Antonym: The Twankydillo Edition
Bring your own brain.
Dear Reader
As you can see I survived the neo-pagan Apple Day last Sunday. Lolling in the shade of the orchard with a litre of scrumpy and a bowl of Apple Crumble listening to a choir sing Twankydillo, I resubscribed to the idea of Apple Day as one of my therapy-mandated “happy places”.
We begin with a quote that links this newsletters twins obsession as Succession showrunner Jesse Armstrong gives his thoughts on AI on the Adam Buxton podcast:
I’ve managed to put it in that box of 'oh-I-don’t-fucking-care' — Jesse Armstrong
OK. Whereas I, you may be aware, have put AI in the box of “what-I-think-about-all-the-time”.
Speaking of which, I’m running a free webinar on Thursday (19th October) about AI trends and “how to plan for the unplannable” (i.e. what will happen with AI in 2024). Come along or message me and I’ll send a link to the video afterwards.
How to automate everything: Don’t.
Most organisations are at a naive stage of automation of complex knowledge work.
We have got thinking machines - generative AI – and some, especially non-specialists, think they can do everything. For now, they are wrong. Others think they can do none of the hitherto human-only work and they are definitely wrong. To find out we’re going to need to get into how we make things in more detail.
One place we can learn from is complex automotive manufacturing, like when Tesla scaled its manufacturing for the first time.
The Elon Musk biography by Walter Isaacson has a lot on Musk's management principles for Tesla, which Musk says are:
1. Question every requirement. Each should come with the name of the person who made it. You should never accept that a requirement came from a department, such as from “the legal department” or “the safety department.” You need to know the name of the real person who made that requirement. Then you should question it, no matter how smart that person is. Requirements from smart people are the most dangerous, because people are less likely to question them. Always do so, even if the requirement came from me. Then make the requirements less dumb.
2. Delete any part or process you can. You may have to add them back later. In fact, if you do not end up adding back at least 10% of them, then you didn’t delete enough.
3. Simplify and optimise. This should come after step two. A common mistake is to simplify and optimise a part or a process that should not exist.
4. Accelerate cycle time. Every process can be speeded up. But only do this after you have followed the first three steps. In the Tesla factory, I mistakenly spent a lot of time accelerating processes that I later realised should have been deleted.
5. Automate. That comes last. The big mistake in Nevada and at Fremont was that I began by trying to automate every step. We should have waited until all the requirements had been questioned, parts and processes deleted, and the bugs were shaken out.
Principles are useless unless they are put to work. Musk called his list “the algorithm” and repeated it so often in meetings that his teams would mouth the words as he said them, and he also made sure they were being implemented.
When trying to scale production at Tesla to 5,000 Model 3 cars a week, Tesla started by making a Muskian fantasy of an automation-first factory real. Everything that could be done by robots was done by robots.
Tesla did things backwards:
Ever since the development of assembly lines in the early 1900s, most factories have been designed in two steps. First, the line is set up with workers doing specific tasks at each station. Then, when the kinks are worked out, robots and other machines are gradually introduced to take over some of the work. Musk did the reverse. In his vision for a modern “alien dreadnought” factory, he began by automating every task possible. “We had this enormously automated production line that used tons of robots,” says Straubel. “There was one problem. It didn’t work.”
De-automation was a strategy that worked. They had to justify every robot on the production line instead of blindly accepting that machines are better. Things rapidly improved.
Musk flipped from being an apostle of automation to a new mission he pursued with similar zeal: find any part of the line where there was a holdup and see if de-automation would make it go faster. “We began sawing robots out of the production line and throwing them into the parking lot,” Straubel says. On one weekend, they marched through the factory painting marks on machinery to be jettisoned.
PR advice: sometimes, just stop saying things
Elon Musk’s corporate communications and advisors often just want him to stop talking / tweeting. If Walter Renwick had advisors they would likely give him the same advice based on reporting on him in The Times. He has been accused of chopping down a lovely tree next to Hadrian’s Wall.
Renwick, tall and broad-shouldered, has taken to wearing a blond Rod Stewart-style wig to avoid recognition by people furious about the tree. “People walked past me and gave me these looks, like I’m a piece of shit,” he said. Later he looks a little sheepish as walkers ask for directions. “Car park, up the hill, to the right.”
There’s no picture available of the full wig, but Google Images offers these suggestions.
BYOB: bring Your own brain to work
Generative AI tools are very easy to access, but they come with no manual and you have to play a game without rules to figure out how they work. You have to play until the work happens.
There’s a divide between people who are using AI tools in their work and those who can’t access or can’t be bothered to use them. I don’t think either group realises what the other is experiencing. The separation is not just one of attitude. Once convinced of the potential, £20 a month – the price of ChatGPT+ and most other premium AI services – seems like a bargain. Once you’re paying, it’s easy to forget that you’re using a version of the technology that is many orders of magnitude better than the free one. You don’t just get a few extra features, you get to access the future months or years before everyone else.
The unevenly distributed future is made more lumpy by AI blanket-ban policies. This week I spoke with someone who told me it was an age thing at their company: the Gen X/Millennial crowd frowned and banned it, the younger ones carried on using ChatGPT on their phones to get their work done faster. Just as a decade ago corporates had to buckle to the demands of workers when they brought personal smartphones, iPads and Macbooks to work. It was called Bring Your Own Device (BYOD) policy, and meant that people bringing devices that were nicer to work on to work was tolerated. We’ll see something similar with AI assistants, I think. Bring your own second brain to work, or something like that.
Try Adobe Firefly 2
Adobe Firefly 2 is almost as good at creating images as MidJourney and, crucially for most people, a lot easier to use. The app which is free to use at the moment for subscribers to Adobe’s pricey software, has lots of simple settings to get usable images and it is more mindful of copyright risks than competitors (more aware but not risk free from a legal or ethics perspective, depending on your moral and legal frameworks).
Here’s some examples:
And that’s all we have time for this week folks…
If you liked it give us a share!
Thank’ee
Antony