Most People Don’t Need a “GenAi for Dummies” Book Anymore

I won’t apologize that this post won’t be a breathe of fresh air in the Ai conversation (although I do get why people want that). I am an Ai skeptic and I don’t hide that fact from anyone. Some things in this world are worth dividing over. I know people are tired of division and sides that everyone takes. It would be pretty easy as a white dude for me to sigh something about everyone taking a side and just accepting my Ai overlords like many do – but I also know there are many people that don’t have the luxury of avoiding sides. They are forced on a side just for being who they were born to be.

So, yes, I keep going back to the same old, same old Ai skeptic / critic / hater / whatever you want to call it side again and again because so many people that are harmed by Ai don’t have a choice to be on that side, or to even avoid either side. There is nothing special about my stance – it should be the baseline for most people: join with the vulnerable in fighting harm and injustice.

Yes, yes, I  know: not everything is that simple. I’m a huge believer in embracing complexity. But I also acknowledge that not everything has to be complex.

Or I could put it this way: the complexity I live with is that I don’t want to contribute to the problems that Ai is causing, but since Ai is almost everywhere these days, I don’t have that choice. I have to know how Ai works in order to know where it is in order to avoid it (or mitigate it’s effects as much as possible). But knowing how Ai works is not the same as actively, regularly using it.

This is what so many Ai Cheerleaders and Ai Both-sides-ers get wrong about Ai critics: we do know Ai. We don’t need a new guide to prompting better. We figured out that better prompting gets better results (but still not good results too often) about 30 seconds into our first time. And writing better prompts isn’t hard for many, especially those in academia that had to write papers constantly for years and multiple degrees.

I have seen well-meaning people get attacked for offering new “GenAi for Dummies” guides. Some of the attacks are pretty vicious and I truly wish that didn’t happen (and yes, I can say that while still disagreeing with the decision to create the guides in the first place). People are a couple of years and thousands of mansplaining hours into being told they just aren’t smart enough to figure out what is basically an autocomplete program (yes, I know Ai is a it more complex than prediction / autocomplete / etc – but that is technically what people are doing with prompting even if the underlying process is complex). Yes, they are smart enough. Please consider that most people that don’t like Ai are not simply in the “prompt harder!” demographic. I’m not justifying the vicious attacks of some, but I see where the anger comes from. Not all pushback is angry attacks, either.

I noticed this article in the Wall Street Journal about an Ai program that helps doctors write responses to patients: “After trying it for a few weeks, doctors said the drafts weren’t helpful and required too much rewriting.” This response to the article on Bluesky (the author set the post to only be read by those logged in to Bluesky, so you will have to go there to read it) is what so many of us say about these tools: yes, we have tried them.

Believe it or not, I do try to get out and talk to people that work with Ai. I try to make sure to talk to people that aren’t white men, mainly because they are the least likely ones to be harmed by Ai. I sat down and ate with someone that works at a very large tech firm here in my area. What they told me about their working conditions with Ai was frightening.

Like many companies of this size, they have an internal Ai solution for their specific field, as well as access to Copilot and other Ai services. They have a dashboard that tracks usage of their Ai tool, and additionally they have made overall Ai usage a requirement of work. Refusing to use Ai of any kind will become and HR and possibly a “reassignment issue.” I know that sounds pretty barbaric, but it is more common than you would realize in larger companies: use Ai or else.

Employees are told that the Ai-tracking dashboard is not punitive, just informational. However, the employee I spoke to had already gotten in trouble with multiple upper level leaders for not using their internal Ai. The kicker is that the internal Ai is designed for a completely different position than the employees works in. Yet they were called out publicly and harshly for not using it. This employee used it on the spot for a nonsense task, and then got praised.

However, that usage somehow did not register correctly and the employee still got flack for “not using” the internal Ai. In fact, several Ai-driven dashboards were not recording information properly. The employee had to have a long, contentious meeting with the upper level admin in charge of the dashboards to point out the problems, and it took a lot of back and forth to break through the guy’s confidence to make him see that it was not collecting data correctly.

On top of this, even when using the internal Ai solution for the tasks it was created for, the output is incredibly sub-par. It is the same story I hear over and over again: it takes longer to fix the Ai output than it takes to just do it without Ai. And since this company has such a specialized marketplace, tools such as Copilot and ChatGPT are fairly useless, because they weren’t trained on detailed enough data.

So, there you have the typical Ai scenario that I hear again and again:

  • The Ai solution (internal or external tools) does not do a great job
  • Ai usage is required by the company
  • Employees are told they are not being tracked for punitive purposes, but all tracking is used to punish and condemn those that don’t meet arbitrary standards
  • These arbitrary standards rarely have anything to do with the work people actually do
  • Explaining these problems to those that run the Ai solutions is usually fruitless
  • Workers get less work done in more time trying to deal with Ai, and then face retribution for not making quotas

I originally wrote this up a while back, and then life got in the way. I followed up with the friend that provided me with this information. It turns out their experience was not an outlier, but the norm for those that work there. However, they did find one way to use Ai: when upper level admins want some random summary or report, and they know that it is not going to be read anyways, they ask Ai to create the report. It is usually nonsense, but they have not heard any negative feedback from those that get the report.

Read what you will into that example of Ai usage and why Ai only seems popular with upper level admins. Everyone I have spoken to that is forced to use Ai is looking for other jobs away from Ai-driven employers.

I know that it is tempting to say this is more of a leadership issue than an Ai issue, and on some levels that is true. But if Ai was doing a decent job of what it was supposed to, all of this goes away. Workers would use it because it is actually helpful and none of the rest happens.

So maybe put yourself into the shoes of someone working in this situation and think about what it is like to be told again and again, over and over that you just don’t like Ai because you don’t understand it, or you need to prompt better. Maybe some are more than just a bit tired of hearing that again and again. Maybe they can’t really see a good side to Ai. Maybe they don’t want to choose sides, but they are forced to choose one thanks to bad leadership. Or maybe even they are just like everyone else in that they don’t like every single thing in the world. Ever wonder why we accept people having personal preference over almost every other type of technology… except for Ai? Why can’t people that don’t like Ai just… I don’t know… be allowed to not like it for any reason they want?