A primer: LLMs work by predicting what comes next in a sequence. Trained on billions of data points, this method has proven to be more effective than our wildest dreams.

And in code? Say you’re the 10,000th person today that wants to add email authentication to your app. Predicting the next line of code is often pretty simple. Coding is full of repeatable patterns and predictable completions; that's why AI is so good at generating working code.

But “likely” is not the same as “correct.” And “working” is not the same as safe, secure, or clinically sound.

So where do LLMs fail?

When results aren’t normal. When edge cases aren’t easily “predictable”.

For the average person in healthcare, understanding a generic diagnosis may lead to a useful average prediction.

But for marginalized communities? For people with rare diseases? Standard prediction can fail in non-standard cases.

If a case is under-represented in the training data—however large that data set may be—the correct answer may be less likely to show up in LLM results.

It’s one of the many ways we can be in awe of what LLMs accomplish today while still being well aware of exactly how they fall short.

As with all things tech, there’s a social justice component to this, too.

It’s common advice online and in business that we must find a niche—we must specialize in order to thrive.

When we look at our own bills to pay, we see AI as a potential immediate threat to our livelihood. What if we get fired and can’t find another job?

But when we look at species of the past that have either thrived or gone extinct, we can see the arc play out from a higher vantage point.

The essential takeaway is this: Specialist species (ones located in a specific biome or tied to a very specific resource) thrive when conditions are stable. They are more likely to do extremely well, but they are also more likely to go extinct when things go sideways.

Generalist species, on the other hand, may not maximize resources in the short term. But they are more likely to survive when conditions change.

We are undergoing a time of massive change.

The answer for you might be to generalize and adapt, as we watch specific old models crumble around us.

I’m not a theoretical builder. I’m a real business owner employing real humans in a time of massive uncertainty and disruption.

And since I was 13, I’ve pushed the boundaries of every piece of technology I’ve owned, starting with programming assembly language on my TI-86 calculator.

So when I talk about AI, it’s not theoretical. It’s not from the standpoint of a newscaster who can barely open Gmail on their phone.

It’s from the point of view of a builder who is constantly navigating implementing technology ethically while maintaining the livelihoods of a team that I love like family.

I talk about the intersection of ethics, philosophy, technology, and real-world leadership.

Because we need fewer opinions from those on the sidelines of these issues. And more from those who are actually building with it.

The bets I make with AI are the areas in which I think it has the greatest probability of improving my business while keeping my team employed.

The bets I don’t make are reflected in what I don’t do.

Hint: Not one of these posts is written by AI.

It’s never been easier to spin up an AI-generated, pseudo-scientific 25-page document and tell your team to memorize it.

But asking someone to study (and spend valuable time digesting) something that we spent very little time creating is the ultimate form of modern disrespect.

When we do that, we’re effectively saying “I don’t value your time, only my own.”

This is guaranteed to rub people the wrong way.

Steve Jobs was famous for extremely short emails. “Thx” “K”. Because previously, busy CEOs had to use extreme brevity to maximize their time.

And it’s not just me who feels this way. Even Paul Graham, co-founder of Y Combinator, admits that he checks out whenever he sees something was written by AI.

Just because ChatGPT can write a ten-page response to every email doesn’t mean that it’s more effective than “Thx.”

The history of tech products is littered with failed product demos.

From Blue Screens of Death in front of thousands at a Microsoft event to the Cybertruck’s shatterproof glass easily shattering on stage, to robots falling down mid-routine and going berserk…

There’s no shortage of embarrassing failures out there.

But a lot of it comes down to expectations.

Deep tech is hard. Doing things that have never been done before is hard.

Creating a rocket that’s never existed before is inherently more likely to cause catastrophic damage than a new sneaker.

But we shouldn’t shy away from (or hide) this type of failure. Instead, we should embrace it.

It’s all part of the process. The failures represent a path towards success, and big victories often come shortly after (and because of) very public failures.