Week 2

Thanks for the votes of support and confidence, everyone. Knowing you are watching really does help me keep focused on the plan. As a reminder to anyone who is new, this is my own personal journal of what I’m learning about AI for marketing, along with my observations about what I think the impact(s) will be. I don’t pretend to be an expert, so please recommend other resources for me in the comments.

So what am I learning this week? (Aside from the ins and outs of setting up my own NAS because my beloved Apple Airport Time Capsule apparently has a death wish).

1. How AIs Learn

I first saw this video before the pandemic at an AI for marketing workshop given by Jim Sterne. I’ve shared it with friends and family on the regular ever since. If you’re new to this subject, let me recommend it to you personally, and you can show it to your parents and kids. It provides a solid overview of the basics of how the learning works, without digging too much into the “I hope you like linear algebra.” The footnote with more details is also worth watching. (total sidebar: I relied heavily on Grey’s “Spaceship You” during the Pandemic. Super helpful framework).

2. Interesting Tweets/Accounts about AI, with or without Marketing

I’ve been on Twitter a long time and was happy to see their algorithm starting to put AI tweets into my feed this week. Musk’s personal antics notwithstanding, I don’t remember old Twitter being this responsive so quickly to my changing interests. Here are some posts I appreciated, and some I’m bookmarking for later. I followed all these folks on Twitter, subscribed to their Substack(s) so you can find them on my profile, and/or followed their podcasts.

• AI Snake Oil Substack: Princeton University Computer Science Prof Arvind Narayanan is writing a book to separate the AI wheat from the chaff while he writes a book. I’ve subscribed and am eager to see where he goes. This thread he’s written to comment on the FTC’s monitoring of AI claims and discrimination in AI is worth every executive’s time.

• Nathan Labenz, who is “building AI video creation” at Waymark, also hosts a podcast about AI that I’ve put into my playlist. He had two great threads this week that I bookmarked. This thread is a link to a number of threads that are super instructive for beginners, including a link to a pretty interesting research assistant called Elicit.org. In the ensuing tweet storm, he muses about what’s coming with ChatGPT4, which is rumored to be in release quite soon. Based on what I have heard from my network, he’s pretty on the money. He also posted OpenAI’s leaked “Foundry” (aka enterprise) pricing and commented on its implications.

• Justine Moore, a consumer partner at A16Z, chronicles Reddit’s ongoing, and successful attempts to jailbreak ChatGPT with a prompt called DAN(do anything now).

• Ethan Mollick, a Wharton professor, posted a monster long tweet and tweetstorm about what’s going on with Bing both during and after “Sydney.” I was blown away by the example of Bing improving its writing on the fly after being told to read Kurt Vonnegut’s advice on writing. The entire tweetstorm is worth your time. Here’s the Substack write-up. I found Ethan Mollick through a tweet from Rob Lennon about requiring the use of AI in the classroom, which Ethan responded to, including a great post about what was learned.

• Here’s Rob Lennon’s Twitter feed, which carries some great practical tips and curation as well. His blog has a bunch of broken links right now so maybe wait a bit before checking that out.

3. AIForAnyone.org

AIForAnyone.org was founded by a pair of well-trained brothers who are passionate about making sure everyone can start to understand what is coming. They have funding from the Mark Cuban Foundation. I’ve signed up for their newsletter and listened to their 1-hour “intro to AI” class. It’s targeted at high school students, and it’s very broad and general so it’s at a level everyone can appreciate but doesn’t address marketing specifically. Here are some thoughts I had while listening.

• If you struggle to understand what the future looks like, pay attention to news coming out of China. Some of the tech there is frightening and unethical by American standards; but the reason to understand it is so we can work together to create the future we want.

• There’s a point where AI stops being “magical” and that means it stops being defined as AI. So the goalpost is constantly moving—another reason to be always learning about it.

• I’m still learning, but it strikes me that one way to understand AI is to think about what specific parts of the brain a particular tech is trying to replace. Most of the things we think of as AI are “ANI” (narrow intelligence) which is basically about a computer being good at exactly one thing. Even Siri falls into this category. Even ChatGPT is good at exactly one thing: scanning the internet and quickly summarizing what it finds, within imposed limits and boundaries.

• Big chunks of this overlap with the “How AI’s learn” below, but are more rudimentary. I liked many examples of how the AI can learn that differ.

• AI is only as good as the data it learns from. I know this is obvious but I think it can’t ever be stated enough.

• When AI is embedded into any technology we have to evaluate the impact of the combined technologies. In other words, it’s what we do with AI that will determine its usefulness and/or danger.

• There is no job or industry that won’t be impacted, but it will be a question of when and how hard. This is why the fundamental understanding of AI is so important for all of us.

• But we have faced this before as a species—computers being the most recent and applicable example. New jobs are being created that wouldn’t have existed at the same time that some are destroyed.

• We will need more people from the humanities than ever to help solve the thorny problems that are going to arise. Take heart, my fellow liberal arts majors!
  

4. Privacy and Ethics

I’ve been working on issues of privacy and ethics in marketing data since 2015. This week I came across this paper from MIT Technology Review about questions relating to deleting problematic data sets. Marketers will need to be thinking about the data their companies collect and be fierce advocates for its ethical use. Based on the conclusions of this article, here are some implications for marketing.

1. Before collecting data, ask how it will be used-really. We should no longer be “collecting all teh thingz” just to do that.

2. Ask vendors and agencies where they get the data they use, and make sure your agreements address the deletion and/or transfer of data collected on your behalf.

3. Data governance isn’t just for public companies anymore. Even startups should have controls in place for sensitive data the way they do for money. If someone needs to access a sensitive set of data, there should be some friction and a record of that access.

4. Larger companies should begin to develop a set of guidelines for the collection, retention, and deletion of data. Marketing should be a key stakeholder in these conversations as engineering rarely understands the unique needs of this department.

5. I’ll talk more about this in future posts, but marketing teams should start having at least one expert in data science/engineering. Not a shared resource, but their own team member, because needs are different.

6. The good news, if you’re an ethical marketer, is that you are taking on far less risk if you’re using cohorts to segment and target.

Playing With Midjourney

This week I learned how to add an aspect ratio and style to my images, and how to access all the images I created. I also spent time reading the Midjourney prompt guides. I learned that I can do this same work in a dm with the bot, so I don’t get lost in the free flow of images in the newbie room. I could sit here all day with this tool. I have always loved working with designers and have a vision of what I want, but lacked the technical skills to make it happen.

She’s evolving. I feel like I might be being influenced by “The Last of Us.” Check out her sardonic style. I’m kinda loving the specs on the owl, too.

Prompt: graffiti style, happy ten year old girl with short dark hair and glasses, in a friendly forest, with a helpful pet horned owl, --aspect 3:2

Oh, and I was on a podcast this week!

Monique Mills had me on her podcast, “The Unpolished MBA,” and I learned about why Jasper.AI is her favorite generative writing product.

What’s Next

I’m taking my first trip after my cancer dx this week, so I won’t be blogging for a bit. However, I do have Jim Sterne’s “AI for Marketing” in my bag and my plan is to write a book review based on my notes when I return. I read the first chapter over the weekend, and it’s a fantastic primer for anyone who wants to get a handle on the basics.

If you work for a company that’s integrating AI into its marketing technology, I’d like to talk with you, on or off the record. Please send me an email or ping me on LinkedIn.