How We Use ‘AI’

I’m tired of hearing and talking about “AI” (quotation marks are on purpose). Yesterday morning Mark Zuckerberg, the most trustworthy personality in tech, wrote a 6,500 word essay titled “The Future is for Everyone.” Reporters are calling it a manifesto and critics are wondering why the heck he didn’t use “AI” to truncate the behemoth of a post. Never use two words when one will do, Mark.

I tried to read it but only made it about four paragraphs before nearly dying of boredom. I don’t want to read a tech bro’s vision for the future. Especially when it’s a tech bro known for creating and then being sued for addictive products that turn our brains to mush.

It’s also off-putting when these tech bro CEOs speak only in the future tense. For almost four years all we’ve heard these guys talk about is “what’s next.” Tell me what “AI” is doing today that isn’t slop. Slop is anything that takes longer to read, watch, or listen to than it took “AI” to create. Slop is crap and nobody likes it. If it’s the crown jewel of “AI” then what the heck are we doing?

I hear that soon it’ll be agentic and act as my very own personal assistant. I can’t wait to see what kind of slop my fake PA puts into my calendar. If it makes a mistake, it’ll tell me, “You’re right. What else can I help with today?” And it won’t take a lick of ownership for anything it screws up. It won't be a PA. It’ll be a sycophant feeding my false confidence.

Does the current version of “AI” provide value? My goodness it does. But only if it isn’t sycophantic slop. I’ve written a bit about how we use “AI” in our business, but I think it’s important to share how we’re using it. We will never feed you slop.

Communications you receive from our team are written by us. Meeting summaries, meeting reminders, market updates, emails, retirement income projections, and financial plan executive summaries are not generated by a robot.

But because we know you’re using the likes of ChatGPT or Claude or Gemini, some of what we send you is sent to a robot with the following prompt first: review the following communication assuming the person/people we send it to will ask an LLM to critique it. Find any holes, errors, or unclear statements.

These posts are reviewed – but not written – by a robot. This business has a ton of rules. There are a lot of things I’m not allowed to say and there are even more things I’m allowed to say but shouldn’t. I upload my completed posts to a robot and ask it to check spelling and grammar, flag passive voice, and make sure the post won't embarrass me later or get me in trouble with the regulator.

Data and history are double-checked, but we always ask for sources. I love this part of “AI.” I can ask it to scour the internet to find me specific details about the past. It’s super google. Instead of me clicking through pages of google search results, it’ll do it for me. But I always ask for the source, because it often makes things up. Does it save me a ton of time? Not really. But it does save me a bit of time.

I don’t want to be fed slop. And I’ll never feed you slop. We’ll use “AI” behind the scenes, but we’ll never use it as a shortcut. What you get from me is created by me and it always will be.