The Beginning of (the) Fall
it's the end of the world as we know it...and I feel fine.
Michael Bierut of Pentagram Thinks AI might Not Be So Bad
I’ve only been using digital tools for 24 years, so I missed out on the classical design skills of x-acto knives and paste up. If you’d like a brief history lesson about what it was like to transition from those tools to the first computers in the early eighties, you might like this excellent interview with Michael Bierut of Pentagram. read more on The Dieline
“If someone would have said to me, ‘What if there was a machine on your desk that could generate typesetting immediately,’ I’m not even sure how I would have answered that question," Bierut says. "I would say, ‘But we have typesetters, who I know and trust and like.’"
Why it matters: Computers have been speeding up graphic design work for 40 years
Quickies
Cezary Gesikowski brings us the best prompts from Midjourney Magazine
Rob Young provides the best prompts for Photorealism in Midjourney
Prompt Dervish explains more about color theory and prompt recipes in Midjourney
Edmond Yip does a deep dive into Stable Audio for all the aspiring AI musicians
Joseph Foley opines that Dalle-3 will be the end of prompt engineering as we know it
My Final Opinion On Generative AI Video (for now)
I’ve spent the entire summer playing with generative video tools and the four major players in the middle of the bell curve are RunwayML, Kaiber, Pika Labs, and Decoherance. RunwayML is the media darling that gets lots of press, but I spent a lot of money on credits and received inconsistent results and I’m canceling my subscription to focus on Pika Labs which is free (for now) and gives me better results. The same goes for Kaiber. It gets a lot of buzz on YouTube, but produces the same bad acid trip garbage over and over. Decoherence uses the same Stable Diffusion engine, bur gives the user more control. Dumping my Kaiber subscription and still paying for Decoherence.
Why it matters: Generative AI video is going to be huge in 2024
Microsoft and Google’s Big Announcements
I know this happened like 5 days ago, which is an eternity in AI news, but both Microsoft and Google made pretty big announcements last week. Microsoft’s new Copilot is an OS-wide assistant that promises that you can now find anything on your computer instantly. And while Google isn’t quite on OS, many of us have adapted our lives to the Google ecosystem. This week Bard announced something called extensions, which when enabled, will now allow you to find anything within your mail, docs, flights, maps, etc. So if you’re one of those people who live outside the Venn diagram of Microsoft and Google products, carry on with your relaxed life. The rest of us are probably going to be swept up in a new wave of productivity.
Why it matters: If your boss knows about these updates, the expectations will be higher
YouTube’s Big Announcement
I’ve spent most of the last year trying to upgrade my video editing skills, with the hopes that if I could just solve that last piece of the puzzle I might finally get my YouTube channel launched. It turns out that might have been a waste of time, since each week I read about some new AI tool that can automate this process. More importantly, YouTube announced a new suite of tools that make it easier and faster for just about anyone to create content (mainly shorts). Soon you will be able to generate backgrounds with a prompt, dub your video in a different language, find the right soundtrack with an AI assistant, or receive suggestions on what type of content to create based on analytics.
Why it matters: AI tools that allow more people to pursue their creative passions are a huge win
My Expectations for Dalle-3
Based on what I’m reading, the release of Dalle-3 next month will be a seismic shift in the image generation tool landscape. With all that money I’m saving from cancelling RunwayML and Kaiber, I’m going to pay for a month of Chat GPT plus so I can start using Dalle-3 right away (they work in conjunction). I’ll have a full report on day 1.
Why it matters: Image generation tools will improve exponentially by the end of the year
Dystopian Headline of the Week
Morgan Stanley says that Generative AI will make it easier for people to have 2 or 3 jobs! So if you’re a millennial and you’ve lived through a “once in a lifetime recession” and a “once in a life time pandemic” you’re about to live through a “once in a lifetime economic restructuring”, but hey, at least you’ll be able to hold down three jobs and still make rent.
Side hustles where generative AI can help include affiliate marketing, e-commerce dropshipping, content creation, tutoring, graphic and web design, programming, and renting rooms or other space, according to Morgan Stanley. The average multi-earner is making $8.50 an hour when using generative AI, and $7 an hour without the technology, Morgan Stanley's surveys found.
Why it matters: It doesn’t matter. It’s just awful. Seriously, who thought this was a good idea?
That’s a wrap! It’s supposed to rain all day today, hopefully you’re snuggled up with a nice cup of tea and getting some value out of this newsletter. Did you know I have a paid version as well? If you want a deeper dive and more content and analysis, you can upgrade to the paid plan.