What a pleasure to read. Just last night I was with our attorneys discussing Cal's work and your sports article, so both of your names were dinner conversation.
Back to this post: I agree with much of this, especially the warning about protecting effort. Where I would push it even further is on the assumption that the primary danger is that AI makes us average. The deeper risk is that it can make us increasingly derivative of ourselves, I'm afraid.
AI is very good at helping us extend whatever pattern we already bring to it. If you are curious, disciplined, well-read, and willing to resist its defaults, it can become an extraordinary cognitive instrument. If you use it primarily to avoid uncertainty, difficulty, or authorship, it can become an extremely sophisticated avoidance mechanism.
As such, the problem begins when we stop noticing which parts of the struggle were actually doing developmental work on us. Writing is an obvious example when, for instance, a sentence you cannot write is telling you that you do not yet know what you think. Asking AI to solve that sentence may produce better prose while allowing you to bypass the encounter that would have produced a better mind, and the same applies far beyond writing.
I have spent much of my life deliberately entering environments where efficiency is almost irrelevant. Mountaineering, endurance training, psychology, serial entrepreneurship, and even relationships (ha!). The friction in all of them reveals the person who is meeting them.
I want to know and get much better at what deserves amplification and what must remain mine. I am much less interested in being “pro-AI” or joining an AI resistance. I use AI to remove drudgery, interrogate assumptions, expose blind spots, organize complexity, test ideas, and expand the range of what I can examine. I am very careful when it begins offering to do the becoming for me.
After years of using it, I am grateful that I can still recognize which thoughts, judgments, commitments, and choices are genuinely my own (and even better so).
Thank you, Ruben, for being here, for reading, and for sharing this thoughtful reflection!
"I have spent much of my life deliberately entering environments where efficiency is almost irrelevant. Mountaineering, endurance training, psychology, serial entrepreneurship, and even relationships (ha!). The friction in all of them reveals the person who is meeting them.
I want to know and get much better at what deserves amplification and what must remain mine. I am much less interested in being “pro-AI” or joining an AI resistance. I use AI to remove drudgery, interrogate assumptions, expose blind spots, organize complexity, test ideas, and expand the range of what I can examine. I am very careful when it begins offering to do the becoming for me."
Where you ultimately come out, what I'd call something like an "intentional AI user" or maybe even an "AI minimalist" (in contrast to many others, at least) is probably the best bet for all of us.
That’s the risk, surrendering to AI. The alternative is to exploit it with creative genius. The difficult bit - knowing which parts of the thinking you must never surrender
Nope. Though I did get a transcript from the software we used to record! But I probably spent 3 hours editing it down to this. What the transcript spit out wasn't very good. Which kind of gets to the point of the piece itself! I wouldn't want to hand transcribe the entire 75-minute conversation! But I am also not going to tell AI to "edit it to make it great for a reader" either. I want to do that work myself.
I noticed while playing with AI is that it never goes beyond average and if average is mediocre that sets the standard. It can be good on boring tasks like grammar etc In statistics what always amazed me is why on a graph that small dot is way outside the average and why average is seen as good and the dot an anomaly. Besides, i cought AI inventing research, quotes, and more. I believe all this has deep implications in terms what it means individualization of the human spirit, most people are second hand, speak whats suposed to be said and not actually have an opinion or critical thinking about a theme. I alwayd feel amazed when discussing some theme people start regurgitating common sense and then suddenly stop when they hit a wall of not knowing. Not knowing is great, it stirs curiosity. But most people feel lost when not knowing or seeing they really dont know much. Its humbling but in my xperience people dont like to feel they dont have control. Its more easy to believe in a lie, its comforting. This kind of normalcy is what i see in AI. More of the same, disguised as a new argument. In the end its about being comfort with the unknown and most humans feel panic with the unknown. The unknown is full of possibilities. The known is ashes most of the time. Life requires starting over every single moment.
Awesome. Yeah, I've fallen for the exact cognitive replication trap myself. I started using it to outsource even which movies I go see, which podcast episodes to listen to (dropped in the transcript), and which books to read. Then I would go do those, but I lost having my own opinion for a time. It was pretty scary, honestly.
Thanks, as someone who quit AI about 6 months ago and also just uses it as advanced Google, I appreciate some of these ideas and framework to talk about this.
My AI test went a lot further (about 2 years) and has landed me at a place where I think LLMs shouldn't be used in knowledge work at all.
Thoughts as to why in no particular order:
- The free and $20 models are very different from the business and enterprise models. The low end ones have the energy of the world's worst intern. The pricier ones don't have the grating tone, the inability to learn what you want, the formulaic writing (as obviously). It's a very slippery slope once you are using a better product, from research/late-stage checks to getting it to do the task for you and I think it's very addictive.
- I would go as far as to say I went from someone very cynical about AI to mild AI 'psychosis', for want to a better word, in a scarily short time.
- I'm good at my job because 20 years ago, I was working in customer service, writing repetitive emails, formatting reports, cleaning spreadsheets, doing data entry. Hated all those tasks and had no way if knowing they would led to the unique, marketable skill set I have now. It's not just bad for writing, it's bad for literally all workplace skill development - spreadsheets and lines of data too.
- It is leading to interpersonal workplace dynamics that are awful. People who lack self confidence spiral. People who are lazy bullshit harder. It brings out the worst in people.
- It breaks trust and moral. As cringey as team building is, it exists for a reason and can't be done when colleagues communicate through their bot intermediaries, even for little tasks.
- The outputs and tone reflect the Silicon Valley data fed into it, so LLMs replicate those processes and work culture.
- As well as being bad for the individual's skills, it's bad for the company. I'm a freelancer and see the same outputs and same turns of phrase popping up in documents produced by different clients, different teams, different companies, different countries. Been a while since I took an Econ class, but that doesn't seem like the road to our upcoming economic utopia.
Thanks again, good to discover your writing via the pod.
I never would have written two books (one in finance and the other on leadership and self-improvement) if I hadn’t read Deep Work—a game-changer. On AI, I finished writing The Psychology of Leadership right before ChatGPT came out. I used Grammarly for a final round of copyediting before submitting to my editor. Brad, that might be a good solution if you just want minor edits. Look…writing isn’t my day job. I use AI as an editor. It gives me confidence and speed. I don’t allow AI to use its annoying tics like “it’s not X, it’s Y”, overuse of rule of three, the short first sentence on every paragraph, often with a colon, etc. I always write a full first draft myself. I keep human writing (as measured by Pangram) above 95%.
AI is not a race to the bottom but a new race to the top.
I agree with the observations in this post, however, I believe they are symptoms of a) "incorrect usage of AI" which are due to it being new, and b) we (all) will get used to using AI as a tool in the right way. In the hands of "the right kind of users", it helps increase quality a million fold. People who are putting in the struggle and the effort are able to achieve unbelievable things.
Competitiveness, and the need to remain excellent and better than others, doesn't go away just because of AI.
Calculators or spreadsheets might have made us more lazy to do mental arithmetic, but that's fine. We don't live in a world anymore where we need to compete on that. We compete with complex spreadsheet models (among others).
And we'll express our humanity in new ways, creative people can now do crazy amazing things.
Manual writing might go the way of horseriding: much less popular and widespread, those who do it are really into it, and everyone else that's using a car is able to go much further much faster. Right now we're still in the transition phase where we can say that it feels weird to "ride" in a car, but we'll get used to "driving".
HI Brad,
What a pleasure to read. Just last night I was with our attorneys discussing Cal's work and your sports article, so both of your names were dinner conversation.
Back to this post: I agree with much of this, especially the warning about protecting effort. Where I would push it even further is on the assumption that the primary danger is that AI makes us average. The deeper risk is that it can make us increasingly derivative of ourselves, I'm afraid.
AI is very good at helping us extend whatever pattern we already bring to it. If you are curious, disciplined, well-read, and willing to resist its defaults, it can become an extraordinary cognitive instrument. If you use it primarily to avoid uncertainty, difficulty, or authorship, it can become an extremely sophisticated avoidance mechanism.
As such, the problem begins when we stop noticing which parts of the struggle were actually doing developmental work on us. Writing is an obvious example when, for instance, a sentence you cannot write is telling you that you do not yet know what you think. Asking AI to solve that sentence may produce better prose while allowing you to bypass the encounter that would have produced a better mind, and the same applies far beyond writing.
I have spent much of my life deliberately entering environments where efficiency is almost irrelevant. Mountaineering, endurance training, psychology, serial entrepreneurship, and even relationships (ha!). The friction in all of them reveals the person who is meeting them.
I want to know and get much better at what deserves amplification and what must remain mine. I am much less interested in being “pro-AI” or joining an AI resistance. I use AI to remove drudgery, interrogate assumptions, expose blind spots, organize complexity, test ideas, and expand the range of what I can examine. I am very careful when it begins offering to do the becoming for me.
After years of using it, I am grateful that I can still recognize which thoughts, judgments, commitments, and choices are genuinely my own (and even better so).
Thank you for a great piece. I'm a fan.
Ruben
Thank you, Ruben, for being here, for reading, and for sharing this thoughtful reflection!
"I have spent much of my life deliberately entering environments where efficiency is almost irrelevant. Mountaineering, endurance training, psychology, serial entrepreneurship, and even relationships (ha!). The friction in all of them reveals the person who is meeting them.
I want to know and get much better at what deserves amplification and what must remain mine. I am much less interested in being “pro-AI” or joining an AI resistance. I use AI to remove drudgery, interrogate assumptions, expose blind spots, organize complexity, test ideas, and expand the range of what I can examine. I am very careful when it begins offering to do the becoming for me."
Where you ultimately come out, what I'd call something like an "intentional AI user" or maybe even an "AI minimalist" (in contrast to many others, at least) is probably the best bet for all of us.
@Alex Roddie
Thanks for sharing the piece!
Cheers Brad, I enjoy your and Cal’s work!
Oooh cheers for the tag
That’s the risk, surrendering to AI. The alternative is to exploit it with creative genius. The difficult bit - knowing which parts of the thinking you must never surrender
The future for all of us, probably.
I hope this summary of the video is not writtern by an AI
Nope. Though I did get a transcript from the software we used to record! But I probably spent 3 hours editing it down to this. What the transcript spit out wasn't very good. Which kind of gets to the point of the piece itself! I wouldn't want to hand transcribe the entire 75-minute conversation! But I am also not going to tell AI to "edit it to make it great for a reader" either. I want to do that work myself.
Great stuff.
Thank you for the kind words and being here.
I noticed while playing with AI is that it never goes beyond average and if average is mediocre that sets the standard. It can be good on boring tasks like grammar etc In statistics what always amazed me is why on a graph that small dot is way outside the average and why average is seen as good and the dot an anomaly. Besides, i cought AI inventing research, quotes, and more. I believe all this has deep implications in terms what it means individualization of the human spirit, most people are second hand, speak whats suposed to be said and not actually have an opinion or critical thinking about a theme. I alwayd feel amazed when discussing some theme people start regurgitating common sense and then suddenly stop when they hit a wall of not knowing. Not knowing is great, it stirs curiosity. But most people feel lost when not knowing or seeing they really dont know much. Its humbling but in my xperience people dont like to feel they dont have control. Its more easy to believe in a lie, its comforting. This kind of normalcy is what i see in AI. More of the same, disguised as a new argument. In the end its about being comfort with the unknown and most humans feel panic with the unknown. The unknown is full of possibilities. The known is ashes most of the time. Life requires starting over every single moment.
"Not knowing is great, it stirs curiosity." Love this.
Awesome. Yeah, I've fallen for the exact cognitive replication trap myself. I started using it to outsource even which movies I go see, which podcast episodes to listen to (dropped in the transcript), and which books to read. Then I would go do those, but I lost having my own opinion for a time. It was pretty scary, honestly.
I've had that experience too. It's a very easy road to go down, makes me wonder how many people that happens to.
Thanks, as someone who quit AI about 6 months ago and also just uses it as advanced Google, I appreciate some of these ideas and framework to talk about this.
My AI test went a lot further (about 2 years) and has landed me at a place where I think LLMs shouldn't be used in knowledge work at all.
Thoughts as to why in no particular order:
- The free and $20 models are very different from the business and enterprise models. The low end ones have the energy of the world's worst intern. The pricier ones don't have the grating tone, the inability to learn what you want, the formulaic writing (as obviously). It's a very slippery slope once you are using a better product, from research/late-stage checks to getting it to do the task for you and I think it's very addictive.
- I would go as far as to say I went from someone very cynical about AI to mild AI 'psychosis', for want to a better word, in a scarily short time.
- I'm good at my job because 20 years ago, I was working in customer service, writing repetitive emails, formatting reports, cleaning spreadsheets, doing data entry. Hated all those tasks and had no way if knowing they would led to the unique, marketable skill set I have now. It's not just bad for writing, it's bad for literally all workplace skill development - spreadsheets and lines of data too.
- It is leading to interpersonal workplace dynamics that are awful. People who lack self confidence spiral. People who are lazy bullshit harder. It brings out the worst in people.
- It breaks trust and moral. As cringey as team building is, it exists for a reason and can't be done when colleagues communicate through their bot intermediaries, even for little tasks.
- The outputs and tone reflect the Silicon Valley data fed into it, so LLMs replicate those processes and work culture.
- As well as being bad for the individual's skills, it's bad for the company. I'm a freelancer and see the same outputs and same turns of phrase popping up in documents produced by different clients, different teams, different companies, different countries. Been a while since I took an Econ class, but that doesn't seem like the road to our upcoming economic utopia.
Thanks again, good to discover your writing via the pod.
I never would have written two books (one in finance and the other on leadership and self-improvement) if I hadn’t read Deep Work—a game-changer. On AI, I finished writing The Psychology of Leadership right before ChatGPT came out. I used Grammarly for a final round of copyediting before submitting to my editor. Brad, that might be a good solution if you just want minor edits. Look…writing isn’t my day job. I use AI as an editor. It gives me confidence and speed. I don’t allow AI to use its annoying tics like “it’s not X, it’s Y”, overuse of rule of three, the short first sentence on every paragraph, often with a colon, etc. I always write a full first draft myself. I keep human writing (as measured by Pangram) above 95%.
AI is not a race to the bottom but a new race to the top.
I agree with the observations in this post, however, I believe they are symptoms of a) "incorrect usage of AI" which are due to it being new, and b) we (all) will get used to using AI as a tool in the right way. In the hands of "the right kind of users", it helps increase quality a million fold. People who are putting in the struggle and the effort are able to achieve unbelievable things.
Competitiveness, and the need to remain excellent and better than others, doesn't go away just because of AI.
Calculators or spreadsheets might have made us more lazy to do mental arithmetic, but that's fine. We don't live in a world anymore where we need to compete on that. We compete with complex spreadsheet models (among others).
And we'll express our humanity in new ways, creative people can now do crazy amazing things.
Manual writing might go the way of horseriding: much less popular and widespread, those who do it are really into it, and everyone else that's using a car is able to go much further much faster. Right now we're still in the transition phase where we can say that it feels weird to "ride" in a car, but we'll get used to "driving".