Jun 19 2023
Understanding The Meganet
How much does AI really understand? How much have we made ourselves its slave through our social media usage? And what does this mean for our marketing efforts?
In this final episode of the second series of Talking About Marketing, we turn our attention to the endeavour of human craft as well as our proclivity for finding the laziest way possible to do things with AI.
David Auerbach is a writer, technologist, and software engineer, whose latest book has shone a light on the way our social media usage has helped the large language models that power Artificial Intelligence to have richer insights into how humans intone and communicate.
As we discuss, the current state of technology and social media is a double-edged sword for marketers.
Meanwhile, back in the "good ol' days", we catch a glimpse of all the background effort applied by Michelangelo to enable him to be the artist we admire. Yes, it's another case of an overnight success who put in thousands of invisible hours to make it all look so easy. There's something in this for all of us!!
And we close out with a tip about getting more views for your YouTube videos and why we don't really need to envy AI robots.
Talking About Marketing podcast episode notes with timecodes
01:36 Person This segment focusses on you, the person, because we believe business is personal. What Would Michelangeo Do?
In his book, Resilience, Eric Greitens, a Navy SEAL, shares a series of letters he'd written to another Navy SEAL who was hitting rock bottom and struggling to adapt to civilian life.
What separates this book from other books written by Alpha males, is that Eric is very well read and has a knack for weaving ancient writers, poets, and philisophers into his letters.
The book powers forward and is probably best consumed one letter/chapter at time rather than than being binged because it is dense with observations, suggestions, and lessons.
One that piqued Steve's interest was letter 8, where Eric is talking about the importance of honing our habits. One great insight was that the key to success is actually to train at enjoying training because once you enjoy training, the rest is easy.
But by way of example, he shares some thoughts from the master artist, Michelangelo, who had detailed notes on things to practice and consider when painting. The end result being that when it was time to paint, the artist was in his flow state.
All of us could probably do with extra tweaking around our habits, whether that's the habit of blogging, or reflecting on our enterprises, or making sure we're looking after our health so we CAN keep producing the output we desire.
11:03 Principles This segment focusses principles you can apply in your business today. Meganets
Is there any way of shaking the dark shadows of our addiction to social media? Is AI something to be scared of? And have we helped make the AI monster smarter?
There is a lot we could discuss arising from David B. Auerbach's book, MEGANETS: How Digital Forces Beyond Our Control Commandeer Our Lives and Inner Realities.
Steve decides to centre our discussion around three things of relevance to us as marketers:
The reason why many marketers and users find social media puzzling (it's because we're all in filter bubbles)Why it's impossible for executives to control the junk and lies in social media sitesThe shallow depth of AI "understanding"
David Auerbach is a writer, technologist, and software engineer. He previously worked as a software engineer at Google and Microsoft for many years before turning to writing. He has written on technology, literature, and philosophy for many publications.
One of his opening observations is that people are not constant. When we interact, it is quite chaotic.
Within all that content, there is more poisonous content in social media than ever before in human history. This is because in the past, when we were restricted to interacting with people in our locale, there would rarely be a concentration of people who agreed with our excesses. If we had fringe ideas, it would get blunted by others around us who would challenge it.
However, social media has meant we can connect with likeminded people no matter where they are in the world. This has the result of making dangerous and fringe voices much louder than they would otherwise had been due to Volume, Velocity, and Virality.
For marketers, this is a double-edged sword. On one hand, if we manange to get "in" with our people, we can surf on a really strong current of ready made customers. However, if we upset a certain person or group, they can quickly "pile on" with their fellow travellers; many of whom will comment anonymously.
Secondly, David argues that when we complain that Mark Zuckerberg should do more to stop the poison in our networks, we are wasting our time. He cites many internal papers noting how little control they have over the networks. Every minute of the day, Google conducts 5.7 million searches, Discord uses send 668,000 messages, TikTok users watch 167 million videos, Facebook users share 240,000 photos. Algorithms just can't process the resulting chaos with any great accuracy.
On top of all of this, us humans actually change the machines. Users search, Google watches what they like and don't like, the algorithm updates itself to change results, website owners change content, and the cycle continues. The same happens everywhere.
And the scores given to how offensive certain terms are, need to be updated because often it is not the surface value but the cultural context that makes something offensive. For example, the phrase "I f*cking love you man. Happy birthday" was rated as 93 in 2017 as very vulgar, but that had dropped to 60 in 2021, while "drop dead" only rated as a 40 in 2017 but jumped to 71 in 2021.
He explains how AI is not all that effective because it does not UNDERSTAND human language. The language models being used just work on the probability of what word is more likely to go after this word in this context.
That's why the third point of interest was his explanation for how AI learns using deep learning.
He explained that if we gave AI a pile of pictures of correctly labelled animals and asked them to separate them into separate containers, it would look at what all the dog pictures had in common and all the cat pictures had in common, as well as where the cat and dog pictures differed the most. These similarities and differences are not things humans would understand. It would be certain arrangements of patterns and pixels. It basically is a huge pile of variation about what makes things more cat-like or dog-like, without it being clear as to WHY. But because we have trained it, it becomes accurate. A new unlabelled image comes in, and it is correctly labelled.
Eerily, David says our daily activities in social networks is a gift to AI. We are giving it a huge library of human interaction and, because the networks give us icons to click to show like and anger and love, the AI deep learning machines get an extra head start in categorising.
In fact, David's forecast is that the mega networks are moving towards dumbing down the flair in human language so that we can feed the machines better.
What do we do with this?
It's a reminder that we are about to go through a time of vanilla writing and things will get very bland for consumers. Our job will be to find ways to keep bucking trends and staying human.
Listen to the full conversation.
33:21 Problems This segment answers questions we've received from clients or listeners. Video Thumbnails
If you've noticed videos on YouTube having catchy images with bold titles, it's done for a reason.
Video thumbnails help grab our attention and if you haven't experimented with them yet, now might be the time.
An easy way to start is in Canva which has a set of video thumbnail templates.
36:01 Perspicacity This segment is designed to sharpen our thinking by reflecting on a case study from the past. A Hard Earned Thirst Unless You're A Robot
A fun note to finish on. The Michelob ULTRA - Robots - Super Bowl 2019 Commercial features an AI-driven robot doing everything better than humans, until it comes to enjoying a beer.
This ad has a bittersweet ending as the robot is left bewildered while humans enjoy the human experience of having a beer.See omnystudio.com/listener for privacy information.