> Except questions/comments in this form preclude the possibility of human connection. (Something to think about for later.)
I think that interacting with your bot is a gradual on-ramp to interacting with you or with other humans on the topic. Even if the bot says something offensive, it is an opportunity for me to share this with you. (And if it a positive interaction, then I am also encouraged) Basically, interaction with the bot is a safe way to say something without being judged by a human celebrity and is the first step towards a connection for which I may now have had enough activation energy otherwise. So, I believe your bot will increase human connections.
I like your hopeful thinking. It's something to pay attention to. I can imagine interactions that would increase connection--People Asking Similar Questions, or Groups Like This or some such.
P.S. I've just in the last few weeks "jumped into AI" with both feet. I started with an iPhone app called "ChatBox," which worked amazing well. After a lot of tinkering, I eventually became reasonably proficient at interacting with it, i.e., "Conversing" with it, if you will. The peak of my progress so far has been to learn how to phrase my questions in such a way that the bot it will be compelled to produce it's "answers" in the form of numbered LISTS. Huge win! "Here we GO !!!!!! ..." THANKS! -gs
interesting. It seems like that is the case with any tool, past and present. You start expecting to use a tool in certain way but if designer of the tool intends it to be used in another way (may be the opposite of affordance in some ways), then the tool can change the way you use it. I see that with certain programming languages as well.
These seem like a set of points together pose a significant gain through "automation by AI"
* don't have to wait for a reply.
* preclude (or at least ... filter) interaction
* don't be an *ss
I think the AI accomplishes automating the hardest task at all, esp. an author/blogger at your level/circle of reference : there are 100's (1000's?!) of potential comments/references(link-backs?) that fall under the TLDR or missed-concept category. The AI in this case will (hopefully) do a great job of pointing out the didnt-read or re-framing the facts to help understanding of the concept.
The AI (hopefully) would also do a good job of summarizing the comment/references into a few categories that would allow you to pick out the more engaging questions or comments.
I think this part defeats the "precludes human interaction" part. We're only human, and answering 1000 comments is not feasible.
I agree with your training sentiment, and I think the whole internet is doing this "experiment" live in real time, so we'll definitely see over the coming months/years how well AI advances toward accuracy and bibliography.
> Except questions/comments in this form preclude the possibility of human connection. (Something to think about for later.)
I think that interacting with your bot is a gradual on-ramp to interacting with you or with other humans on the topic. Even if the bot says something offensive, it is an opportunity for me to share this with you. (And if it a positive interaction, then I am also encouraged) Basically, interaction with the bot is a safe way to say something without being judged by a human celebrity and is the first step towards a connection for which I may now have had enough activation energy otherwise. So, I believe your bot will increase human connections.
I like your hopeful thinking. It's something to pay attention to. I can imagine interactions that would increase connection--People Asking Similar Questions, or Groups Like This or some such.
Wow! Both your work so far and your plans for the future are SERIOUSLY impressive!
"Oh fellow wizard, I salute you!"
-gs (G. S. Khalsa, kulsayogi@gmail.com, 510-575-6695)
P.S. I've just in the last few weeks "jumped into AI" with both feet. I started with an iPhone app called "ChatBox," which worked amazing well. After a lot of tinkering, I eventually became reasonably proficient at interacting with it, i.e., "Conversing" with it, if you will. The peak of my progress so far has been to learn how to phrase my questions in such a way that the bot it will be compelled to produce it's "answers" in the form of numbered LISTS. Huge win! "Here we GO !!!!!! ..." THANKS! -gs
I'm seriously concerned that the net effect of chat bots will be to train us to speak differently. I wanted my computer to adapt to me.
interesting. It seems like that is the case with any tool, past and present. You start expecting to use a tool in certain way but if designer of the tool intends it to be used in another way (may be the opposite of affordance in some ways), then the tool can change the way you use it. I see that with certain programming languages as well.
Tyler Cowen did something similar with his latest book: https://goatgreatesteconomistofalltime.ai/en Very cool!
I heard about that. I wasn't consciously copying him but it was definitely in there.
TLDR;
OTOH, I have thoughts, at the conceptual level.
These seem like a set of points together pose a significant gain through "automation by AI"
* don't have to wait for a reply.
* preclude (or at least ... filter) interaction
* don't be an *ss
I think the AI accomplishes automating the hardest task at all, esp. an author/blogger at your level/circle of reference : there are 100's (1000's?!) of potential comments/references(link-backs?) that fall under the TLDR or missed-concept category. The AI in this case will (hopefully) do a great job of pointing out the didnt-read or re-framing the facts to help understanding of the concept.
The AI (hopefully) would also do a good job of summarizing the comment/references into a few categories that would allow you to pick out the more engaging questions or comments.
I think this part defeats the "precludes human interaction" part. We're only human, and answering 1000 comments is not feasible.
I agree with your training sentiment, and I think the whole internet is doing this "experiment" live in real time, so we'll definitely see over the coming months/years how well AI advances toward accuracy and bibliography.
Thanks for your post.
hi,
this sound pretty interesting to me.
i really appreciate your thinking and approach to software development and business, so i am really eager to try it out.
in the past i felt always depressed about the answers of chatgpt, because they are very mediocre.
so there is light for my use of LLMs now.
Amazing!
These are Heady Times... Let's all hope for the best -🙏