Great analysis as always. As a professional in Artificial Intelligence/Machine Learning/Natural Language Processing, I am well aware of the differences between the Chatbot Large Language Models (LLMs) that are all the rage today and human cognition.
I agree with Gary Marcus that LLMs can't think. "Hallucinate" is too good a word. To use the direct terminology of the philosopher Harry Frankfurt, they Bullshit. People worry that they will be smarter than us. Not a chance. An Artificial General Intelligence is still a pipe dream.
This essay is a beautiful example of this. LLMs throw petabytes of data into the hopper and come out with a pile of conceptual hamburger. On the other hand, you are mining the Small Samples for tidbits of usable, actionable data. Sometimes you're right and you walk the bases. Other times, you walk back to the dugout. But you are doing something that a computer has yet to accomplish.
Antony, I always appreciate your insight especially on this piece that relates to your expertise. It is heartening to know there will always be a space for human excellence no matter how well we try and model it. I found the absence of a pattern, a connection can still yield a transformational moment in its singularity. In baseball, I have found so many of those standalone moments that ultimately told stories as well as the collective. It made us speculate, dream, hope, just as much as it kept us on edge or motivated. There are times when all we have, all we get is a small sample. That small taste can nevertheless change lives.
Thank you, Doug. I’ve been telling students this for years. Have we become too reliant upon data in our reporting? Why doesn’t it tell the whole story?
Great analysis as always. As a professional in Artificial Intelligence/Machine Learning/Natural Language Processing, I am well aware of the differences between the Chatbot Large Language Models (LLMs) that are all the rage today and human cognition.
I agree with Gary Marcus that LLMs can't think. "Hallucinate" is too good a word. To use the direct terminology of the philosopher Harry Frankfurt, they Bullshit. People worry that they will be smarter than us. Not a chance. An Artificial General Intelligence is still a pipe dream.
This essay is a beautiful example of this. LLMs throw petabytes of data into the hopper and come out with a pile of conceptual hamburger. On the other hand, you are mining the Small Samples for tidbits of usable, actionable data. Sometimes you're right and you walk the bases. Other times, you walk back to the dugout. But you are doing something that a computer has yet to accomplish.
Antony, I always appreciate your insight especially on this piece that relates to your expertise. It is heartening to know there will always be a space for human excellence no matter how well we try and model it. I found the absence of a pattern, a connection can still yield a transformational moment in its singularity. In baseball, I have found so many of those standalone moments that ultimately told stories as well as the collective. It made us speculate, dream, hope, just as much as it kept us on edge or motivated. There are times when all we have, all we get is a small sample. That small taste can nevertheless change lives.
Beautiful stuff! Great baseball writing on so many levels ❤️⭐️
Fascinating insight. I'll be reading this one again.
Thank you, Doug. I’ve been telling students this for years. Have we become too reliant upon data in our reporting? Why doesn’t it tell the whole story?