📈Rise of the Machines

Alex is up this week to talk about AI’s impact on a big part of his job.

Did someone forward this newsletter your way? Do us a favor and please subscribe. We grow through referrals and would love to have you on board. We promise we didn’t use AI to write this either.

---

I changed jobs about seven months ago.

I left a large, corporate broker-dealer where I managed the content distribution system to join a smaller investment management firm to serve as their vice president of marketing.

In doing so, I dodged AI eating a good portion of the corporate job’s responsibilities. I don’t know if my job would have been on the line, but it certainly would have changed.

Some context: translated out of corporate HRese, “content distribution system” basically meant I was responsible for a library of pre-written emails, blogs, white papers, and social media posts that financial advisors who subscribed to the system could use in drip campaigns to their clients. I had to keep all these pieces organized and up-to-date, as well as write new items.

The topics for most of these pieces were mostly personal finance-related. Think emails covering such scintillating topics as “should I invest in a Roth IRA or Traditional IRA?” or “what’s the deal with whole life insurance?” etc.

Unlike in this newsletter, I didn’t get to use memes or crack jokes or make anything fun. It was almost always “Joe Friday” style: just the facts (or use references that showed their age thirty years ago). The emails were the longest pieces we wrote, and they usually came in at around 400ish words. For the average reader, it takes about 2.5 mins to read 400 words.

2.5 mins to read what could take me 2 hours to write, with maybe another hour to edit with a colleague. On the one hand, the steady cadence of ‘write-edit-write’ made noticeable improvements in my writing; it also helps to have a fantastic editor (shoutout to Corey). On the other, it took time to grind out some of the pieces that required tracking down the most up-to-date facts and figures. Since the emails would be sent by financial advisors to their clients, the information had to be correct.

All in, composing or revising content in the system would take 10-20 hours per week, or around 40-120 hours per month. It may not have been the most creative content in the world, but it was accurate, reasonably fast, and as engaging as I could make it, given the constraints.

Then, OpenAI released ChatGPT. Watch this (sorry the quality kinda sucks¹):

If you can’t see it, I asked ChatGPT to write me a 200-word email discussing the benefits of life insurance in a simple way (the red square is my prompt). The text appearing below the green square is ChatGPT’s response.

This internet widget accomplished a task in 30 seconds that would have taken me at least 90 mins. And, in this case, it’s 100% accurate. In other words, this internet widget just collapsed 120 hours of work into maybe 2 hours of editing time.

It’s a little jarring to see natural-sounding writing appear on the screen like magic. Like I said, I don’t know what the implications of ChatGPT would have been for my corporate job. I do know that ChatGPT just drove the final nail in the coffin for freelance personal finance writers who generate content for sites like NerdWallet, the Balance, and Simple Dollar (some of whom I’ve written for). I say “final nail” because the owners of these sites were already paying effectively nothing for content; think 10 to 15 cents per word, or about $100 for 1,000 word article that might take you a few hours to complete. The market was already signaling that Joe Friday-style personal finance content is practically worthless before ChatGPT showed up.

Personally, I view the arrival of AI in this space as a good thing. Like any tool, if used effectively, ChatGPT will enable its user to accomplish mundane jobs more efficiently at a lower cost.

I think of it this way:

Most business writing needs to be fast, accurate, and engaging, in that order.

Humans haven’t been faster than computers for a long time. Computers are only as accurate as the parameters that guide their functions, so, there’s some room left for human oversight. “Engaging” will be the hardest for AI to tackle as it’s the most subjective.

If the difference between me and a computer is my capacity to express my thoughts idiosyncratically, then I still like my chances. In a way, that’s the big lesson I take from the Rise of Tech over the last 20 or so years: computers are scary and look like they’re going to eat the world, and then they do sh*t like this:

What’s the Upside?

AI tools, like ChatGPT, are here to stay and their abilities are only going to grow. CodePilot, a ChatGPT-like tool by GitHub has already automated a lot of the coding process. And yet, engineers are still handsomely paid by tech companies. Ultimately, computers are really good at doing the same thing over and over again; they’re really good at calculating probabilities in a narrow setting; they’re still not so good when they have to make a decision without clear parameters.

Thankfully, there’s still room for human error.

Now is probably a good time to let you know that an AI wrote this whole piece.

Just kidding.

For now.

---

For Your Weekend (AI Edition)

Read

The Rise of A.I. Fighter Pilots by Sue Halpern ($ New Yorker)

Algorithms are already good at flying planes. The first autopilot system, which involved connecting a gyroscope to the wings and tail of a plane, débuted in 1914, about a decade after the Wright brothers took flight. And a number of current military technologies, such as underwater mine detectors and laser-guided bombs, are autonomous once they are launched by humans. But few aspects of warfare are as complex as aerial combat. Paul Schifferle, the vice-president of flight research at Calspan, the company that’s modifying the L-39 for darpa, said, “The dogfight is probably the most dynamic flight profile in aviation, period.”

Watch

The Matrix (streaming HBO Max, for rent elsewhere)

Seriously, when was the last time you watched The Matrix? It’s the rise of the machines played out in the most mind-bending way. The movie came out in theaters in 1999 and still holds up 24 years later.

Chuckle:

1. I had to take a video of another video I had recorded earlier because ChatGPT was at capacity at the time I went to write this newsletter. Our robot overlords aren’t here yet!