the format.

But if those promises are going to live up to their potential, the tech behind Bard needs to be better than good.

And as VentureBeat shares “Shakespeare isn’t the power behind the Bard—it’s PaLM 2.”

Why? PaLM 2, Google’s bespoke LLM, is much stronger in logic and reasoning compared to the leading neural network models, according to Pichai.

To start, it has a serious advantage in data sources alone, as it continually draws information from the internet, and most importantly, doesn’t limit itself to English.

Comprehensive support is already available in Korean and Japanese, with more languages launching in the coming months.

These changes are evidence that Google has clearly deliberated long and hard about how to best position itself as a worthy ChatGPT competitor.

So, does that mean now ChatGPT is old news? No. Not yet. Because ChatGPT keeps getting upgrades.

ChatGPT’s extension, Code Interpreter, totally blows the default version of the app out of the water. And even better, It’s only $20 per month.

Those questioning the significance of programming in marketing—think bigger.

Code is so much more than software development.it’s essential for effective marketing strategies.

In one example, professional audience strategist David Boyle demonstrated how to use Code Interpreter to analyze a spreadsheet of music markets to tailor an impressively credible business strategy.

But before you replace your data analysis department with chatbots—remind yourself that AI still has a long way to go.

Just look at HBO's latest national marketing campaign. They recently launched their rebrand, under the mononym Max. Effective name, sure, but they’ve already been caught out for sloppy copy.

Seriously, just look: “Max is already using AI to write their marketing copy.”

Remember, there’s a time and a place for AI.

Amorai, a digital relationship coach intended to “help people grow and foster real-life connections” recently came under fire also due to the fact the app is, in essence, totally oxymoronic.

According to Carnegie Mellon University's social intelligence specialist Maarten Sap, the idea of an AI-powered relationship coach can be downright “dangerous.” 

LLMs learn by training. Input is processed, uploaded to a database, and used as a reference to interpret the human psyche.

That’s why, when AI gets involved with human psychology, there’s a dangerous amount of room for error.

Just remember when Bing added ChatGPT support? Within weeks, it began urging users to divorce their spouses and run away with them instead.

That’s exactly why human touch is a necessity for AI to reach its full potential.

But you still shouldn’t overlook the potential.

With AI’s accelerating complexity, dwindling cost and rapidly increasing ease of access, a fifth of sales team functions can already be automated with current technology, according to McKinsey’s latest report.

This means, and I repeat, in marketing, more than any other industry, you can’t afford to not use AI.

And it’s not just prose, language and good copy. AI can now create incredible demos of any given product with just a few clicks.

And that potential knows no bounds. PaLM 2 has given Google the leverage it needs to pair up with a range of huge names, including a partnership with Wendy’s.

Starting next month, an Ohio Wendy’s outlet will run a trial. Each interaction with a hungry punter will train the AI to understand the language customers use when ordering food, while factoring in background noise and attempting to upsell orders.

It does beg the question:

What won’t AI transform?

Everyone from drive-thru servers to Python developers have some serious competition arising. But just remember: capturing an audience is a cognitive exercise.

Working with AI is a creative exercise. It's a combination of structure, freedom, and feedback that provides an environment where ideas can flourish.

Let’s double down on the importance of AI and creative freedom.

It’s a scary time to be at the heart of the world’s most rapidly transforming industries.

But remember:

You’re better off evolving and growing, than risk stagnation and decline.

Funding:

€18M: Vienna's Prewave has already been using AI to help big shots like BMW, PwC, and Lufthansa predict, find, understand, and categorize risks within supply chains. Their latest funding round aims to launch the tool across the U.S. and Asia.

$250M: Money talks in the no-code application development platform—just look at Builder.ai raking in cash from Qatar Investment Authority (QIA) and an undisclosed amount with Microsoft, bringing their number to $450 million, with 2.3x revenue growth claims.

BIONIC TOOLKIT

🤖 Cheat Layer is the ultimate productivity tool for marketers.

It streamlines your workflow, for easy access and organization of that crucial marketing information.

Store and retrieve cheat sheets, notes, and snippets of marketing knowledge in one centralized location. Use smart search and tagging to find information quickly and collaborate with your team effortlessly.

 🤖 Is your website looking dull? Let me introduce you to Crito Design.

Layout suggestions and automated color palettes are intelligently generated from your existing branding—meaning you can craft professional-quality graphics, logos, and presentations in a flash.

Then, triple those CTRs with AI-powered code generation for the perfect user experience in seconds.

But the best feature is the emphasis on collaboration. The site can share project roadmaps, gather feedback, and unleash unparalleled teamwork in real-time.

Great templates, an amazing UX and a super-powered AI—what could be better?

 🤖 With all those new customers, you might need a helping hand. There is an AI for that: WebWhiz, the ultimate customer support wizard.

It pulls data directly from your website for tuning ChatGPT. Those parameters are translated into demographics for the perfect chatbot for you, your clientele and your business.

Built-in Javascript support means adding the bot is easy. A few simple lines of code adds the emotionally intelligent, lightning fast customer support representative to any site in seconds—with 24/7 availability for those pesky late-night shoppers.

Whatever you choose, when you make the most of AI, you win.

THE LAST BYTE

Large Language Models, like GPT-4, are fundamentally changing the way we interact with computers and technology.

It’s easy to be dazzled by endless text, video, images, and audio at your fingertips.

But what’s happening with AI and Machine Learning goes far deeper:

So far, technology has been used on tech’s terms.

But with LLMs, we can now interact with and use technology on human terms.

Through voice or text alone, you can prompt LLMs (and models for video, images, and audio) for whatever you want—and you’ll get it. You don’t even need to know how to code anymore. Just describe the kind of app you want made and ChatGPT, for example, will code it for you.

Large Language Models are the new Software in the Age of AI.

In the same potential for unlimited creativity and creation, through your words, also lies the greatest danger:

Who controls the LLM?

Whoever does is in charge of language—and your ability to express and formulate your thoughts as you’re using an LLM.

This is one reason I’m excited about open source alternatives.

And it’s why I’m adamant about you using these tools, as tools, and for good things.

See you next week,
Sam Woods
Editor