Nowism - Edition 35

Blimey, it's nuts out there.

You didn't miss my last newsletter, I only send when it's good enough and worth your time. Sometimes that takes two months."

The Big One.

I’m in two minds.

I always thought that fully self driving cars would be impossible. How would a car ever know the difference between a paper bag and a rock in the road, or get a sense of when a kid would chase a ball into the road.

It turns out with enough data and enough compute, rather like a person, you can (mostly?) solve that problem.

Some people in the world think this sort of applies to everything. With enough data and compute you can have humanoid robots that do any job, from laundry to mining to teaching to nursing, everything comes down to more data and more processing power.

They think with AI, you can invent new materials, develop new drugs, you can predict the ROI of a tagline, or pretest a new Kombucha variants success or answer the big questions like what’s best, Finish Quantum Powerball , or Ultimate Powerball, or Power Powerball.

This week, more than any other, which we didn’t think possible , all manner of craziness is afoot, Openclaw, 5.3 Codex, Claude Cowork, it’s relentless waves of “breakthoughts” and ever more dramatic claims. Are these things becoming emergent or are they just doing precisely what we precisely told them to do.

I’m in the godforsaken middle AGAIN.

I think 97% of people are not thinking about AI enough, or they underestimate it. They are a somewhat vulnerable, ignorant, unprepared for some true changes that will almost inevitably come. From being scammed by clones to sharing fake news, it will impact everyone.

And we’ve a painfully loud 3% of the world thinking about it far far too much. People mesmerized by promises, who underestimate the truly VAST real world challenges of turning even the most magical tech into something that can be implemented or add value, in the slightest of ways.

It’s all taking on the characteristics of religion.

Evangelicals saying the end of the world is nigh, exasperation over the non believers, the heretics who ask questions, it would be easier to take seriously if they weren’t the same people who barked like seals at the Metaverse, Web3 and how FIAT was going to fall to NFT’s. There’s a lot of Gel Mann Amnesia too, people who know least about real world jobs, with no clue that more people use fax machines that Slack. Who continue to think truck drivers jobs or Architects are at risk because they have no idea what they actually do.

If we look at real world implementation there’s all manner of shocking stats.

Everyone loathes MS Copilot, OpenAI seems to use remarkably little of their own AI, 95% of Gen AI pilots fail, 85% of AI projects show no significant impact, even the people who boasted most about integrating Vibe coding went a bit quiet after the IPO.

So it seems like OpenAI is struggling, Anthropic will say anything to get noticed, Microsoft have to talk up their own book, no wonder they keep promising the next release will change it all or that “we’re making something so good, we’ve had to make it less powerful”, it feels like a con job.

Right now we’re at the inevitable loggerheads of the algorithmic age, where only extreme opinions get noticed. The sides are talking past each other. Again “Have fun staying poor’ from the NFT era feels like the vibe.

For what it’s worth I remain very much bemused by both sides.

I think

• Something can be incredibly magical but not particularly valuable.

• Something can keep getting better, but a tiny error rate may make it worthless.\

• I don’t think the world is particularly rational in ANY way.

• Tech people are staggeringly odd and their thoughts and behaviors have no impact on 97% of the world.

• Companies are motivated by risk avoidance over efficiency

• The depth of integration which creates change, not the power of the technology.

• Something can change the world, but take a VERY long time to, like EVERYTHING EVER before.

At the heart of this are many interesting questions, but 2 big ones for now.

1) To what extent can every problem be solved by enough compute & data.

A lot of the promises of AI don’t sit right with me.

Maybe I’m a hippy but I think a lot of the world is more unknowable than we think, most data isn’t accessible to AI, most things are chaotic.

I just don’t believe, for the foreseeable future, AI will be able to do anything like as many complex things as we think. A lot of limits to the world are the wrapping around it, privacy concerns, regulations, business models.

2) How fast and how does the world really change ?

Technology changes the world based mainly how easily it fits into the world.

In order to fit in, it tends to

1. Find small little cracks to dominate “voice to text”, make an image, write an email

2. Do something thats been done before but cheaper / faster

These two things make it quite unremarkable.

The business case for AI adoption is not particularly strong with tasks.

Well done, the office barbecue flyer went from Clipart to Photo realistic and now it’s a Michael Bay movie.

Change to the world involves taking on Integration issues, existing contracts, secure, technical debt, meetings, fear of losing power, explainability, audibility.

The vast majority of vast software, think of SAP, Workday ,VMWare, Salesforce, Jira, Teams, PeopleSoft, WebEx, Citrix, Sharepoint is absolutely fucking awful. And it doesn’t change because they get flown to F1, and a software license audit will be called the moment they try to take it out.

We are more likely to 3D print our own furniture than develop our SAAS

We’ve spent billions and decades keeping bots off the internet and out of our sites, but now we’re going to let agents with read write permission roam free, get out and stroke a dog.

I’ve got this weird idea that the big division in the world is no longer rich and poor.

It’s people who live online and people who live in the real world.

It’s really, really easy to think that AI will change the world fast, until you place one foot in reality.

AI is a bundle of very different tools that are all quite profound in nature.
In the same way that the internet changed things slowly, AI is a mere inflection point in this journey.

The progress will happen (1) Cheaper/Faster, (2) Better, (3) Different

And the leapfrogs will happen when people rethink workflows and systems rather than applying AI to what they do now.

AI will ultimately be 50% culture change, 30% company design , 20% new workflows, and 10% new tools

There will be many missteps, initially AI will absolutely turbo charge bureaucracy, not eliminate it.

In fact, I feel like the prevailing emotion in the next five years will be enormous tension.

The tension between what we know AI makes possible & what we routinely experience.

We will be told that image recognition can predict crime before it happens, and detect Cancer at scale, while the bag scanners at airports are slower than ever.

We will be told AI and Big data will offer personalized ads, but my Credit card will decline a $10 charge in Germany, a week after buying a $10,000 flight TO Germany

We will lose another Boeing 777 , we will get an MRI scan on a DVD, we will need to do a Wet signature after a phone has scanned 30,000 data points on my face, picking up a rental car will be as awful as ever, all while we’re also experiencing the most magical moments we can ever imagine.

Little Ones.

1. Turbocharging the wrong things?

This piece in HBR is fantastic and very timely

It shows that AI doesn’t so much reduce work as make it more chaotic and intense.

One of the really weird things about the modern era is that productivity still seems related to effort and outputs more than outcomes. The big fear from AI for me, is that we do vast amounts of things that don’t need to happen.

Because focusing on what matters takes balls.

2. What could go wrong with a bot controlling everything?

You’d be an idiot to try Claude bot unless you know precisely what you’re doing. I don’t, so I relied on pieces like this, to show me what I’m missing

3. Personal AI
   
   This is Googles first foray, but I’m sure the next phase of AI will be about trading privacy You share more, you get more help in return. Then someone uses this for ads that you find quite helpful.

4. Book burning at scale.

It’s always a bit weird to me that if you copy one book, you get in trouble, if you scan a million books, burn them, then charge people to make copies of them, you’re fine

How anthropic threw away millions of books

5. Look mum, no hands.

Wonderful human and great friend of mine, Alex Roy did something amazing, well, the computers did.

I’m just putting an AI generated video of my dog in here, because I feel like it. It’s based on a photo, but the video somehow looks nothing like him.

Bites

• Wellness but on steroids, well, peptides Longevity and body hacking is changing

• Audi Naming mess - I don’t get how car companies make so many dumb big decisions.

• OpenAi is quite screwed - One day Sam Altman and Sam Bankman-Fried will have more in common than just a first name.

• For me AI is mostly recursive, It’s based on the (A)verage of the (i)nternet. Can AI create new? And can it do new math?

• The number of scams Meta happily take money from is incredible.

• What It Actually Takes to Build a Keynote” - superb piece from Zoe Scaman

• 3D printed wood furniture

• The paper passport is dying ( slowly)

• 250,000 drone deliveries from a company that should be celebrated a lot more

• Probably the best, and therefore most depressing, fake music.

• Financial nihilism will probably be a big thing this year.

• Learn from the Startup graveyard

• Flying taxis are coming to Florida.

Blast from the past

Some trends I wrote 6 years ago. The future is not that unknowable but the timing is impossible.

Me Me Me

This year is a bit quiet, it always starts off that way.

I’m on the edge of lots of proposals being accepted or declined.

I’ve 15 events that may say yes or no.

So as of now I’m open to:

• Keynote speaking or more intimate powerpoint based affairs.

• Workshops / Training on all elements of AI

• Consulting where I actually solve problems you have or show you new possibilities

• Consultant Agitator- Got McKinsey’s B-team in? Your boss thought Bain could help, but it’s basic? Average thinking from Accenture. Bring me in and spend 5% more to get 50% more out.
  Make them work harder with me as an ally and therapist.