Nowism - Edition 31

So where is AI in the hype cycle, nobody knows, but some questions arise. Arlo says hello.

It all feels a bit overwhelming these days. My attempts to keep up to date with the daily developments in AI have been exhausting, which explains the irregular nature of this newsletter.

“Nobody knows anything" was coined by William Goldman, a screenwriter and novelist, as a way to explain how Hollywood “worked”, and it always seems like a perfect description of the current AI world, the more you read, the more you experiment, listen, talk, learn, the more it seems to make less and more sense.

I’ve learned to take a step back and try to see things in context, over a longer period, then some patterns but also questions emerge.

Big Ones

Where are we on the AI hype curve?

Is it the law of diminishing returns or exponential growth, Zeno’s paradox or the runaway effect? Where are we?

It’s easy to get swept up in the wave that is AI. I returned to Facebook to buy rare plants from locals on Marketplace, but my entire feed was flooded with ads for AI courses from prestigious institutions or tips on how to make money as an “AI transformation specialist.” It felt rather scammy—very web3 or metaverse. My Twitter feed is an endless stream of proclamations: vibe coding means anyone can do anything, universities are dead, or AI-first startups will eat the world, despite employing just two people—one of whom might be an enthusiastic 8-year-old.

There’s plenty of evidence to the contrary, too. Klarna rode the PR wave of replacing humans before almost entirely walking it back. Models that once seemed to improve dramatically are no longer doing so—or is this a classic case of our expectations of technology advancing faster than the technology itself? What’s magical becomes ordinary, then disappointing, in about 43 hours.

Three questions in my head:

1. Not so fast, the world is messy.
   There’s a certain Gell-Mann Amnesia effect with AI: the less you know about a field, the more impressive AI seems to be at it. It’s easy to think it can replace a lawyer, architect, or copywriter if you have no idea what these jobs actually involve. The only exception seems to be coding, where many software engineers are the most impressed by what “it” can do. Regardless of its ability to handle specific tasks, we often forget that most jobs aren’t just tasks—they’re far messier. The role of a receptionist, for instance, probably involves 24 different things, of which an iPad can replace only one.

2. We know how the movie ends……or do we?
   If there were one chart I could make go viral, it’s this: technology has always progressed in three stages—(1) cheaper and/or faster, (2) better, and then (3) different.

   
   First, it’s used to make the same things, but cheaper: the plastic button, the steam engine, the computer—all were initially used to do the same things, just more cheaply.
   
   Then, it’s used to do the same things in a slightly different but better way: the electric lightbulb, the electric iron, synthetic dyes, steel.
   
   Finally, it’s used to create something radically different: film, microwaves, lasers, the TV set.

So, we kind of know how AI will go:

• 2025: Automate customer service

• 2027: Improve customer service

• 2030: Reimagine customer service
  

• 2025: Automate ads

• 2027: Create much better forms of advertising

• 2030: Redefine what ads should be

  
  etc

  

  3. There are really three types of companies when it comes to AI adoption
     

  Simply sprinkling AI onto a legacy process won’t work as well as redesigning that process from the ground up with AI at its core.
  
  The differences between these three types of companies are stark:

  • A large, traditional multinational trying to adopt AI.

  • A large, digital-first firm aiming to integrate AI.

  • An AI-first startup built to leverage it from day one.
    

  AI can be deeply transformative, but it’s incredibly hard to get big, established organizations to change. Even famously nimble, entrepreneurial companies in the second category will struggle with challenges like security, data governance, and new workflows.

  So, yes, AI-first startups (type 3) will likely harness AI’s power first. But then the question becomes: you’ve mastered AI—do you know how to make money in your sector? Will an AI-first retailer, bank, insurance company, energy firm, or marketplace automatically dominate? Probably not. Most winning companies succeed by understanding their sector deeply, not just by mastering the tech.

Non -AI things.

Make Advertising Great Again.

I got a column in the drum, in my second post I swore about how shit ads are now and how we can Make Advertising Great Again.

”To be advertised at today is to be insulted, the promise of adtech is a lie, we don’t offer one on one conversations at scale, we offer random precise guesses as to who you are, that result in it thinking you’re an old, pregnant male suffering from poor health. It will engage you in conversational Cantonese to stoke your passions for luxury handbags and industrial rivets.

Take a few seconds to unblock those ads, and then take a look and see what you get. Did a human see this? Did anyone care? Did anyone expect me, a human, to click?

I don’t think we realize how deeply nasty it all is; between the lies, the fraud, the scams, the bots, the targeting, the following, the harassment, the automated ways to spank budgets on fake clicks, faster than ever. It all makes pig butchering or 419 scams look rather sophisticated and above board” - Read more here.

Sponsored post ( but still good) - Spinnaker technology

Nokia. Kodak. Blockbuster. Borders.

Remember the relentless tech disruption narrative from 2010?

These companies were cautionary tales, and the question on everyone’s mind was who’s next?

Surely Walmart & Target were doomed, Airbnb would crush Marriott, & Toyota would fall to Tesla, Uber, Waymo, or all three.

Who needed a Gym in the age of Peloton, or Bank when we've got an neoBank.

But then… nothing much happened.

For all the talk of “Disrupt or Die,” “Digital Darwinism,” and “Creative Destruction,” reality turned out far less dramatic.

Carry on reading the post here.
https://www.linkedin.com/posts/tomfgoodwin_nokia-kodak-blockbuster-borders-remember-activity-7287526658012565504-D1fU

Little ones

• An interesting piece on various companies completing changing how things are made.

• Love this piece on Cognitive debt. It’s when you skip deep thinking to get quick answers, but you don’t understand why those answers are correct. It’s like borrowing knowledge without paying the “interest” of truly grasping it

• Robots flopped at a half marathon, proving they need wheels, not legs. Human-like robots are pointless unless deception’s the goal. Why do people keep forgetting this?

• Quite a useful case on how a Legacy Financial Institution Went All In on Gen AI. HBR is often nothing like as good as it used to be, but this is decent.

• A lot of the time it seems like the jobs the US wants to bring back don’t really exist any more and that’s prob a good thing, coal mining anyone, but sometimes it’s not that simple. Turns out robots making shoes isn’t that easy.

• I love how you can art from the weather in real time with this , just wish people made Digital Art Screens for this stuff.

• I’m always fascinated by bulky, large, outdated infrastructure. Sometimes I do think that deep transformation is essential, but there's never a business case to be made for it. Things like Covid showed sometimes urgency, desperation and disruption is a good combo. Doge’s work is quite interesting to show the chaos that ensues. Sometimes you have to Take steps back to go forward

• People seem to think startups like this will threaten management consulting, I think it’s nonsense, people don’t choose McKinsey for ideas or work done, they choose it for career insurance and to show you tried , and to get existing work sold in by outsiders.

Thought for now.

What are we saving all this time to do?
The more technology I have in my life, the most determined I am to be bored.
I’m cooking more than ever, doing more things slowly, I’m getting surprisingly obsessed with house plants and watching them grow. I also have a puppy and watching him attempt attack a tennis ball gives me a rush that no notification can currently compete with. So much of live isn’t to be automated or streamlined, it’s to be entranced by. Miami is about to enter the season of the very best clouds, it’s exiting.

All about me.

I’m speaking at AntiCon 2025 in London this week about AI and Advertising.
Then some private events in Paris before another talk in London for Mailchimp on June 12th.

Currently planning a bit of a global tour to Australia, maybe Cape Town and somewhere else on the way in August 2025, and prob the Middle East in November, if you want to add me en route.

This is my puppy. He owns me now. He’s called Arlo. He’s tiring but magnificent. Everyone should have a dog.