I started my own business when I left tech. It's a bar.
I do my own mechanic work, but I have old trucks. I would never sign up to work on modern vehicles and the entitled customers trying to get one over on you.
I drove a route (not delivery) many years ago and I liked that. I tend to like jobs where you start with a pile of work, and finish with none left, rather than perpetually working towards something that's never done, like maybe a software project.
Delivery drivers and train conductors are both headed for automation. I think trains are mostly already there. They mostly have a human for emergencies, as far as I know.
This is something I’ve been thinking about, but nothing speaks to me enough to jump ship. Anything I think of also means taking a huge pay cut, so it seems to make sense to stay on the current track until the wheels fall off to maximize my savings to soften the transition. It’s also possible that I may never need to switch careers before retirement, so it could be a moot point.
I think people would do best to retool away from “coder” towards some specilaization where coding is part of the job. Don’t be a coder, be an X that codes.
If we think of coding becoming like writing, then you can see how it might be a ubiquitous skill - but still some differentiation. Some small number of highly skilled people do it full time. Some people do is as a big part of their job (journalists, etc). Still others it’s not core to the job, but it helps (a lot of white collar work).
I have felt that shift too. One thing that helped me was focusing on skills where you build and ship something yourself — even small tools or utilities. That sense of ownership and direct usefulness is hard to replace and still very satisfying.
I keep a short list of jobs where the output is visible by the end of the day: electrician, field technician, rail operations, even property maintenance. Shadowing someone for one real shift helped me separate fantasy from fit much faster than more online reading.
If you are an enterprise developer working at a bank, insurance company, airline etc, you have been easily replaceable commodity for well over a decade and comp has stagnated for a decade. Those types of companies are reaching out to me today offering the same comp for the same type of role I was getting in 2016-2018.
If someone hoped to make a career by turning well defined requirements into code forever, the writing was on the wall when it was easy enough for anyone to become a good enough full stack developer/mobile developer/web developer etc.
You are now seeing the same in BigTech where knowing how to reverse a b tree isn’t good enough to get that $225K+ a year mid level job.
AI has just accelerated that trend. The transition I started making a decade ago was not to be just an undifferentiated ticket taker and get closer to dealing with the “business” and taking on projects with larger scope, impact and ambiguity - a true senior developer.
Even though part of my job has been to produce code for 30 years, it’s been well over a decade since I had to interview based on coding tests. I’ve had 7 jobs since then including one at $BigTech.
AI can’t talk to customers. It can’t solve XYProblems. It can’t navigate corporate politics and conflicting priorities.
In my 25+ years programming, I don't yet think I've met a programmer who hasn't dreamed of "just being a lawnmower" (or some equally equiv, "mundane" job). Just the idea of ending the day having accomplished something and being able to 100% switch off work and just relax. The same is probably true of other professions like lawyers and doctors.
AI is definitely going to change things. It already has, but I don't think what it's done is the real change coming. Right now it's just a new tool in the toolbox. It does great at some things and terrible at others. Note: I'm speaking purely about LLMs/generative AI, because other forms of AI are -incredibly- useful and have been for a while (think DeepFold). I only mention it because it's important people remember there's different types of AI models used for all sorts of different things.
What's important to keep in mind are a few things:
* We all stand on the shoulders (abstractions) of giants. People of the past lamented about C, Java, ... every abstraction that came before. Did those abstractions cause the next batch of programmers to lose knowledge about what was really happening? Sure, but each also enabled an explosion of new ways of thinking and problem solving that brought us to where we are now. If you can recognize that, then that will help.
* You became a programmer/engineer for reasons (what I'll call "itches"). Only you know what those are. Buy, my guess is that - like most of us - there was something about the -problem solving- and logical analyses/thinking taking place, and seeing others gaining benefit from that work. First, if you look to another job, pick something that scratches those same itches, otherwise you'll be bored and depressed again very quickly. More importantly, the reason you're even asking the question is because something about AI feels like it's taken that "scratch" away from you. Keep in mind that your "itch" can be abstracted, too. Instead of solving a (for example) network byzantine problem, go one level higher, which has the same problem at the human-scale. Now AI becomes a tool to help you come up with and solve the problem instead of one that's replacing you. Same itch, different level, different tools.
* The comment mentioning "shadowing someone" is some solid advice. Right now you're succumbing to a "grass is greener" effect. That will disappear quickly if you do switch, and you'll still be in the same place if those itches are not being scratched. Again, identify the itch first.
Finally... and I can't stress this one enough... whenever I've been down due to the "itch" not being scratched at work (for any reason, not just AI), I've personally found the best solution to the problem is to volunteer and help others. This could be anything: food bank, big brother, translation services, helping to teach your native language to someone else, or even teaching something you consider important (i.e., programming) to kids at your local school via an after-school program. Every single time I do this I walk away feeling great. Inevitably there was a single point in the day in which I dramatically affected one person's life positively, and they - in turn - affected me similarly. In a class of 10 kids who want to learn to make games, there was 1 who had no way of being able to do this on their own due to at-home problems. At a food bank it was an elderly woman who couldn't speak English and I happened to be the one person who could speak her language (she was so excited to talk to someone). Or being in the children's ward at the hospital over x-mas, delivering toys. And sometimes I'd even walk away with a new set of ideas where my programming skills could solve a new set of problems I discovered existed and lead me to a whole new career.
The only thing I desired to do full time outside of programming was real estate investing and teaching fitness classes personal training.
I was a part time fitness instructor from 1999-2012 and was a (very bad) real estate investor from 2002-2008. I wouldn’t say fitness instructor is just for the young, my wife is now 50 and she has been teaching for 5 years. But it’s definitely not something you can do full time, the money isn’t there and trying to do it too much and is a wear and tear on your body. Even owning your studio isn’t that great as far as income.
But now at 51 and I have been working remotely between three companies since 2020, there is nothing I would rather do than wake up, wash up, and walk to my office in the other bedroom.
We have also done the digital nomad thing for a year and this year we will be away from home for a total of 3-4 months. There is nothing else that I would rather be doing.
AI interaction is fundamentally unsatisfying. (I have a theory as to why; see below.) An AI can get you results, maybe even the results you wanted, but it can't give you a genuine smile. Even when we're surrounded by things that get us results, we still want the smile. So if you're doing something where you interact with people face to face, at least some of the time, that's a decent place.
In my large town/small city, there's an ice cream shop that closes between Thanksgiving and roughly March 1st. I would love to own a business that people cared as much about as they care about that place opening for the year.
-----
Why is interacting with an AI unsatisfying? I wonder if it doesn't come down to information theory. AI output has too little information in it.
You compute the information in a text by going symbol by symbol. Normally "symbol" means one character or byte. But you could do it by going word by word (or word part by word part) instead. Here you can see the problem: an AI gives you whatever word is the most probable next word, given the context. That's lower information than human output, which would give a wider variety of next words, and therefore a higher (Shannon) information content.
Yes, you can change that by increasing the temperature of the AI. Can you do that enough to give human-equivalent information content without destroying the coherence of the output? (Merely high Shannon information isn't the only goal. The insane may have high information in their output, but they are incoherent enough that it doesn't matter.)
I started my own business when I left tech. It's a bar.
I do my own mechanic work, but I have old trucks. I would never sign up to work on modern vehicles and the entitled customers trying to get one over on you.
I drove a route (not delivery) many years ago and I liked that. I tend to like jobs where you start with a pile of work, and finish with none left, rather than perpetually working towards something that's never done, like maybe a software project.
Delivery drivers and train conductors are both headed for automation. I think trains are mostly already there. They mostly have a human for emergencies, as far as I know.
This is something I’ve been thinking about, but nothing speaks to me enough to jump ship. Anything I think of also means taking a huge pay cut, so it seems to make sense to stay on the current track until the wheels fall off to maximize my savings to soften the transition. It’s also possible that I may never need to switch careers before retirement, so it could be a moot point.
I think people would do best to retool away from “coder” towards some specilaization where coding is part of the job. Don’t be a coder, be an X that codes.
If we think of coding becoming like writing, then you can see how it might be a ubiquitous skill - but still some differentiation. Some small number of highly skilled people do it full time. Some people do is as a big part of their job (journalists, etc). Still others it’s not core to the job, but it helps (a lot of white collar work).
I have felt that shift too. One thing that helped me was focusing on skills where you build and ship something yourself — even small tools or utilities. That sense of ownership and direct usefulness is hard to replace and still very satisfying.
I keep a short list of jobs where the output is visible by the end of the day: electrician, field technician, rail operations, even property maintenance. Shadowing someone for one real shift helped me separate fantasy from fit much faster than more online reading.
If you are an enterprise developer working at a bank, insurance company, airline etc, you have been easily replaceable commodity for well over a decade and comp has stagnated for a decade. Those types of companies are reaching out to me today offering the same comp for the same type of role I was getting in 2016-2018.
If someone hoped to make a career by turning well defined requirements into code forever, the writing was on the wall when it was easy enough for anyone to become a good enough full stack developer/mobile developer/web developer etc.
You are now seeing the same in BigTech where knowing how to reverse a b tree isn’t good enough to get that $225K+ a year mid level job.
AI has just accelerated that trend. The transition I started making a decade ago was not to be just an undifferentiated ticket taker and get closer to dealing with the “business” and taking on projects with larger scope, impact and ambiguity - a true senior developer.
https://www.levels.fyi/blog/swe-level-framework.html
Even though part of my job has been to produce code for 30 years, it’s been well over a decade since I had to interview based on coding tests. I’ve had 7 jobs since then including one at $BigTech.
AI can’t talk to customers. It can’t solve XYProblems. It can’t navigate corporate politics and conflicting priorities.
In my 25+ years programming, I don't yet think I've met a programmer who hasn't dreamed of "just being a lawnmower" (or some equally equiv, "mundane" job). Just the idea of ending the day having accomplished something and being able to 100% switch off work and just relax. The same is probably true of other professions like lawyers and doctors.
AI is definitely going to change things. It already has, but I don't think what it's done is the real change coming. Right now it's just a new tool in the toolbox. It does great at some things and terrible at others. Note: I'm speaking purely about LLMs/generative AI, because other forms of AI are -incredibly- useful and have been for a while (think DeepFold). I only mention it because it's important people remember there's different types of AI models used for all sorts of different things.
What's important to keep in mind are a few things:
* We all stand on the shoulders (abstractions) of giants. People of the past lamented about C, Java, ... every abstraction that came before. Did those abstractions cause the next batch of programmers to lose knowledge about what was really happening? Sure, but each also enabled an explosion of new ways of thinking and problem solving that brought us to where we are now. If you can recognize that, then that will help.
* You became a programmer/engineer for reasons (what I'll call "itches"). Only you know what those are. Buy, my guess is that - like most of us - there was something about the -problem solving- and logical analyses/thinking taking place, and seeing others gaining benefit from that work. First, if you look to another job, pick something that scratches those same itches, otherwise you'll be bored and depressed again very quickly. More importantly, the reason you're even asking the question is because something about AI feels like it's taken that "scratch" away from you. Keep in mind that your "itch" can be abstracted, too. Instead of solving a (for example) network byzantine problem, go one level higher, which has the same problem at the human-scale. Now AI becomes a tool to help you come up with and solve the problem instead of one that's replacing you. Same itch, different level, different tools.
* The comment mentioning "shadowing someone" is some solid advice. Right now you're succumbing to a "grass is greener" effect. That will disappear quickly if you do switch, and you'll still be in the same place if those itches are not being scratched. Again, identify the itch first.
Finally... and I can't stress this one enough... whenever I've been down due to the "itch" not being scratched at work (for any reason, not just AI), I've personally found the best solution to the problem is to volunteer and help others. This could be anything: food bank, big brother, translation services, helping to teach your native language to someone else, or even teaching something you consider important (i.e., programming) to kids at your local school via an after-school program. Every single time I do this I walk away feeling great. Inevitably there was a single point in the day in which I dramatically affected one person's life positively, and they - in turn - affected me similarly. In a class of 10 kids who want to learn to make games, there was 1 who had no way of being able to do this on their own due to at-home problems. At a food bank it was an elderly woman who couldn't speak English and I happened to be the one person who could speak her language (she was so excited to talk to someone). Or being in the children's ward at the hospital over x-mas, delivering toys. And sometimes I'd even walk away with a new set of ideas where my programming skills could solve a new set of problems I discovered existed and lead me to a whole new career.
The only thing I desired to do full time outside of programming was real estate investing and teaching fitness classes personal training.
I was a part time fitness instructor from 1999-2012 and was a (very bad) real estate investor from 2002-2008. I wouldn’t say fitness instructor is just for the young, my wife is now 50 and she has been teaching for 5 years. But it’s definitely not something you can do full time, the money isn’t there and trying to do it too much and is a wear and tear on your body. Even owning your studio isn’t that great as far as income.
But now at 51 and I have been working remotely between three companies since 2020, there is nothing I would rather do than wake up, wash up, and walk to my office in the other bedroom.
We have also done the digital nomad thing for a year and this year we will be away from home for a total of 3-4 months. There is nothing else that I would rather be doing.
My top 3
1. Saunameister in public saunas. AI won’t be replacing that any time soon
2. Stay-at-home dad
3. Personal trainer for elderly ladies
Now I’m building software
won't matter since we're about to reset anyway
Touch humans in some way.
AI interaction is fundamentally unsatisfying. (I have a theory as to why; see below.) An AI can get you results, maybe even the results you wanted, but it can't give you a genuine smile. Even when we're surrounded by things that get us results, we still want the smile. So if you're doing something where you interact with people face to face, at least some of the time, that's a decent place.
In my large town/small city, there's an ice cream shop that closes between Thanksgiving and roughly March 1st. I would love to own a business that people cared as much about as they care about that place opening for the year.
-----
Why is interacting with an AI unsatisfying? I wonder if it doesn't come down to information theory. AI output has too little information in it.
You compute the information in a text by going symbol by symbol. Normally "symbol" means one character or byte. But you could do it by going word by word (or word part by word part) instead. Here you can see the problem: an AI gives you whatever word is the most probable next word, given the context. That's lower information than human output, which would give a wider variety of next words, and therefore a higher (Shannon) information content.
Yes, you can change that by increasing the temperature of the AI. Can you do that enough to give human-equivalent information content without destroying the coherence of the output? (Merely high Shannon information isn't the only goal. The insane may have high information in their output, but they are incoherent enough that it doesn't matter.)
People don't take this seriously, but i do think it's serious.
There are industries that one can switch to. Tourism, agriculture, healthcare are things where people will use the robots rather than be used by them.
Jobs that control the money flow will probably be able to make a cut for themselves, finance.
anything handcrafted. machines can build a table, but people will always value handcrafted tables. same with planting/gardening, breweries, etc.
I am sick of tech, been doing it my whole life. built a company and now I just want to put my hands in the dirt or on a tool.
Retiring, maybe writing a novel
zoo keeper
deep sea diver :)