Hi Mike. I like how you walked us through your train of thought and then stopped- meaning that you didn’t feel compelled to offer up an idea in a nice tidy package.
Thanks for the shoutout in this piece. As I wrote in Trust is the New Oil, I think about email as an analogy:
As an analogy, think about what’s happened to email over the last 20-30 years. In the late 1990s-early 2000s, it was still relatively novel to receive email and it was therefore high signal. Today, the noise has drowned out signal. We are overwhelmed by email spam, with spammers and email filters engaged in an never-ending cat-and-mouse game. Those filters are pretty good, but the costs to send bulk mail are so low that the spammers still have an economic incentive to try. Even with the best filters, spam slips through and, just as bad, legitimate emails get filtered out. (I recently found a trove of responses to the emails I send out from this Substack buried in my spam folder.)
The result is that email has been devalued as a communications medium. Click-through rates on legitimate emails have declined from 4-6% in 2010 to about 2% today. While it is still used at work and for a lot of marketing, its prior role has splintered into numerous apps: Discord, Insta, group chats, Slack, etc.
So, under this theory, open platforms like social collapse under the weight of all this stuff. Still used, but their utility will diminish. So, what’s the Discord/Telegram/Slack of content in the future?
I compare AI content to traditional content the same way I think about buying mass-produced cookies versus going to a local bakery. Sometimes I don’t care much about the end result, and a $3 box of Chips Ahoy fills the void. Other times I want something interesting, spectacular, or memorable—so I’ll spend $3 on a single bakery cookie.
Sometimes we’ll care if content is AI, and sometimes we won’t.
It’s like within the membrane between attention and content these massive models of reality are being formed—shaping a larger consciousness, and we can sense a handing-off of agency. But then maybe we never had as much agency as we thought we had anyway.
I look at AI the way I look at all man-made things - if I can't see truth or goodness within the thing itself or in the intentions of its creator, then the primary motive for the thing to exist is likely of little or no value to my life. And in today's technological society, my adoption or acceptance of the thing is likely only going to result in money escaping me on a monthly basis or until designed obsolescence requires me to replace the thing with a newer version of itself, on and on, until I lose the ability to do that which the thing was created to replace in the first place.
Mike I want to add this post is excellent. Also I may have misinterpreted you. I think you meant "Computers can read and write as well as people." in the context of posting and generating online activity, which they indeed can.
What’s a person if not language? God? Breath and pulse? The homeostasis that exists until he needs to speak again? The second a person is recognized at all his best description is the language he uses and the language used to describe him. “Marriage of Many Years” says “you are a language I have learned by heart.” In the absence of language the gates on insanity are wide open.
When a critical mass turns apathetic toward colloquial language and colloquialisms, language loses meaning and/or converges on newspeak & doublethink (1984) — it makes one wonder if the ultimate ability for anyone creating media (AI assisted or not) is simply that of persuading toward a language-bound-goal the audience of oneself, family, friend group, community, cult, political party, or nation? In the case of a solitary thinker, the to and fro converges on schizophrenia. As an aside I am also intrigued by how something like the diagnostic and statistical manual of mental health disorders will evolve to respond to the AI sphere.
Pewdiepie released a video “Stop. Using AI” in which he displays something similar to reaction agents. His chatbots are organized as members of a council that deliberates on his queries and then responds unanimously with an answer. There is a slight twist though; the council members die if they are useless and that made them resort to lies and manipulation to retain their position which ultimately reduces the trustworthiness of the responses.
Seems to me even the bounds of trustworthiness are political and monetary—both pinned by language unless we fall back to protohuman grunty sounds in defense or against physical behaviors of only those in our immediate proximity.
I guess at that point trust comes down to being brothers and that’s about it? Or being able to absurdly allow anything at all from a place of complete indifference toward the quality of abstractions used to frame it? People typically operate in alignment with the former when money is involved and the latter when they seek attention.
What is the pursuit of attention anyway if not a wish to be perpetually cast with favorable and trendy adjectives via negative feedback loops? I won’t take away all credit from everyone but I suspect the ratio of people who have a plan for what they’d do with fame and money is smaller than those who simply chase it under a spell of hysteria. I really want to say it’s something like 1:99 but I’ll be happy to posit a 49:51 ratio for now. And maybe as the money and fame flows in more and more the recipients tendencies start bordering on fascism? Wherever there’s attention, a dictator will appear? The poor thing is often just a lived experiment in the arbitrary divisions of an ever-fragmenting dictionary though.
—
As much as this reads like a treatise defending humans being language I am absolutely heartbroken by it. I want to design something more primitive that allows less noise.
I like the topics Mike choses to write about. I also think he’s better exposed and informed than I am. I write here sometimes almost as though it’s a conversation which stands to contribute to his and my own inquiry.
This particular instance was an accidental comment. I wrote what I wrote and I would have likely deleted or moved it to my own notes for further scrutiny but given the weirdly placed left side publish button for comments in this app it got posted while I tried to edit and I ultimately opted to leave it.
As for not publishing my writing: I’ve had a relatively turbulent life and within it I’ve been unable to set a publishing system. I also feel estranged from the two potential audiences I once thought I’d have. My best hope as a writer now is to write for the mentally challenged and I haven’t been able to recompose much to that end since I’ve come to terms with the realization. Money complicates things further and I wonder if I should just operate according to what I write instead of publish?
I take your comment as equal parts compliment and etiquette check. For the former, I am grateful, and the latter, I agree—I didn’t mean to bother.
Very thoughtful stuff Mike. I think you are on to something with the trust idea - I think it is like going back in time a bit - there was a time when a person's reputation was totally critical, which is why people fought duels over them. I am not sure I would like to be you and Ian's age - the pace of change is creating huge opportunities and huge potential problems. Keep thinking and writing.
"Computers can read and write as well as people." They can't produce original thinking. AI data mines and repeats. If that's your definition of writing, then yes, AI can write.
Writing and thinking would normally be thought of as synonymous. One presupposes the other. Otherwise you just have copying, which yes, AI is very good at.
That's exactly the presumption I'm curious to pull apart. Many people presume writing = thinking. When computers can write so well (and not just copying, they can create net new sentences), it begs the question what is writing vs thinking. It's a much more interesting question than "is AI sentient" which clearly it is not.
AI can recombine an existing sentence, or duplicate one from a data-scraped source. These are not "new sentences", unless you consider new to be sentences that are recombined or duplicated. People think and write. Machines aggregate and repeat. So again, AI can "write", but only in that sense. It can't "think".
Besides parsing and interesting discussions on definitions, which I'm always up for, there's some real danger here to all of us writers. I wrote about it here, if I may https://richarddonnelly.substack.com/p/whats-new-in-ai
Hi Mike. I like how you walked us through your train of thought and then stopped- meaning that you didn’t feel compelled to offer up an idea in a nice tidy package.
Thanks for the shoutout in this piece. As I wrote in Trust is the New Oil, I think about email as an analogy:
As an analogy, think about what’s happened to email over the last 20-30 years. In the late 1990s-early 2000s, it was still relatively novel to receive email and it was therefore high signal. Today, the noise has drowned out signal. We are overwhelmed by email spam, with spammers and email filters engaged in an never-ending cat-and-mouse game. Those filters are pretty good, but the costs to send bulk mail are so low that the spammers still have an economic incentive to try. Even with the best filters, spam slips through and, just as bad, legitimate emails get filtered out. (I recently found a trove of responses to the emails I send out from this Substack buried in my spam folder.)
The result is that email has been devalued as a communications medium. Click-through rates on legitimate emails have declined from 4-6% in 2010 to about 2% today. While it is still used at work and for a lot of marketing, its prior role has splintered into numerous apps: Discord, Insta, group chats, Slack, etc.
So, under this theory, open platforms like social collapse under the weight of all this stuff. Still used, but their utility will diminish. So, what’s the Discord/Telegram/Slack of content in the future?
Fantastic! 'Language is not me' is a key insight.
I compare AI content to traditional content the same way I think about buying mass-produced cookies versus going to a local bakery. Sometimes I don’t care much about the end result, and a $3 box of Chips Ahoy fills the void. Other times I want something interesting, spectacular, or memorable—so I’ll spend $3 on a single bakery cookie.
Sometimes we’ll care if content is AI, and sometimes we won’t.
It’s like within the membrane between attention and content these massive models of reality are being formed—shaping a larger consciousness, and we can sense a handing-off of agency. But then maybe we never had as much agency as we thought we had anyway.
I look at AI the way I look at all man-made things - if I can't see truth or goodness within the thing itself or in the intentions of its creator, then the primary motive for the thing to exist is likely of little or no value to my life. And in today's technological society, my adoption or acceptance of the thing is likely only going to result in money escaping me on a monthly basis or until designed obsolescence requires me to replace the thing with a newer version of itself, on and on, until I lose the ability to do that which the thing was created to replace in the first place.
Thank you, Mike, for sharing this phenomenal analysis.
If I understand correctly:
• You see AI as a supplement to human creativity, not a replacement.
• The era in which unfocused AI content garners attention will prove transitory.
Mike I want to add this post is excellent. Also I may have misinterpreted you. I think you meant "Computers can read and write as well as people." in the context of posting and generating online activity, which they indeed can.
What’s a person if not language? God? Breath and pulse? The homeostasis that exists until he needs to speak again? The second a person is recognized at all his best description is the language he uses and the language used to describe him. “Marriage of Many Years” says “you are a language I have learned by heart.” In the absence of language the gates on insanity are wide open.
When a critical mass turns apathetic toward colloquial language and colloquialisms, language loses meaning and/or converges on newspeak & doublethink (1984) — it makes one wonder if the ultimate ability for anyone creating media (AI assisted or not) is simply that of persuading toward a language-bound-goal the audience of oneself, family, friend group, community, cult, political party, or nation? In the case of a solitary thinker, the to and fro converges on schizophrenia. As an aside I am also intrigued by how something like the diagnostic and statistical manual of mental health disorders will evolve to respond to the AI sphere.
Pewdiepie released a video “Stop. Using AI” in which he displays something similar to reaction agents. His chatbots are organized as members of a council that deliberates on his queries and then responds unanimously with an answer. There is a slight twist though; the council members die if they are useless and that made them resort to lies and manipulation to retain their position which ultimately reduces the trustworthiness of the responses.
Seems to me even the bounds of trustworthiness are political and monetary—both pinned by language unless we fall back to protohuman grunty sounds in defense or against physical behaviors of only those in our immediate proximity.
I guess at that point trust comes down to being brothers and that’s about it? Or being able to absurdly allow anything at all from a place of complete indifference toward the quality of abstractions used to frame it? People typically operate in alignment with the former when money is involved and the latter when they seek attention.
What is the pursuit of attention anyway if not a wish to be perpetually cast with favorable and trendy adjectives via negative feedback loops? I won’t take away all credit from everyone but I suspect the ratio of people who have a plan for what they’d do with fame and money is smaller than those who simply chase it under a spell of hysteria. I really want to say it’s something like 1:99 but I’ll be happy to posit a 49:51 ratio for now. And maybe as the money and fame flows in more and more the recipients tendencies start bordering on fascism? Wherever there’s attention, a dictator will appear? The poor thing is often just a lived experiment in the arbitrary divisions of an ever-fragmenting dictionary though.
—
As much as this reads like a treatise defending humans being language I am absolutely heartbroken by it. I want to design something more primitive that allows less noise.
Why aren’t you publishing pieces yourself? Given the expended effort above, sure seems like the upside outweighs the downside.
I like the topics Mike choses to write about. I also think he’s better exposed and informed than I am. I write here sometimes almost as though it’s a conversation which stands to contribute to his and my own inquiry.
This particular instance was an accidental comment. I wrote what I wrote and I would have likely deleted or moved it to my own notes for further scrutiny but given the weirdly placed left side publish button for comments in this app it got posted while I tried to edit and I ultimately opted to leave it.
As for not publishing my writing: I’ve had a relatively turbulent life and within it I’ve been unable to set a publishing system. I also feel estranged from the two potential audiences I once thought I’d have. My best hope as a writer now is to write for the mentally challenged and I haven’t been able to recompose much to that end since I’ve come to terms with the realization. Money complicates things further and I wonder if I should just operate according to what I write instead of publish?
I take your comment as equal parts compliment and etiquette check. For the former, I am grateful, and the latter, I agree—I didn’t mean to bother.
Very thoughtful stuff Mike. I think you are on to something with the trust idea - I think it is like going back in time a bit - there was a time when a person's reputation was totally critical, which is why people fought duels over them. I am not sure I would like to be you and Ian's age - the pace of change is creating huge opportunities and huge potential problems. Keep thinking and writing.
"Computers can read and write as well as people." They can't produce original thinking. AI data mines and repeats. If that's your definition of writing, then yes, AI can write.
You'll notice I didn't say "think". This was a deliberate choice of the author :)
Writing and thinking would normally be thought of as synonymous. One presupposes the other. Otherwise you just have copying, which yes, AI is very good at.
That's exactly the presumption I'm curious to pull apart. Many people presume writing = thinking. When computers can write so well (and not just copying, they can create net new sentences), it begs the question what is writing vs thinking. It's a much more interesting question than "is AI sentient" which clearly it is not.
AI can recombine an existing sentence, or duplicate one from a data-scraped source. These are not "new sentences", unless you consider new to be sentences that are recombined or duplicated. People think and write. Machines aggregate and repeat. So again, AI can "write", but only in that sense. It can't "think".
Besides parsing and interesting discussions on definitions, which I'm always up for, there's some real danger here to all of us writers. I wrote about it here, if I may https://richarddonnelly.substack.com/p/whats-new-in-ai