In a continuation of the Lego analogy, Ian explores the two main classes of artificial intelligence. Learn why custom-manufactured Lego pieces are like narrow AI, and why general AI is more like a bucket of liquid plastic.
- Welcome back to our LEGO workshop. So we've discussed how generic LEGO bricks approximate enterprise software tools. Now here's where things get exciting. Today's LEGO sets, like this Hogwarts Castle set, now comes with custom LEGO pieces, like this magic crystal, this owl, and this sorting hat. Which really can only can be well, a sorting hat. I think of these custom LEGOS like narrow artificial intelligence.
Today's AI capabilities are all more or less narrow in the sense that they're algorithms that serve a discreet and specific function, given context and data. Much like this sorting hat can only really be, well, a sorting hat, an algorithm that scans medical images for traces of cancer for example won't be of much use in recommending a movie for you to watch or interpreting an insurance claim. So these custom pieces can come in handy, but are generally reserved for specific situations in which generic blocks don't suffice.
Let's go back to that employee onboarding example for a moment. In that scenario, there are times when we may need to interpret resumes that have been scanned and sent by our external recruiter. Traditionally, we've used people in this role, like Hermione here. For that, we'll now use an intelligent OCR, or Optical Character Recognition tool, which will be represented by the magic crystal. And here, where Hagrid was assessing candidates, perhaps we can use another algorithm trained to look for certain skills we know to make the ideal candidate.
Hello sorting hat. And later on in the process where we'll find Dumbledore, we may now employ AI to oversee and look for suspicious patterns in payroll across the whole organization. For that, we use this owl piece. I've seen OCR and machine learning deployed in exactly these HR solutions. Now, to extend the LEGO analogy one step further, we might consider how general AI fits.
General AI, or Artificial General Intelligence, is amazing. It would be able to perform any task that you and I would be able to do. It's the thing dreams or nightmares are made of, and what Hollywood movies are all about. But it takes a lot of work. While narrow AI was represented by these specific plastic pieces, think of general AI like a cup of liquid plastic.
What I can do with it is nearly endless because it isn't constrained to being a block, nor stuck being a specific crystal, owl, or sorting hat. But what it requires having is the right people and much more expertise in design, planning, and knowledge to shape it into the thing we're going for. In other words, we often have an enthusiasm around what's possible with general AI, but an underappreciation for the difficulty in turning such a malleable substance into something useful.
So look, this is clearly an overly simplistic analogy and a great way to justify playing with LEGOS. But I hope it illustrates some basic points about how your tool kit will include configurable enterprise software and automation tools, cognitive tools like narrow AI, and progressively more powerful, general AI technology in the future. This sets the stage for the following lessons on how to make the most of the tools you have available to your organization today.
- Understanding the digital operations landscape
- Where digital ops fits in your business
- Digital ops front-office and back-office use cases
- Drafting a strong, top-down mandate
- Investing reasonably in digital transformation
- Managing expectations for AI projects