Cautious Optimism About AI
Nov 17, 2023 3:11 pm
It's almost Thanksgiving here in the US, which means lots of planning for food and cleaning the house for guests. In fact, I'm in a hurry to get this newsletter out because I have folks coming to clean the carpet in an hour.
AI is all the rage right now, which was preceded by Machine Learning, and before that, Data Science and Data Lakes, and so on.
However, where AI is different from the other hype trains that people hopped on is how immediately accessible it is and how much promise it shows at first glance.
I've been playing around with my subscription to ChatGPT and asking all kinds of things about it, and I find it fascinating and troubling at the same time.
First, the fascinating part is how rapidly it can come up with a reasonable answer to many questions. It is like a conversation with an approachable Wikipedia. I think that it is really quite handy to have even less friction when it comes to getting information that has been put together coherently.
As models grow, change, and specialize, and new variations on this particular form of AI emerge, we're going to see incredible growth in the use and maturity of AI in all areas of technology.
However, there are some serious issues that are already apparent that may never fully work. Today, what AI can create is far from perfect. Sometimes, it is dangerously imperfect. In fact, people have shown that current models still fall victim to human bias in their responses, which can cause them to make mathematical mistakes. Its products, while interesting, show a lack of actual expertise in its knowledge or ability to create.
The bottom line here is that AI still needs human experts to analyze, edit, and rework the basic products of AI. While this isn't bad, most experts spend their career on the creation and not solely the editing of whatever their field is.
Let's take software as an example. You can ask ChatGPT to create code for you, and it will even run. It will have problems, though. It will take an experienced developer to spot the problems as they incorporate it. This would be akin to a code review, which many companies leverage, but is a very leaky bucket for solving quality issues.
This use of AI puts software developers, whose careers are mostly around the creation of code, into a position of editing code. This is not what they're actually skilled in or experts in, which means they will miss many of the issues that come with leveraging AI coding tools.
So, while I'm excited about what AI can do and will grow into, I'm nervous about what will happen in the next few years as folks trust the products of AI far more than they may deserve and as industries everywhere shift to needing editing skills instead of creation skills.
Here's my weekly update for November 17th, 2023...
For years, folks have struggled with phrases like “self-organizing” and the newer “self-managed” regarding teams.
I have met plenty of leaders and managers who have tried to figure out how to create teams with these qualities and find their own place when they work with teams like this.
In this article, I want to talk through a few things to consider if you’re a leader interested in building self-managed teams.
**Scattered across almost any consultant’s PowerPoint decks will be several quotes.
There’s the infamous, “If we asked what they wanted, they’d have said a faster horse,” and “Everyone has a plan until they get punched in the mouth,” and the one I’ll be talking a bit about today is, “A bad system beats a good person every time.
Let’s face it: agile conquered the world of software development.
Often, though, I find people are dissatisfied with the results, and in this article, I want to point out a few common things that folks struggle with and what you can do to get some of the results that Agile or Scrum promised.
PS: I'm looking for new clients. I know companies everywhere are preparing for a recession, but the expectations to deliver results aren't changing either. If you find yourself needing to rapidly grow what your teams are capable of, need help getting to market sooner, or solving quality issues we should talk.