The Issue with AI
This is my newsletter issue that covers AI. But I also mean it the other way.
This issue contains about 1,800 words. It features 26 links that lead to pages with more words and medium-length videos. What I have picked for this issue comes from many months of reading about AI. I spent the time to read each piece, think about them, forget about them, then go back to them to see if they are valuable. I then picked the ones that I thought are worth your attention.
So if you are considering using a language model to summarize this material, please don't. Because that instinct is the part of the problem. Here is why:
AI of today is more buzz than technology
The field of artificial intelligence is amazing. For many decades, devoted scientists worked hard to tackle crazy complicated math and philosophy problems to advance computational intelligence. They worked on many competing and at times contradictory approaches to unlock the AI technologies that we use today. We should recognize them and thank them for their efforts.
Watch this 17-minute video with patience and attention to appreciate the level of sophistication these scientists have been working at for decades. Please see for yourself that there was significant AI work many years before ChatGPT.
But today, we talk less about the fantastic scientific endeavors that made today's AI systems possible and more about the buzzy side of AI.
Do you realize that we call everything AI today? Detecting what is in an image used to be called computer vision, now it's AI. Creating layer-based computational models for processing data used to be called neural networks, now it's AI. Processing data for predictions and insights used to be called machine learning/deep learning, now it's... Heck I saw an AI-powered muffin the other day.
Why do we do that? Because the companies and venture capitalists who have poured millions into an important but consequential step in computer science wants to pump this progress into a frenzy. People in power needs a reason to raise and spend more. Someone needs to pay. That's us.
Intelligence in "AI" isn't the same intelligence in humans
The intelligence in generative artificial intelligence isn't like human intelligence, cat intelligence or even fish intelligence. These models do not think. Instead, they predict likely outputs based on inputs and parameters through intricate, complex mathematical models.