(This is part 2 of a 5-part series of posts attempting to explain the AI landscape in a practical and simple way. See other posts here.)
When I first started exploring AI, I found that my biggest problem was figuring out where to start. I’d have one conversation where I felt like I was making progress, and then I’d stumble across a paper that would say something like:
It’s easy to give up when the paper you’re reading starts talking in wingdings. But just like we learn by doing, I’ve found it helpful to mix theory with practical things I can be thinking about or action steps I can take to make concepts real. If you’re like me, it may help to start with basic addition before you get into trigonometry.
I have a few tips on what you can be doing today:
Forget stealth mode; be inclusive.
If you’re a machine learning developer building new technology, there may be instances where stealth mode is the right move. But if you’re a machine learning developer building new technology, you’re probably not reading this.
What I’ve seen happening is that, in most organizations, there has been enough buzz about AI to have some secret project team having secret meetings of top-secret importance. In my view, these organizations have the cart way before the horse, and the only real purpose the secrecy serves is ego-driven (after all, only the smartest of us are invited into the top-secret project room 🙄).
Rather than trying to figure out future applications of a technology most people in the room don’t even understand, you’d be much better served figuring out where you’re already behind today, which is going to be an inherently inclusive process that consists of…
As of December 2023, there were more than 180 million ChatGPT users, and there’s probably a lot more than that today. That means whether you are aware of it or not, you probably have team members using GPTs to help with some day-to-day tasks.
Now in early 2024, I’ve heard of very few companies working to foster widespread AI Literacy in their teams through education and training or that have taken any real steps to provide AI governance.
Do all of your team members know what happens to the data they enter into a GPT prompt? Maybe James, the Intern, uses it to help him write a nice summary of his meeting notes, and now those internal meeting notes are helping to train the next iteration of ChatGPT.
If you have a design team, is every person aware of the most recent US Copyright Office guidance that any image produced even partially by generative AI cannot be copyright protected? This is essential info if you’re designing a logo for a client.
You’re already behind on this stuff. You don’t need to panic, because everyone else is, too. But your organization would likely be better served by taking the AI conversation as public as possible, educating your people and implementing AI governance, rather than being too narrowly focused on and secretive about some clever application that will likely only become relevant after the other things I’ve mentioned become real problems to deal with.
You don’t usually buy a new gadget and then figure out what to use it for, so don’t approach AI that way, either. If you’re thinking about how to apply AI within your organization, start by understanding what your various teams spend the most time on. Don’t assume you know, either. Include them.
Once you identify the most time-consuming (or least enjoyable) aspects of their workflows, you can assess what AI-powered applications can assist. More than likely, the first AI-enabled programs you scale in your organization aren’t going to be sexy, but rather basic stuff to help your business run more efficiently. And that’s okay, because we all need to learn to write–full stop–before we can write Hamlet.
In case you’re wondering, I pulled the equation at the beginning of this post from this paper talking about how a hypothetical, entirely-AI tech company (“ChatDev”) was able to design, code, test, and document complete software on its own (in under 7 minutes). If you’re interested, there’s a clever video explaining it here.
Next post I’ll get into more practical implications Gen AI will have within the next few years.