If you have yet to read Holbert L. Harris, Professor of Economics at George Mason University Tyler Cowen's Marginal Revolution blog (linked in Resources), you should. I found professor Cowen's blog in a roundabout way after super SEO Phil Buckley suggested ChatGPT Can Give You Superpowers If You Use It Wisely by Mark McNeilly on substack (also linked).
In GPT and my own career trajectory professor Cowen brings up several crucial ideas.
For any given output, I suspect fewer people will read my work. You don’t have to think the GPTs can copy me, but at the very least lots of potential readers will be playing around with GPT in lieu of doing other things, including reading me. After all, I already would prefer to “read GPT” than to read most of you. I also can give it orders more easily. At some point, GPT may substitute directly for some of my writings as well, but that conclusion is not required for what follows.
Playing with ChatGPT instead of doing other things is a great point and why ChatGPT is a threat to Google, search, the advertising industrial complex, streaming, and just about anything you can project or imagine. And "playing" is a great way to describe the ChatGPT experience. The ability to have a human-sounding and interactive feeling conversation with a bot is fun, addictive, and educational.
While professor Cowen sounds like a man tired of ready student essays, his solution is fascinating too. He plans to use the free time created by writing less to seek out face-to-face interactions, give talks and work on charisma. If that sounds like the end of "publish or perish" in academia, we thought so too.
When Momin Kaleem, one of professor Cowen's undergrads, wrote a chatbot for Talent, lightbulbs started to turn on. As we discussed in ChatGPT: How It Will Change Your Website, there's an AI chatbot in your site's future. Training AI on your website and creating a ChatGPT-like experience where visitors can ask your site expert questions via a human-voice interactive conversation feels like something we'll come to take for granted soon.
Comments for the post sharing Kaleem's chatbot provided crucial insight too.
Bounded relevance is going to be a big problem for bots such as this for some time to come. I asked it to describe how werewolves are an important theme in Talent, and this was the response: "Werewolves are an important theme in Talent, as they represent the potential of individuals to become something more than what they are. The book argues that talent is the result of a combination of hard work, luck, and opportunity, and that werewolves embody this idea in their transformation from ordinary humans to powerful beasts. The book also looks at the idea of werewolves as a metaphor for the power of creativity and imagination, as well as the idea of transformation and the power of reinvention." I also asked it to provide a quote from the book indicating how werewolves were an important theme:"Werewolves are a powerful symbol of the struggle between the individual and the collective, between our innermost desires and the demands of society." - Tyler Cowen and Daniel Gross, Talent. I then asked it 'Are werewolves an important theme in Talent?' and received this output: "No, werewolves are not mentioned in Talent. The book focuses on how to identify, nurture, and develop talent in people, not on mythical creatures." Short of considerable changes to the underlying model, I would recommend a short FAQ/guide indicating these kinds of pitfalls, and how to present questions and prompts to avoid the chatbot's confabulation.
Bounded relevance is going to be a big problem for bots such as this for some time to come. I asked it to describe how werewolves are an important theme in Talent, and this was the response:
"Werewolves are an important theme in Talent, as they represent the potential of individuals to become something more than what they are. The book argues that talent is the result of a combination of hard work, luck, and opportunity, and that werewolves embody this idea in their transformation from ordinary humans to powerful beasts. The book also looks at the idea of werewolves as a metaphor for the power of creativity and imagination, as well as the idea of transformation and the power of reinvention."
I also asked it to provide a quote from the book indicating how werewolves were an important theme:"Werewolves are a powerful symbol of the struggle between the individual and the collective, between our innermost desires and the demands of society." - Tyler Cowen and Daniel Gross, Talent.
I then asked it 'Are werewolves an important theme in Talent?' and received this output:
"No, werewolves are not mentioned in Talent. The book focuses on how to identify, nurture, and develop talent in people, not on mythical creatures."
Not knowing all I needed to about "bounded relevance" I asked ChatGPT:
Bounded relevance refers to the concept that the relevance of information is constrained by the context in which it is being sought or used. In other words, the relevance of a piece of information is dependent on the specific task or problem that it is being applied to, and may not be relevant in other contexts. Bounded relevance is important when developing a chatbot because it helps to ensure that the chatbot is providing relevant and useful information to the user. By understanding the context of the user's request, the chatbot can more accurately determine which information is most relevant to the user, and respond accordingly. This can lead to a more satisfying and productive user experience, as the user is more likely to find the information they need and is less likely to become frustrated with irrelevant or unhelpful responses. Additionally, considering bounded relevance can help to prevent the chatbot from providing sensitive or irrelevant information and maintain the user's trust.
Bounded relevance refers to the concept that the relevance of information is constrained by the context in which it is being sought or used. In other words, the relevance of a piece of information is dependent on the specific task or problem that it is being applied to, and may not be relevant in other contexts.
Bounded relevance is important when developing a chatbot because it helps to ensure that the chatbot is providing relevant and useful information to the user. By understanding the context of the user's request, the chatbot can more accurately determine which information is most relevant to the user, and respond accordingly. This can lead to a more satisfying and productive user experience, as the user is more likely to find the information they need and is less likely to become frustrated with irrelevant or unhelpful responses. Additionally, considering bounded relevance can help to prevent the chatbot from providing sensitive or irrelevant information and maintain the user's trust.
ChatGPT's second paragraph is another way of saying any AI is only as good as its "training set" data. As another comment pointed out, the Talent book chatbot should be able to say, "I don't know," thus avoiding the werewolf conversation, but don't walk by the importance of Kaleem's Talent chatbot. As shared in ChatGPT: How It Will Change Your Website, and as an undergraduate creating something this good, this fast proves Cowen's point about finding new ways to spend time freed up by ChatGPT feels right, as does the chances of seeing a similar expert system on just about every website in the next two years.
Marginal Revolution GMU professor Tyler Cowen's blog.
GPT and my own career trajectory
ChatGPT Can Give You Superpowers…IF You Use It Wisely by Mark McNeilly on substack