Listen to this article:
I don’t think I’ve ever seen a technology develop and expand and proliferate this quickly. Remember back in the ancient days of 2023 when it was pretty much just ChatGPT? Did you notice the following year when Microsoft added Copilot into many of its products and Google started incorporating AI Overviews into their search results? And then new models for image generation, video generation, text to speech, speech to text, application development, application deployment, and more. I can’t afford subscriptions to all of these. Neither could anyone else, apparently, because now there’s a proliferation of AI model aggregators. Be sure to subscribe to one of the hundreds, perhaps thousands, of AI email newsletters to keep you up to date on the new models. We’re going to need email newsletter aggregators to keep us up to date on the newsletters keeping us up to date on the models. And then there’s the AI agents automating and accelerating routine business tasks. My graduate research focused on Artificial Intelligence and Neural Networks, so it is very exciting for me to see the technology blossom.
It’s also a little exhausting.
The state of AI today reminds me of an experience I had at an Australian Rules Football game.
I started watching Australian Rules Football in middle school. The neighborhood had just been wired for cable TV, and a new all-sports startup channel, ESPN, was trying to fill 24 hours of airtime with whatever programming they could inexpensively acquire. They regularly featured high school football games and rodeo. And Australian Rules Football (AFL). I loved watching Australian Rules Football.
On weekends, I would get together with neighborhood friends to play (American) football. My next door neighbor, the same age as me, had moved in a few years earlier from Scotland. He played rugby before his family moved to the States, and he also liked Australian Rules Football. So, while we were waiting for everyone to arrive, several of us would go running up and down the field making “shocking hand passes,” bouncing the ball very 15 meters (or at least trying to since an American football doesn’t bounce the same way as an Australian football), and kicking the ball just outside the goal post for a behind.
Many years later, I was invited to give a conference presentation in Sydney. The one request I had for my hosts was to arrange for my wife and me to go to an AFL game. A whole group went to see the Sydney Swans play the Hawthorne Hawks at the Sydney Cricket Ground. I loved every minute of it. On TV, the camera follows the movement of the ball, but there’s a whole lot more going on elsewhere on the field; probably more than any other sport I’ve watched.
What does this have to do with AI? I’m getting there.
It was a thrilling game, with the Swans winning by a single point, 80 to 79. Ordinarily fans aren’t allowed onto the field, but on Sundays you can bring your ball and play after the game. Considering the number of fans on the field, I’m not sure that anybody left the stands through the exits that afternoon. The oval was full of people kicking and punching balls to each other. People there could kick the ball to each other 20 years apart just as easily and accurately as folks here might throw a baseball or football.
My wife and I decided to walk a lap around the edge of the field where there might be fewer balls coming at us. That was mostly true, until we got back behind the goal. Balls were coming in from every direction. Of course they were. Everybody wanted to kick a goal. A few times I had to punch away an inbound ball. It felt like being inside a popcorn machine.
Get the AI connection?
Incoming approaching from every direction at the same time, being in pure reaction mode, and having to make last-second saves to avoid being hit exactly describes AI right now. At least for me.
My wife and I managed to make it all the way around the oval intact and unscathed. Maybe AI will be like that, too.
We have to be aware, and we have to avoid falling into the trap of continually pursuing the next thing.
Keep in mind that your objective is to accomplish something. Never ending tool evaluations don’t accomplish anything other than evaluating tools. I’ve known many that have been comprehensive and produced great research, but by the time they had produced their polished presentations, the technology had moved on. The conclusions and recommendations were well-reasoned but obsolete. I’ve seen evaluations reopened, with the vendors leap-frogging each other until management makes a (too often arbitrary) selection.
When it comes to AI, make the best choice you can. Quickly.
Pick a direction and accomplish something.
Deliver business value. There will always be something better or faster or more sophisticated than what you picked. And if there isn’t today, there will be tomorrow. But once you start, you can rationally and objectively assess the benefit of moving to a different tool against the cost, including the lost opportunity cost associated with diverting resources.
After all, the AI isn’t the objective. Delivering business value is.
