Remember when we invented “cloud”? Remember when everyone made a scramble to move to “cloud”? Build faster! Build cheaper! Get rid of your sysadmins, those mean salty dogs! And so, a lot of your CEO types (not all, but enough) had to go cloud native, a lot of them thinking they’d save money. After all, you don’t pay rent on a cloud. You don’t need heating, space inventory, access keys, facilities to deal with leaking pipes, and all those other headaches when you “live in the cloud”.

Except those savings never happened. But the savings weren’t the point. The point was the capabilities that cloud unlocked for you. You can scale systems up (and even down, if you’re mindful), you can push features faster, you can get more creative because throwaway prototypes are cheaper. Moving into “cloud” is an amazing feature unlock. But if you did it to save money, you did it for the wrong reason.

From where I’m sitting, how we think about AI in a business sense is going to go in a similar fashion. Prices for AI (however you want to measure it) have been artificially low. As soon as I saw I could get higher priority and larger limits for $20/month, I couldn’t pay you quickly enough. $20 a month is an insane value for what we’re getting. Even with the recent price increases, we’re still getting way more than our money’s worth. Eventually, physics and reality will set in and we’ll be paying ‘appropriate’ market rates (whatever that will even mean).

$20 a month is an insane value for what we’re getting

But to pull on the cloud analogy - if you’re “doing AI” to save money, you’re doing it for the wrong reasons. If you’re offloading humans in favor of AI, you fall into one or more of the following:

• You haven’t thought it through and are losing money and talent
• You’re a monster who doesn’t care about the people you’re trying to replace (and you still probably haven’t thought it through)
• You’re replacing a call center whose entire purpose can be automted by machines

True story on that last part - I had to schedule a service at my house not too long ago. I called the hotline and was greeted by what was definitely an AI model responding to my request. Completely without a human, I described the problem, scheduled the next available slot, and notes were made for the reason for my call. That last item is decidedly in the bucket of “no one really wants to do this, so we should get robots to do this”. But that’s another topic.

So what do you get when you let people have access to AI tooling? Everyone is their own developer (for better or worse), so ideas can quite literally come from anywhere now. You also have machines calling out blind spots in your schedule. And if you’re careful about it, you can actually save a little time after some thoughtful experimenting. Regular email updates come to mind here. There’s an art in making sure those emails still have some human warmth to them but, again, that’s another topic.

As we’ve read before, AI systems are the worst they will ever be today. It only gets better from here. Are there real-world consequences and pitfalls and dangers to this? Hell yeah there are. But in this post, we focused on the (incorrect) mindset that “AI will save you money”. We can talk about a lot of those other topics later. For now, keep learning.