The AI squeeze is real. While AI may or may not lead to widespread job losses, I think people are underestimating the number of basic job it will kill, jobs that were done by entry level e

While AI agents may or may not led to widespread job losses, I’m

AI agents will continue to get smarter and more powerful. In my company, this means it no longer makes sense to hire interns or new grads as software engineers. It’s easier and cheaper and faster for us to manage AI agents. That’s pretty crazy! At the same time, the value of super-senior software engineers is higher than ever. Big tech pay top talent millions and millions in salary to programmers. We’re still hiring, but only the best of the best, like everyone else, so there are very few of these super-smart-folks to go around. The market for intelligence (human and machine) is going to get wonky!

I think this sort of thing will happen in every industry over the next 2 years. Everyone will realize that they can delegate more and more of their work to AIs, and become managers of whatever job they used to do.

All of this is just three and a half years since the launch of ChatGPT:

While leadership sleeps four hours a night generating 37,000 lines of bloated code, the New York Times coined a term for what’s happening downstream: “++tokenmaxxing++.” It’s a competitive status game where employees race to consume the most AI tokens. OpenAI has an engineer who processed 210 billion tokens in a single week. Anthropic has a single Claude Code user running a $150,000 monthly bill. Shopify’s Tobi Lutke made AI usage a ++factor in performance reviews++ (Meta did the same). Some companies have literal internal leaderboards tracking who burns the most tokens.

The leaderboard measures consumption, not output.

The age of weaponized interdependence:

Please use the sharing tools found via the share button at the top or side of articles. Copying articles to
The post-cold war assumption that dependencies were broadly shared and the open economy was a global public good has quietly come apart. We are living through an age of asymmetry, a transitional period in which power flows less from size or wealth than from the ability to convert imbalance into leverage. Strategic advantage now accrues to actors that recognise the imbalance and act on it before a new equilibrium settles.

While Tehran and TSMC compound advantage, others are absorbing costs they haven’t yet learned to price, from the EU’s limited sovereignty over its own consumer data to Volkswagen, whose EV transition rests on Chinese components.

This is from Colgate-Palmolive’s recent earnings call. If the Strait of Hormuz crisis continues, we’ll start seeing serious inflationary pressures. Companies will take a hit to their margins, but beyond a point, they’ll start passing on costs:

Let me just address it from a macro standpoint. I’ll let Stan provide some more details. You know, clearly the assumptions that we have embedded into our guidance for the year include the $300 million of additional raw materials. We’re assuming oil roughly at around $110. I think importantly, strategically, as we’ve always gotten ahead of the cost environment, we need to ensure that our operating units are planning for these types of inflationary environments that are coming. Clearly, we’ll wait and see. There’s a lot of ups and downs moving around the world, so to speak, on oil pricing. 

10m 31s

For us, strategically, it’s important that the operating units start to build this into their strategies on how they want to execute some of the strong innovation plans we have for the balance of the year, how we execute funding the growth for the rest of the year. Again, we feel it’s very prudent to get those numbers out there, and we built that into our guidance. Clearly, some of the inflationary environments is forced us to take the gross margin down for the year. Overall, we still feel we’re well in line with our guidance on earnings per share. Stan? 

Stan Sutula, CFO, Colgate-Palmolive

11m 1s

Yes, let me pick that up. Our assumptions for the year embedded in our gross profit margin guidance includes oil at roughly $110 on average for the remainder of the year, and the associated impact that has on raw and packaging materials. Since the fourth quarter call, you know, we’ve seen an additional raw materials and logistics impact for the year of roughly $300 million. You should think of that as roughly 2/3 raw materials and 1/3 logistics. The biggest incremental impact, Filippo, is coming from oil byproducts, resins, petrochemicals, fats, and oils. We now expect that spending in those areas to be up more than 20% year-on-year for the full year. You can see the impact that that has. Our logistics costs are up nearly 10%, impacting both ocean and land freight.

One useful tool

Scrapling for scraping sites.

A few nice charts

This is amazing!

image.png

We’ll long be discussing the impact of this.

image.png

Arindrajit Dube on his experiences using Claude and CODEX in economics research:

AI and economics research To date, I’ve seen a lot more on how AI can help revolutionize economics research than I’ve seen how concretely AI made new research happen in economics. In my own experience, I’ve had Claude Code markedly increase my productivity by cutting down on coding time. But it has failed to give me a single new economic insight. (I’ve tried.)

—

In contrast, Claude Code or Codex can really cut the time to code painfully routine coding by 80%. Non-routine stuff maybe by still 30%. That’s a real gain. It can’t produce deep insights, but it can really help. I wish we focused on this type of productivity gains instead of the breathless singularity discourse that just permeates this platform when it comes to AI.

Can LLMs identify new tax loopholes in the same way that they identify software flaws?

Even more interesting are the broader implications. The same searching, pattern-matching and reasoning capabilities that make these models so good at analyzing software almost certainly apply to similar systems. The tax code isn’t computer code, but it’s a series of algorithms with inputs and outputs. It has vulnerabilities; we call them tax loopholes. It has exploits; we call them tax avoidance strategies. And it has black hat hackers: attorneys and accountants.

Just as these models are finding hundreds of vulnerabilities in complex software systems, we should expect them to be equally effective at finding many new and undiscovered tax loopholes. I am confident that the major investment banks are working on this right now, in secret. They’ve fed AI the tax code of the US, or the UK, or maybe every industrialized country, and tasked the system with looking for money-saving strategies. How many tax loopholes will those AIs find? Ten? One hundred? One thousand? The Double Dutch Irish Sandwich is a tax loophole that involves multiple different tax jurisdictions. Can AIs find loopholes even more complex? We have no idea.

I’m bullish on this.

A lot of the debate on AI and science focuses on whether LLMs can generate new ideas. Yet 20–80% of existing papers are never cited and presumably not read. By digesting vast uncited literatures, LLMs may meaningfully advance knowledge simply by absorbing old ideas.