What AI is Not: A Simple Introduction

To talk about what AI is, it may be helpful to outline first what it isn’t.

It’s not magic. It’s not some kind of devious robot preening its fiber optic mustache in the dark corner of our lives. Truth be told, it’s not even that smart.

Despite impressive achievements and headlines, today's AI systems are far from the all-knowing, conscious entities often portrayed in science fiction.

Here's why:

1. Narrow Intelligence: Most AI is designed for specific tasks. An AI that's great at chess can't suddenly decide to write poetry or drive a car.

2.  No True Understanding: AI doesn't truly "understand" information like humans do. It processes patterns in data without grasping meaning or context.

3. Data Dependence: AI's performance is only as good as its training data. It can't really innovate beyond what it's been taught.*

4.  Lack of Common Sense: AI struggles with simple tasks that require common sense reasoning, which humans find intuitive.

5. No Consciousness: AI doesn't have self-awareness or emotions. It can't think abstractly or have original ideas.

6. Fragile: AI systems often fail when faced with scenarios slightly different from their training data.

7. Bias and Errors: AI can and often do perpetuate or amplify biases present in its training data, leading to flawed outputs.

*If I had more space (or was smarter), I’d provide more nuance here: Some super advanced AI systems can generate so-called novel combinations or ideas within their domain, which I suppose can be seen as a form of innovation.

Here is what AI is pretty good at:

1. Process complex information

2. Recognize patterns

3. Make decisions based on data

Try to think of AI as a skilled mimic. It can replicate human-like behaviors and outputs based on what it has observed, but it doesn't truly understand or internalize the meaning behind these actions.

Some of the most common use cases for modern AI:

· Email spam filters

· Netflix movie recommendations

· Voice assistants (Siri, Alexa)

· Autocorrect on your phone (dear iPhone: it’s never “duck”)

· Fraud detection for your credit card

The reason I start this series here is because I believe one of the most important things an and individual and/or an organization can do prior to implementing AI is, as AI Advisor Jason Gulya notes, “deal with the emotions” (which are well-founded in many cases) surrounding AI: fear and stress of displacement, mistrust of algorithmic decision making, etc.

This will be subject of our next two-minute installment.

Andrew, North Light AI

www.NorthLightAI.com

The Lightwave strives to deliver practical AI insights to busy professionals in concise, two-minute reads. We cut through the jargon and hype to provide clear, balanced explanations of AI concepts, applications, and implications for business. Whether you're AI-curious or AI-skeptical, our newsletter equips you with the basic essential knowledge to navigate the evolving AI landscape efficiently and confidently.