Week 4???
Or, WTF did you all do while I was out?
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.
A girl takes a ten-day vacation and AI has a field day! The girl gets slightly distracted by a potential banking crisis, comes back, gets Covid, and tries to make sense of it all. I have a lot of things to share including my book report, but I’ve deliberately kept this installment short so you will engage with what I think is one of the most salient pieces of writing on AI and language I’ve found.
What I Learned This Week
1. GPT4: The Humans Bite Back
It showed up even earlier than my sources suggested, but as powerful and disruptive as was promised. I’m somewhat enjoying the freakouts. Still digging into all that’s being written and I’ll come back to this shortly, I promise.
In the meantime, join me in the schadenfreude for a second, but not for long. Because Marketers, think long and hard about anything you’re doing that you rely on friction to evade punishment for. The era of spammy growth tactics is coming to a close, and not a day too soon.
2. In Which I Interview A Potential Web Analytics Employee Named Bing
After spending a solid decade-plus on Chrome, I’m now using Microsoft Edge again as I got access to the New Bing. Sidebar: I don’t think the tech press gave enough credit to Microsoft for putting Reid Hoffman on their board after the LinkedIn Acquisition, nor do I think most people appreciate the extent to which he influenced Microsoft’s ability to get its paws on GPT. They certainly are playing the long game and using it to pull us into their browser has to be scaring the stuffing out of Google. (As it should. It was becoming bloatware.)
Now, I don’t think it’s productive to play gotcha with AI chat in this forum, because I want to stay focused on what marketers need to know. But I want to illustrate the state of the art right now, so I used a case study that I solved for a client many years ago: “what is the cause of my site’s low bounce rate if that bounce rate is not low due to user experience?”
In Bing’s defense, this is a bit of a trick question, but it is an answer that is obvious to anyone with expertise in web analytics. I’ll reveal the answer at the end (at the cost of losing one of my absolute favorite interview questions, but all in the name of progress).
Here’s what Bing had to say:
I played back and forth with it a bit.
Yes, it got caught up in the bland “average” problem as I call it. (I’m also starting to appreciate the definition of this as “bullshitting.”) So I tried to help it:
Then I tried arguing with it in the way I would with a digital analytics professional.
I gave it the information it requested, and it still hadn’t accurately answered the question, and given that we were now in an infinite loop, I gave up. (Bing you can’t get hired as a web analytics pro just yet).
(Here’s the actual answer: The Google Analytics tag was set up incorrectly. In this specific instance, anyone logging into the site was redirected to a new GA property, and thus was logged as a “bounce.”)
I’m sure this will get picked up someday soon and be included in the list of potential causes of a high bounce rate; but it shows a very important gap in AI at present: the inability to make leaps on knowledge not in its current training set. My point is that there’s no end to the possibility of leaps, and AI will thus remain behind humans in many critical ways that matter for those of us who make our living being curious and thinking.
This Week In Privacy And Ethics
I’ve been working on privacy and ethics issues in marketing data since 2015. I also chair the committee on Privacy and Ethics for the Digital Analytics Association, a committee I created in 2020.
Literally, the night before my trip, this piece in NY Magazine came out about Emily Bender. I have been itching to share it with you since then. It’s an absolute must-read for anyone who wants to start wrestling with ethics around AI.
“But language — how it’s generated, what it means — is about to get very contentious. We’re already disoriented by the chatbots we’ve got. The technology that’s coming will be even more ubiquitous, powerful, and destabilizing. A prudent citizen, Bender believes, might choose to know how it works.”
I’m deliberately keeping this installment of Let’s Get Real short so that you will all take the time to pour a beverage and sit down with this piece. It will teach you so much about the issues at hand. We are marketers, and we deal in language. We have a responsibility to understand.
Playing With Midjourney
In honor of the allegory Emily Bender is famous for, and as an additional enticement for you to read this article, I prompted an image about an Octopus eavesdropping on a conversation. I love art and I love working with designers, but couldn’t create it myself. I can’t begin to describe the satisfaction I’m having with these tools. (what’s up with her feet and the Octopus’ extra eye, though? that’s called, no joke “an AI hallucination” though there’s a lot of debate around it; similar to what I said above, this is the visual version of “bullshit” that the chat does in the LLM called Bing or GPT.)
What’s Next
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.