AI or Die

“You cannot overtake fifteen cars in sunny weather. But you can when it’s raining.” —Formula One driver Ayrton Senna

Did you read Dario’s Amodei’s (founder/CEO of Anthropic) post from late last year? This is what he suggested might be available from AI models, as early as 2026. The quote is long, but it’s worth reading in full.¹

“I have in mind an AI model—likely similar to today’s LLMs in form, though it might be based on a different architecture, might involve several interacting models, and might be trained differently—with the following properties:

In terms of pure intelligence, it is smarter than a Nobel Prize winner across most relevant fields – biology, programming, math, engineering, writing, etc. This means it can prove unsolved mathematical theorems, write extremely good novels, write difficult codebases from scratch, etc.

In addition to just being a ‘smart thing you talk to’, it has all the ‘interfaces’ available to a human working virtually, including text, audio, video, mouse and keyboard control, and internet access.

It can engage in any actions, communications, or remote operations enabled by this interface, including taking actions on the internet, taking or giving directions to humans, ordering materials, directing experiments, watching videos, making videos, and so on. It does all of these tasks with, again, a skill exceeding that of the most capable humans in the world.

It does not just passively answer questions; instead, it can be given tasks that take hours, days, or weeks to complete, and then goes off and does those tasks autonomously, in the way a smart employee would, asking for clarification as necessary…

The resources used to train the model can be repurposed to run millions of instances of it (this matches projected cluster sizes by ~2027), and the model can absorb information and generate actions at roughly 10x-100x human speed. It may however be limited by the response time of the physical world or of software it interacts with.

Each of these million copies can act independently on unrelated tasks, or if needed can all work together in the same way humans would collaborate, perhaps with different subpopulations fine-tuned to be especially good at particular tasks.

We could summarize this as a ‘country of geniuses in a datacenter’.”

A country of geniuses in a datacenter.

Is your company positioned to take advantage of access to a country of geniuses in a datacenter?

Is your company positioned to compete with the company that does take advantage of access to a country of geniuses in a datacenter?

The game on the field is about to change. The leaderboard in your industry is going to shift violently over the next few years.

If you have prepared your company for this, congratulations. You can stop reading.

Still here? Great.

What do you do now?

The first thing to do is to become much more familiar with what AI can do today. Most professionals think of AI as a cute chatbot that can replace wikipedia for them. They don’t realize that it’s already capable of being a virtual colleague and collaborator.

OpenAI recently released Deep Research. This is what Sam Altman posted about it:

A single-digit percentage of all economically valuable tasks in the world. In the world! Today! If Sam is right, the percentage of economically valuable tasks it can do in your company today is almost certainly higher. I don’t know if that means it’s 5% or 50%. (But I’d find out if I were you.)

So, if the models are capable of all of this, what actually stops people from getting the most out of them?

1. Ambition: Not enough people ask the model to help them with real work. They ask it to help them with drafting email or coming up with funny songs.

2. Patience: How much context are you willing to provide it to help you? How does it compare to the amount of context you would give a brilliant new hire who was joining your team? Is that the context you’re giving to the models when you ask it to help you? Have you ever asked a model if it understands what you’re asking it to do or if it has any clarifying questions?

3. Clarity of thought and communication: Spend some time sharpening the axe before you try to cut down the tree.²

I grant you that it’s not easy to make AI a virtual colleague today. But that will change. Entire companies of brilliant people are working on bringing you virtual co-workers and ushering in the future that Dario describes.

It will happen sooner than you think.

Let’s say you agree with me. The virtual colleague and collaborator is here. The country of geniuses in a datacenter is coming. What should you do about it?

1. Take a week off with the sole purpose of becoming more clear about what AI models can do now and where they are going.

2. Imagine the AI-native version of your product that could be built with a team that is one-tenth or one-hundredth the size of yours. Build it (or buy the one that might have already sprouted up.)

3. Understand that your customers are going to have dramatically higher expectations of you. They are going to expect a product that is dramatically better and dramatically cheaper than yours is today. If you don’t believe me, think about what you believe your customer support will be like in a couple of years. I bet you think it’s available all the time with someone friendly and knowledgeable, at a fraction of the price. This improvement distance you expect your customer support team to travel is the same distance your customers are going to expect you to travel.

4. Earn the right to be the AI strategy for your customers. This is much much much harder than it seems…but also much much much more lucrative than you might expect.

5. Become an AI-first company. There is a great story about Mark Zuckerberg refusing to look at desktop mocks when the world started shifting to mobile. What is your equivalent of that for AI? How can you show your team you are serious about transforming the team around an AI future?

6. Go person by person in your company and figure out who is going to be augmented by AI agents, and what roles are going to be replaced by them. For those that are going to be augmented by the AI agents, tell them that their status as “high-performer” depends on their ability to leverage the new technology.

These are not tweaks or subtle changes. This is not adding a slide in the board deck about how you are using AI.

Winning in this new world requires a full “re-founding” of your company.

This is neither simple nor safe. It is indescribably hard to re-found a company. In fact, in many ways, the bigger and more successful your company is today, the harder this will be.

But as my wife and I tell our sons…we don’t always get to choose what happens. But we do get to choose how we respond.

With that in mind, go back and read Dario’s quote again. And then read this from Sam Altman about AGI and what it means. Because I’ve already assigned you a fair amount of reading today, I’ll give you a quote that sums up Sam’s post pretty well in case you don’t read the whole thing:

“In a decade, perhaps everyone on earth will be capable of accomplishing more than the most impactful person can today.”

Wow.

It’s the most inspiring future that one could imagine.

It’s also the most challenging one to prepare for. Whatever industry you are in, the weather is changing. And when the rain comes, there’s going to be a lot of passing going on.

Where you end up in that race is up to you.