Alan Shimel wrote a post that really got into my head. It’s titled And the Cockroaches Shall Inherit the Earth. You should read it, it’s good, but the TL;DR is that even though AI does a mediocre job at most things, that mediocre job can be good enough. The idea of “good enough” stuck with me and really has me thinking about what it means. But probably not in the way Alan intended.
The thing I keep wondering is: What if we just stop doing “good enough?”
Lately, we keep hearing an argument that AI can do certain tasks good enough. We hear all the time that AI doesn’t need to be great, it only has to be good enough to replace a human also doing work that is good enough. Why should I pay a human to do the job when I can have a robot do it for a fraction of the cost? This is the classic mechanization idea that started with the Industrial Revolution. A machine that can do the work of many humans is valuable. An excavator is worth tens or hundreds of people with shovels.
But there’s another angle to this in the world of tech. How many of the good enough things do we do because everyone else is doing it? Does everyone need a blog, and a YouTube channel, and a newsletter, and a podcast, and every possible social media account? These days in the world of business it’s easy to have Fear of Missing out (FOMO). If everyone else is doing something, FOMO will tell you need to be doing those things too.
With AI we can solve our FOMO by publishing lots of good enough content in all the places. It doesn’t matter if nobody will ever look at that content, the point was to publish.
Let’s use Techstrong as an example, or maybe the anti-example. The interviews that Techstrong does aren’t “good enough”; they are well beyond good enough, they are great. Nobody wants to watch a good enough interview, or watch a good enough video, or read a good enough article or blog post. It’s boring and you’re not going to learn something new. It doesn’t matter if it was created by AI or a person. Bad content is bad and just because you can create bad content cheaper and faster than ever, that doesn’t somehow make it good. One thousand bad interviews don’t add up to one good interview.
If you look through YouTube or iTunes, you will find a wasteland of good enough content. Low production quality, people who didn’t care but still published something. People who did care but didn’t stick with it. High production quality but terrible content. It’s a near infinite combination of things. The one trait they all have in common is no audience. Nobody wants podcasts and videos that are only good enough.
Back to the Techstrong example. How many other companies tried to make a business out of interviewing people and failed? It’s probably a pretty large number. When those companies started, not one of them thought, “This is the most mediocre idea ever!” Everyone who tries to start up a new company, blog, project, or anything really does so because they think they have an exceptional idea. But even an exceptional idea with mediocre execution will never be great. We also need to understand that we can’t be exceptional at everything; we can be exceptional at one thing. Everything else is a distraction. There’s a reason Techstrong doesn’t also sell shoes. They know what they’re good at.
So why do we think we think it’s OK to have a website filled with mediocre content? Or a blog that we know is watered down to a point there’s nothing useful on it? If your website is a gigantic word salad, who is it for? This isn’t really an AI problem; it’s a bad content problem. It’s just more obvious now because of AI. So why are you doing it?
I’ve asked the “who are you doing this” question in the past, unrelated to AI, and the answer is always the same: But I need the content so the internet robots can find me! The general idea was that you need a lot of pages that can trick the magic gremlins that scrape the internet to feed the search engine and AI models. The more videos, and podcasts, and blog posts you have, the more internet points you will earn!
But there is another way. A way we don’t seem to hear about all that often. While you can try to game the system, you can also earn it, for real. If you are creating high-quality material that others will seek out, and tell their friends, and talk about it on social media, and link to from other blogs. You’re not playing the SEO game anymore, you ARE the SEO game. Great content builds an audience and makes you a trusted source. Who is your audience or customer, and why should they come back for more?
I’m not an expert in the world of SEO, but it seems to be an accepted idea that getting mentioned in large, popular sources is much more valuable than trying to slop your way into a decent SEO ranking. You know how you get mentioned? By being great. No reporter has ever gone looking for good enough people to interview. They want to interview the people who are on the top of the list, experts in their field.
All of this brings us to the question: What if instead of publishing two AI-generated blog posts per day, you just stop doing it? Would it really matter? Would you know if anything changes?
It’s really hard for us to stop doing things. We have visions in our minds of how taking an action, ANY ACTION, we create a positive result. If we are doing something, it must be important; another blog post, even if it’s mediocre, is progress. A lot of this mediocre content isn’t driving business needs; it’s a Skinner box. A Skinner box was used to study pigeon behavior. When given rewards at random, they would develop elaborate rituals that didn’t work because the reward was random. But the pigeons thought the rituals worked. We are the pigeons looking for a treat, and publishing content nobody reads is our reward.
This doesn’t just apply to generating content. There are a lot of things a typical business can stop doing and it would have zero negative impact on the bottom line. It could be too many meetings, long, useless product documentation and user stories that nobody uses or reads. If you have a task where AI could replace a human, maybe that’s a task you don’t actually need to be doing. Putting energy into things you don’t need to do, by definition, takes energy away from things you should be doing.
So when we put all this together, how can we figure out when “good enough” isn’t good enough but actually pointless. There’s never going to be one simple answer. It can also be really hard to measure what works and what doesn’t work. It’s a variation of that old advertising saying “Half my advertising spend is wasted; the trouble is, I don’t know which half.”
Rather than starting the conversation about how you can use AI to generate more … stuff, maybe the first question should be “do I need to be doing this at all?”
