When I do talks, I often ask the audience “Without saying ‘Artificial Intelligence,’ what’s the definition of “AI?”
The audience often throws out lots of different definitions, all somewhat related, yet somewhat different.
Because of this confusion, we created our own definition. In AI as your Teammate, we defined it as “Data with a task.” I think that we can make it even simpler.
AI is computers doing stuff that humans used to do (or shouldn’t do anymore).
I find this definition supremely useful because it focuses on exactly what the value proposition is. AI is about saving time.
Most of the things that AI does are things that humans are doing now, and theoretically could do with enough labor / brainpower. When humans aren’t doing those things, they can focus on doing new things, and pretty much the whole world is powered on humans having machines do old things so we can do new things. The wheel, the pulley system, the printing press, modern software, search engines. In each of them, humans did the task first, then created a machine to do it better.
Second of all, I think this is a great definition because it grants permission to say, “this task doesn’t use my capabilities enough, and I think that a computer can do a good-enough job.”
What an empowering thing to say.
“I feel I’m being underutilized, but this has to get done, so I’ll have a computer do a good enough job.” With that attitude, the world is everyone’s oyster to maximize their talents and impact.
Third of all, this definition is agnostic of Repetitive Process Automation (RPA, sometimes called “Robotic Process Automation” but I like “Repetitive” better), Neural Networks, Large Language Models, Transformer Architecture, or the simple If/Then/Else statement.
The definition’s focus is on the impact it’s making.
AI is computers doing stuff that humans used to do (or shouldn’t do anymore).
Most AI right now involves automating necessary-but-low-value tasks that humans are doing. Especially in the small and midsized market, this is where the best ROI comes from. In the enterprise, there’s also lots of task automation, but there’s also data operations, insights, and intelligence.
The focus is on the job getting done, not on the means for doing the job.