99 Brown Balloons
You have to laugh, even when your nuclear-armed neighbour is trying to drop hot air balloons full of poo on you.
My brain and I, we are not friends. We are classmates doing a group assignment called life, and it's not going great.
– Frederik Backman
Dear Reader
Like Backman, my brain and I have definitely not been friends this week.
This newsletter is really a collage written by different versions of me. I tried getting ChatGPT to help, but generative AI is a cognitive accelerator, not a miracle worker.
So we start with what was going to be the end piece, because it is funny. Or at least easy to laugh at. I know how South Korea feels.
99 brown balloons
This week Pyongyang has launched a series of poo-balloons designed to drift into South Korea. Analysts say it is retaliation for leaflet drops of anti North Korean propaganda.
However, no one has ruled out the possibility that is a satirical comment on the amazing campaign pledges by the UK’s Conservative party in its opening expectoration of the General Election. For foreign readers, these have included angry-old-person friendly ideas like reintroducing compulsory national service for 18 year olds.
Over-hype
The BBC this week reported on a study by Reuters of AI use around the world. The angle they chose was that only 2% of people in the UK are using AI every day! That’s nothing! Count Binface could do those sorts of numbers in a General Election!
Over. Hyped.
Over.
We can move on.
Except measuring how many people do a thing in a single day is a matter of context.
Other things that “only” 2% of the UK does every day include gardening, DIY and visiting museums.
Here are some other things that only 2% of people in the UK in the UK do every day:
Gardening
DIY and home repairs
Attending theater, musicals or live performances
Visiting museums
And yet there is hype isn’t there?
Is AI over-hyped? Well, that would suggest that there was a correct amount of hype. Do we mean excitement? Expectation. The best mantra is from William Goldman writing about Hollywood in the 1980s: “Nobody knows anything.”
Here are some problems that beset AI as experienced through big tech products:
Trying to fit a 2024 emerging technology into 2010 products. The shapes don’t match.
The potential is clear, but we don’t know what the shape will be or what it will be called yet.
Explosive, radical change doesn’t fit with the attention span of news, markets, investors, or decision-makers.
If you experience it or read about the discoveries scientists are making, then it’s hard not to not be excited.
The big Google, Microsoft and OpenAI announcements of the past two weeks have competed to outdo one another with features and applications of AI that are mind-boggling, but not here yet. Or here but they don’t always work. For an early-adopter like me, I’l tolerate the bumpy experience but for many users it is chronically off-putting. The price of the candle isn’t worth the flame.
The much-quoted Harvard Business School study that showed productivity gains for knowledge workers using ChatGPT-4 last year was named after an insight that came out of the research: generative AI can massively boost productivity and the quality of people’s work, but not in all their tasks, but not evenly.
Testing the new ChatGPT 4o for the New York Times, Brian X Chen said it excels at editorial but is not good enough to be trusted with coaching his child on maths problems. He says:
Free is the right price: Since we are helping to train these A.I. systems with our data to improve, we shouldn’t be paying for them.
That may be true for consumers, but for businesses wanting some security and ability to share techniques in teams, the paid option is still the best.
Chen says that AI companies are releasing half-baked products because they want people to use them so they can improve them. This is true, but there is also a reckless (Open AI) or panicked (air) to the competitive releases.
Using generative AI can be like being an early motorist or aviator – thrilling, but its unpredictability and complexity is to be expected, respected and worked with in a way. It is engineering mindset, an innovator’s mindset, finding what works by doing it. Not everyone is comfortable with that and even fewer of them are comfortable with the idea of large workforces all getting their own cognitive whirligigs.
AI in its early days is moving fast but in complex ways. The Big Tech companies would be better off as would we all if they took a little more time over product development. But they won’t. Trillions of dollars are at stake and the logic of the game won’t let them pause.
Some will make the judgement that all of this is more trouble than it is worth to keep up with. Wait for the final, user-friendly package to arrive as it surely will in a few years time.
Failing at failing
Last week I said I would try Dr Veldsman’s method of applying for things I probably wouldn’t get. Making failures a goal, and noticing what the side-effects of over-reaching would be,
I sort of did. Sort of didn’t. It will take some more practice.
Part of the reason I couldn’t fail as much as I wanted to this week was because my own thinking system had a crisis. I thought I was a bit tired or burnt out, but it’s the cognitive equivalent of a migraine. A storm sweeping through the neural pathways like its rainy season. Hanging around.
This week I recommend…
Watching…
Eric (Netflix). Strong start to a limited series. Benedict Cumberbatch plays the creator of a Sesame Street-like show in early 80s New York. His son goes missing on his way to school.
The Sympathiser (Now TV/HBO). A half-American half-Vietnamese double agent caught in the fall of Saigon. Worth it for Robert Downey Junior’s make up alone, this seems like another HBO bit of brilliance. First episode has me hooked.
The Muppets (Disney+). Everyone needs a happy film for when their brain stops working. This is mine. When I saw it at the cinema with my kids I laughed so much the people in the row in front of me wrote a letter to the Kermode and Mayo show about it. True story. Written by one of the Flight Of The Conchords duo.
Reading
Yellowface by R F Kuang. A nervy culture wars literary heist thriller. Uncomfortable but once past the slow start, gripping.
Birnam Wood, by Eleanor Catton. Woke eco-warriors and tech billionaires making plans for the end of the world in New Zealand. Inspired by the same source material and themes as The Future by Naomi Alderman, and so far just as much fun.
Making
BN Edition. The Brilliant Noise weekly newsletter picks three AI-related stories every week and gives a concise explanation of what they mean. I was disproportionately pleased with the 80s pizza advert image we created in Midjourney, in reference to that Google’s problems with its search overview.
Gov AI report bot. The newsletter also covers the Tony Blair Foundation’s recent paper on governing and AI. We made a GPT to help you riffle through the 70+ pages to find what’s useful. Top tip, ask it to create a table of use cases including the challenge, approach and outcomes. There’s 25 of them in the text.
AI + Marketing - CIM Newsletter. This is a piece based on my webinar for the Chartered Institute of Marketing a couple of months ago about how generative AI would affect marketing.
That’s all for this week!
Thanks for reading. I hope there was something interesting for you here.
Antony
🔗 That Frederik Backman speech in full: