Structured human expertise is fueling the AI race

I joined micro1 because helping the US win the AI race is the most important challenge of our time.‍

AI has already proven to be a technology of unprecedented capability. Those who develop the best AI will have unprecedented impact, control and influence on society at large. Speed is of the essence, because there will come a point when AI superiority turns to insurmountable AI dominance. The leapfrogging may stop, and a winner definitively crowned.

The potential for good is very promising. The human condition could be massively improved. Never has a technology so powerful been available to so many people for so little cost.

But there is also a significant defense angle. Depending on who wins the AI race, our society may face unwelcome changes outside of our control. The race for AI is a national security priority. It’s important to act with urgency to help the US win the AI race.

There are three parts to the AI race

The battle for compute is well documented.

It involves infrastructure, policy decisions and a relatively small group of highly skilled humans. Most humans who can partake in this battle are already in it.

So is the battle for top researchers.

This involves more humans, but it is still a small group. It’s growing as AI becomes more accessible. Most humans who can partake in this battle are also already in it.

But there is a third battle underway: the battle for human intelligence.

‍There are almost no headlines about this battle. I believe it’s the most important one, and it will involve by far the most amount of humans. In fact, just about every human can play a role. Yet only a very small % of them are. That will change with micro1.

The battle for expert human intelligence

AI labs and other organizations rely on expert human intelligence to improve their frontier models and context-specific agents. Thousands of humans outside of the field of AI are now directly responsible for some of the most impressive recent AI advances.

This data is not available on the internet. It can’t be scraped. It exists only in the minds of humans. It comes in the form of direct, live and raw interactions with frontier AI.

Lawyers, doctors, finance professionals, engineers and many other humans from a wide range of skillsets are intently training AI on a case-by-case basis in a highly structured manner. They provide granular feedback, highly complex grading and detailed evaluations. It can take hours to create a single data point.

It is not data-labelling. This work is time-intensive, tedious, and arguably more impactful than if those hours were spent working for an employer creating unstructured data. No other work allows you to impact humans on this scale. Any improvement to a major AI model impacts a user base measured in the hundreds of millions. For example, a doctor could advise hundreds of patients in a day. Or she could train a model to advise hundreds of millions of patients. Market dynamics support this: hourly wages for producing this kind of data are significantly higher across all domains.

This is happening at scale and in real time. Expert human intelligence is fueling the AI race.

A huge responsibility

Platforms, like micro1, who sit at the intersection of  expert human data and AI models play a major role in deciding where this fuel goes. We choose which companies have access to this expert human data.

This decision cannot be made with revenue as the sole or even primary consideration. AI companies have an insatiable appetite for expert human data. And it’s no secret that they have the wallet to go with it. It’s our responsibility to carefully balance our business motives with the impact that serving a potential customer could have. This is an obligation to our experts and our society.

We have made the decision to only serve customers who are in the US camp of the AI race. This may not maximize immediate revenue, but it contributes to future prosperity. We believe it’s a good long term financial decision.

How we can help

Our platform is enabling the best human minds to contribute to this race.

The more brilliant minds participate, the more impactful our role becomes. Everything we do is based on ensuring that our experts are happy. This leads to high retention and attracts more amazing humans to participate.

And expert happiness is a direct function of meaningful work. The economic incentive is only one part of this. We’ve developed a unique bond with our experts, who trust that we put their intelligence to good use.

We plan to onboard many millions of brilliant humans who want to participate. They’ll play a direct role in helping innovative, hard-charging companies build the technology of a lifetime.

It’s a challenge without precedent.