Adopt, adapt, enjoy

📚 PDF ⏳ 7 min 📖 5

Chamath

I suppose it's amusing on a number of levels. Chamath has been super-cosy with the President along with half of Silicon Valley. Now the country has swung blue and it’s time to run for cover. “The wage-earners……”

All told though, he’s right. Asset price growth v wages is the number one issue and everyone, apart from Silicon Valley, has seen it. I have been singing it merrily from the rooftops here “wages are collapsing” because that is very much what we see here in Australia. Wages in the US are some 20% higher than Australia, and it’s still true there too.

“Broad based measures of success are no longer reliable”. GDP for starters is increasingly nonsense. As I observed a few weeks ago, AI is enabling all sorts of things that are GDP destructive, for example software subscriptions are mostly superfluous now because you can simply code your own solution. It is far easier, far better to make your own which is, paradoxically, negative for GDP and positive in every other way that matters.

Another would be GDP per capita, a nonsense derivative of a nonsense number. In an economy where 5% of the population are taking 95% of the AI gains, whether as owners or deployers, it makes no sense. The fact that it is the driving force of the economy currently means nothing because in most cases the net measurable benefit to most people today is negative.

I’d just make a point here about the only place in Australia where salaries are rising fast. The public sector. In Victoria, an experienced teacher is about to see their pay rise from $118,063 to $151,419 by 2029. Not that this is negative, only that it will flow entirely against the trend of the private sector and entirely against the value proposition of high school teachers. We have 162,000 people in Australia essentially preparing the same lessons, every year, the same curriculum and delivering education at one pace. AI can easily do this. The best students now get GPT to teach them what the teacher missed. In reality, a teacher has no chance to know that the 33rd student in their class didn’t understand, how could they?

Hundreds of thousands of math teachers around the world are shut away in their offices every day, painstakingly making identical PowerPoint slides over and over for the same middle school math class. This is the most absurd, industrial-scale waste of human intellectual resources imaginable.

Many people still treat classes taught by star teachers at prestigious schools as a scarce resource, fighting tooth and nail to get homes in good school districts and places in top classes, but Hassabis has shattered that illusion with the coldest first-principles reasoning.

Standardized knowledge delivery never deserved a place in expensive in-person classrooms. Rote memorization and grinding through practice problems should be kicked out of the classroom entirely and handed over to personal AI that can adjust in real time to each child’s millisecond-level response speed. Add a few of the world’s top masters recording the foundational lectures, and that alone is enough to completely replace the teaching component of basic education worldwide in one fell swoop.

For the past two hundred years, schools have essentially been industrial-age barracks, forcing wildly different carbon-based brains onto the same assembly line and making them tighten screws to the same metronome, teachers reduced to human tape recorders, students trapped in a sea of dull problems, competing to see who can be more like a hard drive, and once AI and top masters take over standardized knowledge delivery completely, the only legitimate reason for in-person classrooms to exist will be five dimensions that machines can never calculate through their weights: interpersonal strategy, hands-on work on complex projects, entrepreneurial ambition, creativity, and, most critically, judgment. There is an extremely cold, counterintuitive truth here: when personalized AI drives the marginal cost of high test scores through the floor, students trained to solve problems according to standard answers become computational dead weight the moment they walk out the school gates.

Put another way, the very best teachers should be paid millions, the rest zero. Naturally, we are moving in entirely the opposite direction, which I think is the path of universal basic income. Basically, just a load of people doing stuff that is totally unnecessary. Look around at work, honestly, what are they actually doing?

This is not to attack teachers. My own professional breed, accountants, probably have a worse (read better) future. I believe the profession is entirely redundant already. It was never that much fun, it was just well paid in some cases.

Good riddance to it I say.

Everything’s Pretend

Newest model this week Claude Fable 5.2. Watch this video. Berman is the best reviewer because if the model is no good he says so. In this particular case though, that is not what he says.

Protein design score 85.1%. Not impressed? Well, just worry about the more important score which is the Remote Labour Index (41.3%). It is a massive leap on everything before and is doubling every 3 months. June 2027, this is solved.

The point of the RLI is that they are real projects, what happens is the AI gets some half explanation. Can you do this project? Exactly as would happen in the real world, the instructions are weak, the supervision is weak, the reviewer is weak. You just have to work it out and some people can and some can’t. Essentially this is all white collar work.

You can see how politically sensitive this is because it is called the Remote Labour Index. Now, naturally, most of us feel we are not remote workers. I go to the office, important in person meetings etc. etc. Decisions, judgement etc. Remote Worker does not resonate with you or me. We think of a Filipino making $30k/year. But no, it’s you and me.

Everyone getting off the train on George Street, all those people in Australia Square buying lunch > IT’S YOU.

And yet, it's a big opportunity. So I guess we adopt, adapt, enjoy.

Ueno Bank

What is bitcoin for? When you live in Australia and life is a bit dull, you wouldn’t really know. Lots of us are in it because it might go up, in fact, if we're brutally honest. Why else? There is a view in traditional finance that this is in fact the definition of a ponzi. If that were all, I would agree.

Most of the world is not the first world though. Most of the world lives with political unrest, danger, unstable systems of government. Indeed, in many places politics can get you killed.

Paraguay, in my naive view, was not really one of those places. This week though, one of their neo-banks fell over. Ueno bank is gone and all the depositors’ money with it. Naturally, the president of Paraguay was once an investor (not anymore he says) the company was closely involved with the political structure, allegedly lots of dodgy loans, that kind of thing, and it's all gone south.

Incredibly, 60% of the adult population had an account there. The President has now been impeached and nobody has their money.

Bitcoin is $80,000 in Paraguay right now as people run for cover.

You have to ask yourself, in Australia, where the banks are the five of the top six companies in the country. Could it happen here? Surely they don’t make bad loans, do they?

There is a market for depoliticised, decentralised money. If the major participant in the market is also the one who prints the money, how can that ever work? It never has. It fails again and again and again. Ask Venezuelans, Zimbabweans, Nigerians, ask people in Paraguay.

Leadership

Another week, another win for Australia. The Prime Minister announced that Australia is a first mover in regulating AI, which means a rather European approach of fining American companies who might pay, and Chinese companies who definitely won’t. He isn’t the first though, the Europeans were years ahead. Remember this iconic tweet from 2023. The EU “AI Commissioner”.

We all know what happened next.

At least in Australia’s case we were never really in danger of having an AI industry. So maybe we can use the Google fines for something useful, like a new sports stadium or something.

I can’t help but think that this is largely where Australia’s productivity is vanishing. How many civil servants did it take to come up with these regulations for industries we don’t actually have in this country? I imagine it was vastly time consuming because regulating something you neither understand nor control is hard. The whole circus results in us never having the thing that we currently don’t have anyway, and the regulation is deemed a success. A great many people will share that view too, I guess I’m just not one of them.

Euro-Trash

Who is Pablo Hernandez de Cos? It doesn’t really matter, all that matters is that he is neither French nor German. The issue for the ECB (and the EU more generally) is that both France and Germany are likely to lurch to the right in their upcoming elections. What they don’t want is an ECB President who is susceptible to domestic pressure because the boys back home are on the phone. So the market is swinging behind Pablo, a 60% chance, albeit on thin volume.

In some ways it’s a good move, but ultimately it is going to alienate a lot of European voters, particularly the Germans. When Pablo uses lots of their hard-earned money to bail out the Southern States, which he will. Far, far better a German does it.

They have said they want this nailed down by December. If it is Pablo, there will be relief because he will work to make the Euro survive. He will do whatever it takes. It will be the wrong choice though. The Euro ends when the Germans end it, Pablo makes that more likely, not less.

Wikipedia watchers will appreciate some recent frenzied action on Pablo’s page. Seven edits in three days. Third place candidate Olli Rehn has five edits all year, one in the last six months. Just an observation, it means nothing.

Further information

Our October 2026 report to investors can be found here.