Back in April 2023, I posted a list of principles for creating with AI, attempting to separate fact from fiction and find some universal truths about working with AI.

How did it hold up?

For the most part, I think it stood up well. But there’s one thing I got very wrong and I’m furious about it.

I had the opportunity to talk about it all on the Media and the Machine podcast (subscribe here to get it when it drops). To prep for the podcast, I revisited those principles and added a few notes:

1. AI tools will continue to improve.

2. The rate of improvement will continue to increase.

Still true. If you think the rate increase isn’t happening, you’re probably only looking at generative AI (and only from U.S. corporations) — and even then it’s debatable.

3. Because of 1 and 2, never say, “AI can’t…”

People should know this by now. Extend the timeline out 10, 100, or 200 years. What can you confidently say AI will not be able to do?

4. If everyone has access to the tool, the tool is not a competitive advantage.

I’m surprised by businesses that still don’t understand this point. Using AI isn’t enough. That’s table stakes now. What else are you bringing?

5. AI seems to have a bias towards abundance. It seems to abhor scarcity (the driver of value) and constraints (the driver of creativity). Use this.

Generative AI’s bias towards abundance is probably the catalyst of most hallucinations. A confident, wrong response or action will keep subscribers longer than a thoughtful, “I don’t know.” No one in charge of an AI company can afford to understand the value of scarcity or constraints right now.

6. AI will be incorporated in all your work tools. It will be expected as a part of your process, even if it isn’t. To be competitive for clients, you should master the tools more than the next person.

This may have been the most demonstrably accurate of the bunch.

7. AI companies are still hiring writers, content strategists, and designers for a reason.

Some of the higher-paying content positions I’ve seen lately come from industries that have the highest levels of AI adoption. I guess they learned #4 pretty quick.

8. Who did you turn to for trusted info on AI? Not AI.

“Trusted” does a lot of work in this one.

9. AI will take jobs quickly, but probably not careers — as long as relationship-building remains the focus of the career. Stick to your principles. Principles are useful constraints, not likely to be shared by AI anytime soon.

AI will take it all — at short-sighted companies. AI is already a net killer of jobs on a grand scale. But see #4 and #7 above. After we go through these hard times, if there’s a re-hiring frenzy, it may be met with yawns from empowered free agents. Forcing out employees with a deep knowledge of your industry, and access to the same AI tools used to replace them, is a recipe for creating thousands of competitors — competitors who can complete projects faster and cheaper while earning more. Companies really aren’t thinking this through.

10. Process pays no mind to AI. Your process is guided by the constraints that work for your brain, without regard for whether AI can help. Maybe it can. Most likely it will also remove vital constraints if you’re not careful. But, see number 3. Never say never. Experiment.

True, but too nuanced. It doesn’t make a lot of sense to build models designed for your specific brain…yet. So, don’t assume any kind of automation makes sense for you without experimenting.

11. AI is an executive function aid. It will be embraced for accessibility purposes, and put to use in K-12.

I was dead wrong about this and I’m furious this isn’t happening. Rather than embracing the positives of AI as a learning tool, schools have made it a bogeyman. They are failing to prepare our kids for real life. On top of that, they’ve conditioned students to accept extreme amounts of surveillance in their daily lives — tracking every keystroke (and sometimes much more) to ensure AI wasn’t used. I’d rather our schools graduated individuals capable of independent, critical thought — with experience in using whatever tools help them succeed in the real world. Which leads me to the next principle.

12. AI does nothing to solve the hardest part of creating (for yourself or for clients): having the courage to express an opinion in public. “Voice” is not opinion.

This is so true that it has a name now: slop. There’s a huge opportunity for realness out there.

13. People still prefer to buy from people, even if they also buy from AI. Adding real people to the mix will drive more value over time.

14. In B2B, AI can open conversations and create interest. Closing still happens between humans — sometimes entire committees. The companies who are fastest in getting a human involved with the customer are likely be more competitive. Use AI to quickly get that human in front of that client.

This is another set of principles that’s so true, it led (at least in part) to a new business term: the obelisk structure. I think that will eventually fall away, though. With this new structure, we’re admitting now that some people are indispensable for now. We’ve got a long way to go before businesses just say, “people are indispensable.”

15. With so much new content about to be added to everyone’s information diet, trusted editors will rule for the foreseeable future.

Cutting through AI slop will become a job in media with openings for every niche topic of interest.

16. True curation requires trust and opinion. This is a weakness for AI and an opportunity for creators. Invest in curation. Email newsletters still return the best ROI for curated content.

No changes since 2023. Actually, no changes since the 1990s on this one.

17. Humanization of content beats personalization and automation.

Never truer. Never more ignored.

18. Most businesses have short-term biases. Cost-cutting with AI will come before revenue-boosting.

We are deep within this moment now. I wish I hadn’t been right about this one.

19. It was Content Strategy. It’s now Content Strategy. Even the greatest AI-wielding creatives have terrible implementations. Again, editing will be a more desired skill. Strategy is the competitive advantage.

I think this is now widely understood. It’s just not widely implemented. That’s this era of AI in a nutshell: we understand more, but we’re not acting like it.