Will AI replace most tech jobs?

[twinstudy]

I see your point, but I wonder about the broader societal effects of the continual elimination of basic jobs.  Keeping with the medical coder example, that's a job that doesn't require a college degree (though you may need a technical school certification, if your employer doesn't provide OJT) and that has an annual salary of between $48,000 and $77,000.  Like, that's a solid job for a lot of people.  And I would wager that most of the people that become medical coders in the first place are not going to transition into becoming accountants, doctors, lawyers, or programmers when those jobs go away.  Hell, my father was a reasonably successful architect with his own practice, and he elected to retire 20 years ago rather than continue to adapt to the increasing technological requirements of his field.

I've read enough of your posts to feel confident that your position is generally something like, "Smart people will adapt and succeed, and anyone not smart or flexible enough doesn't deserve to."  But, non-smart and non-adaptable people still need to eat.  They still need a place to live.  They still need medicines and treatments for illnesses and injuries.  There are a lot of jobs out there that don't require a tremendous amount of knowledge or creativity or analysis to perform.  They just require someone to sit down and do them.  As these sorts of jobs get automated away, more and more people get pushed into competition for the jobs that haven't been automated yet. 

I agree that AI has the potential to continue the trend we are already seeing of bimodal outcomes in the job market. To date, that hasn't resulted in what I would consider to be untenable amounts of societal unrest (the US is an edge case, and the UK is not doing great, but if you look at other countries like Canada, NZ, and Australia, they are doing fine). If it gets dicey, I think the solution would be for the government to provide universal basic services, so basically, supplemental income for food and housing as needed by the people, funded by taxes (and of course universal emergency health care which only the US doesn't have).

[LaineyAZ]

But the trick in funding this is, taxes paid by who?  Billionaires don't want to pay and corporations not only don't pay, they actually get refunds.  This administration will only worsen that trend.

So the U.S. continues to borrow money.  Unless tax loopholes are closed and laws changed, this is in a downward spiral.

[Ron Scott]

I see your point, but I wonder about the broader societal effects of the continual elimination of basic jobs.  Keeping with the medical coder example, that's a job that doesn't require a college degree (though you may need a technical school certification, if your employer doesn't provide OJT) and that has an annual salary of between $48,000 and $77,000.  Like, that's a solid job for a lot of people.  And I would wager that most of the people that become medical coders in the first place are not going to transition into becoming accountants, doctors, lawyers, or programmers when those jobs go away.  Hell, my father was a reasonably successful architect with his own practice, and he elected to retire 20 years ago rather than continue to adapt to the increasing technological requirements of his field.

I've read enough of your posts to feel confident that your position is generally something like, "Smart people will adapt and succeed, and anyone not smart or flexible enough doesn't deserve to."  But, non-smart and non-adaptable people still need to eat.  They still need a place to live.  They still need medicines and treatments for illnesses and injuries.  There are a lot of jobs out there that don't require a tremendous amount of knowledge or creativity or analysis to perform.  They just require someone to sit down and do them.  As these sorts of jobs get automated away, more and more people get pushed into competition for the jobs that haven't been automated yet. 

But the trick in funding this is, taxes paid by who?  Billionaires don't want to pay and corporations not only don't pay, they actually get refunds.  This administration will only worsen that trend.

So the U.S. continues to borrow money.  Unless tax loopholes are closed and laws changed, this is in a downward spiral.

Perhaps what’s missing in the discussion is the fact that if all these human jobs are “lost” the actual cost of products will simultaneously be plummeting. And if AI can develop products with cheap synthetic materials of its own design instead of the raw materials we currently mine and refine, prices will plunge even further. Maybe the provision of life’s necessities will become “too cheap to meter”.

Yes, there will definitely be a period of transition—where job losses need to be addressed—but what if the world morphs from being driven by humans competing for scarce resources to humans basking in abundance? What if the amount and quality of material goods readily available to you was more than you needed or wanted? What if the comparative benefits of being wealthy were reduced to di minimis levels and coveted by fewer and fewer people?

It has only been in the past couple, 3 centuries that a sizable portion of the human population in the world had the ability to improve their personal circumstances through their own efforts. What if AIs replaced humans in all activities required to sustain healthy, comfortable, and satisfying lives?

Then everything we ever knew about society will change and nothing will ever be the same.

Then what?

[jrhampt]

What possibly makes you think that prices will plummet instead of corporations just taking more profits and the gains all going to shareholders?

[YttriumNitrate]

What possibly makes you think that prices will plummet instead of corporations just taking more profits and the gains all going to shareholders?

Look at the so called "Robber Barons" of the 19th century. John D. Rockefeller became the first billionaire by lowering the price of oil from 30 cents a gallon to 8 cents.

[Paul der Krake]

What possibly makes you think that prices will plummet instead of corporations just taking more profits and the gains all going to shareholders?

Why do prices ever drop?

[Ron Scott]

What possibly makes you think that prices will plummet instead of corporations just taking more profits and the gains all going to shareholders?

Why do prices ever drop?

Seems the price of computing has plummeted dramatically over the years…and after the breakup of AT&T and the ensuing competition and tech advancements, phone communications are far cheaper now than before.

To generate the computing power of a modern, $1000 iPhone in 1960 might cost hundreds of billions of dollars, take up as much space as the Pentagon, and use as much electricity as a small city did then.

Also, consider that when America was founded, the average family spent up to 75% of their income on food! Advances in agriculture and the Industrial Revolution blew that paradigm out of the water.

So competition and new, more productive technology are key here: Let’s say a business exists in a simple 5 company industry in which each player holds an equal share of the market. If company #1 can use AI to reduce their cost of production significantly they can maximize the value of their company by gaining market share with as small a reduction in pricing as possible. If the AI solution is figured out by the other 4 companies, pricing will reset at a much lower level.

Similarly, just look at the price drops in drugs when their patents expired. Prozac and Lipitor prices dropped more than 90% when multiple generics hit the market.

So yeah, prices drop…

[Ron Scott]

What possibly makes you think that prices will plummet instead of corporations just taking more profits and the gains all going to shareholders?

Why do prices ever drop?

Seems the price of computing has plummeted dramatically over the years…and after the breakup of AT&T and the ensuing competition and tech advancements, phone communications are far cheaper now than before.

To generate the computing power of a modern, $1000 iPhone in 1960 might cost hundreds of billions of dollars, take up as much space as the Pentagon, and use as much electricity as a small city did then.

Also, consider that when America was founded, the average family spent up to 75% of their income on food! Advances in agriculture and the Industrial Revolution blew that paradigm out of the water.

So competition and new, more productive technology are key here: Let’s say a business exists in a simple 5 company industry in which each player holds an equal share of the market. If company #1 can use AI to reduce their cost of production significantly they can maximize the value of their company by gaining market share with as small a reduction in pricing as possible. If the AI solution is figured out by the other 4 companies, pricing will reset at a much lower level.

Similarly, just look at the price drops in drugs when their patents expired. Prozac and Lipitor prices dropped more than 90% when multiple generics hit the market.

So yeah, prices drop…

…per ChatGPT

Has AI yet been responsible for a decline in prices of any products?

AI has already contributed to price declines in several industries and products by improving efficiency, reducing production costs, and enhancing supply chain management. One notable area is consumer electronics and hardware, where AI-powered automation and robotics have reduced labor costs and increased manufacturing efficiency. This has helped stabilize or lower the prices of products such as smartphones and laptops, even as they have grown more complex. Additionally, e-commerce platforms utilize AI-driven dynamic pricing, which adjusts prices based on demand and competition, often leading to discounts for consumers.

In renewable energy, AI has played a role in the significant price drop of solar panels and lithium-ion batteries by optimizing manufacturing processes and streamlining supply chain logistics. Furthermore, AI’s applications in energy management have reduced electricity costs for consumers who rely on renewable energy sources. Similarly, cloud computing services have benefited from AI-driven improvements in server utilization and cooling systems. These advancements have allowed providers like AWS, Google Cloud, and Microsoft Azure to offer more competitive pricing for storage and computing power.

AI’s impact extends to food delivery and grocery industries, where it optimizes supply chains and delivery routes. Companies like Walmart and Instacart use AI to minimize waste and improve efficiency, which sometimes translates into lower prices for consumers. By managing perishable goods more effectively, AI also reduces spoilage and markdowns, further cutting costs.

The field of online education has seen lower costs as AI automates content creation and personalizes learning paths. These innovations have reduced the expense of developing and delivering digital courses, making e-learning more affordable. Similarly, in transportation, ridesharing companies such as Uber and Lyft use AI for route optimization and surge pricing, which can result in lower fares during off-peak hours. AI has also contributed to the declining prices of electric vehicles by driving advancements in battery technology and manufacturing.

Pharmaceuticals have also felt the impact of AI, particularly in drug discovery. By streamlining research and clinical trials, AI has reduced the cost and time required to develop new medications, potentially lowering the price of certain drugs.

Overall, AI’s influence on pricing is most evident in industries where it increases efficiency, reduces waste, and automates processes. Although its effect is not uniform across all sectors, its growing role in technology, energy, and logistics-intensive industries is making many products and services more affordable.

[41_swish]

That literally sounds like mad libs. This stuff still lacks any critical thinking skills.

[GuitarStv]

That literally sounds like mad libs. This stuff still lacks any critical thinking skills.

Yeah, I shudder to think of the quality of e-learning that it says AI is churning out in seconds now.

[41_swish]

That literally sounds like mad libs. This stuff still lacks any critical thinking skills.

Yeah, I shudder to think of the quality of e-learning that it says AI is churning out in seconds now.

I have a friend who is an educator, and he uses AI to make the lesson plans.... And homework assignments.... It is about as generic and formulaic as you think it would be... He uses it for grade 9 biology. He also uses it to auto generate quizzes based off of YouTube videos. The youth is so cooked. He told me multiple teachers at his school shamelessly do this..... It is sad and he sees nothing wrong with it

[GuitarStv]

That literally sounds like mad libs. This stuff still lacks any critical thinking skills.

Yeah, I shudder to think of the quality of e-learning that it says AI is churning out in seconds now.

I have a friend who is an educator, and he uses AI to make the lesson plans.... And homework assignments.... It is about as generic and formulaic as you think it would be... He uses it for grade 9 biology. He also uses it to auto generate quizzes based off of YouTube videos. The youth is so cooked. He told me multiple teachers at his school shamelessly do this..... It is sad and he sees nothing wrong with it

The funny part is, most of the answers he's getting back from his students are probably AI generated too.  :S

[41_swish]

That literally sounds like mad libs. This stuff still lacks any critical thinking skills.

Yeah, I shudder to think of the quality of e-learning that it says AI is churning out in seconds now.

I have a friend who is an educator, and he uses AI to make the lesson plans.... And homework assignments.... It is about as generic and formulaic as you think it would be... He uses it for grade 9 biology. He also uses it to auto generate quizzes based off of YouTube videos. The youth is so cooked. He told me multiple teachers at his school shamelessly do this..... It is sad and he sees nothing wrong with it

The funny part is, most of the answers he's getting back from his students are probably AI generated too.  :S

He knows that they are using AI. They know that he is using AI. It is just deplorable.... It kind of feels like a cat and mouse game. I feel like education will see a pretty big shake up in the next 10 years due to AI.

[Ron Scott]

That literally sounds like mad libs. This stuff still lacks any critical thinking skills.

Yeah, I shudder to think of the quality of e-learning that it says AI is churning out in seconds now.

I have a friend who is an educator, and he uses AI to make the lesson plans.... And homework assignments.... It is about as generic and formulaic as you think it would be... He uses it for grade 9 biology. He also uses it to auto generate quizzes based off of YouTube videos. The youth is so cooked. He told me multiple teachers at his school shamelessly do this..... It is sad and he sees nothing wrong with it

The funny part is, most of the answers he's getting back from his students are probably AI generated too.  :S

He knows that they are using AI. They know that he is using AI. It is just deplorable.... It kind of feels like a cat and mouse game. I feel like education will see a pretty big shake up in the next 10 years due to AI.

I always felt the true test of understanding subject matter that was tested today with essay assignments lay in the ability to debate, take oral exams, and make thoughtful presentations with Q&A.

[GuitarStv]

That literally sounds like mad libs. This stuff still lacks any critical thinking skills.

Yeah, I shudder to think of the quality of e-learning that it says AI is churning out in seconds now.

I have a friend who is an educator, and he uses AI to make the lesson plans.... And homework assignments.... It is about as generic and formulaic as you think it would be... He uses it for grade 9 biology. He also uses it to auto generate quizzes based off of YouTube videos. The youth is so cooked. He told me multiple teachers at his school shamelessly do this..... It is sad and he sees nothing wrong with it

The funny part is, most of the answers he's getting back from his students are probably AI generated too.  :S

He knows that they are using AI. They know that he is using AI. It is just deplorable.... It kind of feels like a cat and mouse game. I feel like education will see a pretty big shake up in the next 10 years due to AI.

I always felt the true test of understanding subject matter that was tested today with essay assignments lay in the ability to debate, take oral exams, and make thoughtful presentations with Q&A.

I think I agree with that.  I've always hated doing presentations, and would spend hours and hours meticulously preparing every second of them.  In university I discovered that skipping the preparation and simply studying the material inside out then winging the presentation usually resulted in a better result.  :P

[41_swish]

That literally sounds like mad libs. This stuff still lacks any critical thinking skills.

Yeah, I shudder to think of the quality of e-learning that it says AI is churning out in seconds now.

I have a friend who is an educator, and he uses AI to make the lesson plans.... And homework assignments.... It is about as generic and formulaic as you think it would be... He uses it for grade 9 biology. He also uses it to auto generate quizzes based off of YouTube videos. The youth is so cooked. He told me multiple teachers at his school shamelessly do this..... It is sad and he sees nothing wrong with it

The funny part is, most of the answers he's getting back from his students are probably AI generated too.  :S

He knows that they are using AI. They know that he is using AI. It is just deplorable.... It kind of feels like a cat and mouse game. I feel like education will see a pretty big shake up in the next 10 years due to AI.

I always felt the true test of understanding subject matter that was tested today with essay assignments lay in the ability to debate, take oral exams, and make thoughtful presentations with Q&A.

I think I agree with that.  I've always hated doing presentations, and would spend hours and hours meticulously preparing every second of them.  In university I discovered that skipping the preparation and simply studying the material inside out then winging the presentation usually resulted in a better result.  :P

I agree, too. Explaining a topic effectively in an essay / paragraph style is tough. When I was in Engineering school, and we had free response questions on exams those were always the toughest. In multiple choice you can get lucky or at least use deductive reasoning. In free response you have to know. Regardless, feeding our kids AI slop is probably not the best for their education, but what do I know? I am not a parent, I guess.

[Ron Scott]

That literally sounds like mad libs. This stuff still lacks any critical thinking skills.

Yeah, I shudder to think of the quality of e-learning that it says AI is churning out in seconds now.

I have a friend who is an educator, and he uses AI to make the lesson plans.... And homework assignments.... It is about as generic and formulaic as you think it would be... He uses it for grade 9 biology. He also uses it to auto generate quizzes based off of YouTube videos. The youth is so cooked. He told me multiple teachers at his school shamelessly do this..... It is sad and he sees nothing wrong with it

The funny part is, most of the answers he's getting back from his students are probably AI generated too.  :S

He knows that they are using AI. They know that he is using AI. It is just deplorable.... It kind of feels like a cat and mouse game. I feel like education will see a pretty big shake up in the next 10 years due to AI.

I always felt the true test of understanding subject matter that was tested today with essay assignments lay in the ability to debate, take oral exams, and make thoughtful presentations with Q&A.

I think I agree with that.  I've always hated doing presentations, and would spend hours and hours meticulously preparing every second of them.  In university I discovered that skipping the preparation and simply studying the material inside out then winging the presentation usually resulted in a better result.  :P

I agree, too. Explaining a topic effectively in an essay / paragraph style is tough. When I was in Engineering school, and we had free response questions on exams those were always the toughest. In multiple choice you can get lucky or at least use deductive reasoning. In free response you have to know. Regardless, feeding our kids AI slop is probably not the best for their education, but what do I know? I am not a parent, I guess.

That’s where I’m at. And, bottom line, I think AI can ultimately yield dramatic improvements in education. Just think of a tireless tutor that understands both your style of learning and the critical points of confusion you might have in mastering a concept. BUT today’s educators are going to have to adapt. They can’t be wedded to the lecture/essay model (boring!) and bitch that AI is taking that away from them.

[41_swish]

That literally sounds like mad libs. This stuff still lacks any critical thinking skills.

Yeah, I shudder to think of the quality of e-learning that it says AI is churning out in seconds now.

I have a friend who is an educator, and he uses AI to make the lesson plans.... And homework assignments.... It is about as generic and formulaic as you think it would be... He uses it for grade 9 biology. He also uses it to auto generate quizzes based off of YouTube videos. The youth is so cooked. He told me multiple teachers at his school shamelessly do this..... It is sad and he sees nothing wrong with it

The funny part is, most of the answers he's getting back from his students are probably AI generated too.  :S

He knows that they are using AI. They know that he is using AI. It is just deplorable.... It kind of feels like a cat and mouse game. I feel like education will see a pretty big shake up in the next 10 years due to AI.

I always felt the true test of understanding subject matter that was tested today with essay assignments lay in the ability to debate, take oral exams, and make thoughtful presentations with Q&A.

I think I agree with that.  I've always hated doing presentations, and would spend hours and hours meticulously preparing every second of them.  In university I discovered that skipping the preparation and simply studying the material inside out then winging the presentation usually resulted in a better result.  :P

I agree, too. Explaining a topic effectively in an essay / paragraph style is tough. When I was in Engineering school, and we had free response questions on exams those were always the toughest. In multiple choice you can get lucky or at least use deductive reasoning. In free response you have to know. Regardless, feeding our kids AI slop is probably not the best for their education, but what do I know? I am not a parent, I guess.

That’s where I’m at. And, bottom line, I think AI can ultimately yield dramatic improvements in education. Just think of a tireless tutor that understands both your style of learning and the critical points of confusion you might have in mastering a concept. BUT today’s educators are going to have to adapt. They can’t be wedded to the lecture/essay model (boring!) and bitch that AI is taking that away from them.

I am sure AI will change the education landscape, and nearly every industry, in some meaningful way. The idea of a tutor/teacher who knows your strengths and weaknesses for material and then tailors the lesson plan to that will be tremendously helpful. However, it just seems like it spits out mad libs type assignments. I am sure it will get better. We have already seen a great improvement in just two years. I actually think AI is the most useful at "textbook type" problems, where a clear solution exists. Helping kids understand that solution seems like something it will be able to do in the nearish future.

[Ron Scott]

Getting back to the original post and tech jobs it might be helpful to consider Cursor. These guys offer and are continuing to improve what they call “AI Code Editor” that helps developers code using AI prompts and promises dramatic productivity improvements.

Anybody play with their stuff?

References:

Website: https://www.cursor.com/

Reddit: https://www.reddit.com/r/CursorAI/

Interesting excerpt from the Lex Fridman podcast with leading techs at Cursor (comments section interesting too):
https://youtu.be/reH9VRMPRpY?si=Q4GjhomY4DWnfHZt

[Ron Scott]

Anybody interested in learning how AI startups are researching business niches to target high labor cost blocks of work will be interested in checking out this video. It is produced by the investment firm Andreesen-Horowitz—who are major Silicon Valley tech investors heavily into AI.

This is how AI will be systematically, replacing a good percentage of jobs, following a very generic B2B vendor business model used for many years:
https://youtu.be/6qQoZsfA_Iw?si=LzdB7UjLiUT6jhX7

[Ron Scott]

That's what I'm saying though. There won't be any cost savings. In law, which is adversarial, greater productivity just means greater weapons on both sides. It's a stalemate arms race.

At least among big law firms, only about a quarter of lawyers' time is spent on litigation.[1] I would expect most of the time savings from AI to come from this like drafting contracts, wills/estate planning, and government compliance.

Yes, this is like DH's job. He's in policy development and there are a few lawyers in his team and what they do is much like law in that it's processing enormous amounts of data and reports and creating more condensed reports and recommendations.

So much of what DH does could be done by AI. It's taken 2 decades of this kind of work for him to even be able to do the kind of work he does, to even know where to look for the kind of data he needs.

At least 50% of his job could be AI automated, which would be very, very interesting.

It is also possible that some of the more talented lawyers who become adept at using AI assistants become stars. Some might be able to do the work that requires 2 lawyers today. These stars could earn much more money, creating even more inequality. And we might expect stars to branch out on their own, possibly leaving large law firms without as much top talent.

It’s hard to say…

[LaineyAZ]

That's what I'm saying though. There won't be any cost savings. In law, which is adversarial, greater productivity just means greater weapons on both sides. It's a stalemate arms race.

At least among big law firms, only about a quarter of lawyers' time is spent on litigation.[1] I would expect most of the time savings from AI to come from this like drafting contracts, wills/estate planning, and government compliance.

Yes, this is like DH's job. He's in policy development and there are a few lawyers in his team and what they do is much like law in that it's processing enormous amounts of data and reports and creating more condensed reports and recommendations.

So much of what DH does could be done by AI. It's taken 2 decades of this kind of work for him to even be able to do the kind of work he does, to even know where to look for the kind of data he needs.

At least 50% of his job could be AI automated, which would be very, very interesting.

It is also possible that some of the more talented lawyers who become adept at using AI assistants become stars. Some might be able to do the work that requires 2 lawyers today. These stars could earn much more money, creating even more inequality. And we might expect stars to branch out on their own, possibly leaving large law firms without as much top talent.

It’s hard to say…

Another potential change due to AI:  law firms charge by the billable hour.  All professional staff in a law firm are required to bill a minimum number of hours per year, and bonuses are typically offered for any hours above that. 
This actually skews towards those who are inefficient because a faster experienced worker can get the same work done in fewer hours, but it hurts the law firm's bottom line even though efficiency saves the client money. 

I'm curious to see how that shakes out:  the idea of a superstar attorney + one AI assistant could be the norm?
or, do clients demand lower costs because a firm only has to employ a few workers?  I predict that attorneys won't be the well-paid professionals they are today.

[Financial.Velociraptor]

  I predict that attorneys won't be the well-paid professionals they are today.

Increasingly, they already aren't.  Operating a Law School is a lucrative business for both traditional universities and for profit schools.  This has resulted in an oversupply of such schools and an oversupply of JDs.  Combine this with rapidly growing outsourcing to India and the Philippines, and JD wages have been declining for decades.  AI is only going to compound the problem for new law grads.  There are some super rich ambulance chasers and some hybrid CPA/tax attorney types that run "family offices" but your regular first year to fifth year JD is dead broke after paying on loans and arguably could have been happier as a career barista. Hella lot more free time and free coffee.

When the accounting thing looked like it was going to be a dead end and before I got my first material work promotions, I looked into Law School.  Figured out real fast that if you don't go to an Ivy or graduate at the top of your class, odds of making 100k+ were actually pretty low.  I've since met dozens of JDs who work in accounting/budgeting, logistics, and other as there are no entry level attorney jobs for the average law grad that pay more than someone with a BBA in "general business studies" can earn as a pointy haired boss in training.

[Ron Scott]

Law is just one example of what I feel is a strong possibility that AI creates a massive productivity boom.

It’s hard for me to envision how (quickly) all the workers will find new good-paying jobs. Time will tell.

But it’s even harder to imagine how the cost of basic essentials would drop pretty fast.

I watched a video the other day in which they predicted we’d be able to lease an AI robot that could do most household chores in a couple years for about $300 a month. The timing would be perfect!

[Wolfpack Mustachian]

Law is just one example of what I feel is a strong possibility that AI creates a massive productivity boom.

It’s hard for me to envision how (quickly) all the workers will find new good-paying jobs. Time will tell.

But it’s even harder to imagine how the cost of basic essentials would drop pretty fast.

I watched a video the other day in which they predicted we’d be able to lease an AI robot that could do most household chores in a couple years for about $300 a month. The timing would be perfect!

I have some limited experience with robotics, and I find this very hard to believe even from a mechanical productive. I can't imagine robots with the physical and programming capabilities to do most household chores in two years on a mass produced scale for that price.

Although, it would be really impressive!

[MustacheAndaHalf]

Combine this with rapidly growing outsourcing to India and the Philippines, and JD wages have been declining for decades.

According to the American Bar Association, the average wages of a lawyer has been increasing for the past two decades, and spiked in the past few years alongside inflation.
https://www.americanbar.org/news/profile-legal-profession/wages/

[YttriumNitrate]

  I predict that attorneys won't be the well-paid professionals they are today.

Increasingly, they already aren't.  Operating a Law School is a lucrative business for both traditional universities and for profit schools.  This has resulted in an oversupply of such schools and an oversupply of JDs.

With the law, there are basically two separate groups and few people actually make the average starting salary.

https://www.lawhub.org/trends/salaries
It will be interesting to see which group is hurt more by AI.

[twinstudy]

  I predict that attorneys won't be the well-paid professionals they are today.

Increasingly, they already aren't.  Operating a Law School is a lucrative business for both traditional universities and for profit schools.  This has resulted in an oversupply of such schools and an oversupply of JDs.

With the law, there are basically two separate groups and few people actually make the average starting salary.

https://www.lawhub.org/trends/salaries
It will be interesting to see which group is hurt more by AI.

It won't be the higher-earning group. Do you think biglaw clients will opt for some dross written by a robot that can't even produce insight? Once you get to the top echelons of law (or most professions), 'value' as a concept loses its meaning and clients are happy to pay for what they perceive to be the best service. Right now Chat GPT puts out content similar to what a smart 8 year old could produce by paraphrasing wikipedia. Maybe if it improves 100 times over it'll be equivalent to what a junior lawyer could output.

[Ron Scott]

  I predict that attorneys won't be the well-paid professionals they are today.

Increasingly, they already aren't.  Operating a Law School is a lucrative business for both traditional universities and for profit schools.  This has resulted in an oversupply of such schools and an oversupply of JDs.

With the law, there are basically two separate groups and few people actually make the average starting salary.

https://www.lawhub.org/trends/salaries
It will be interesting to see which group is hurt more by AI.

It won't be the higher-earning group. Do you think biglaw clients will opt for some dross written by a robot that can't even produce insight? Once you get to the top echelons of law (or most professions), 'value' as a concept loses its meaning and clients are happy to pay for what they perceive to be the best service. Right now Chat GPT puts out content similar to what a smart 8 year old could produce by paraphrasing wikipedia. Maybe if it improves 100 times over it'll be equivalent to what a junior lawyer could output.

I agree this is another likely scenario. We just don’t know.

What I’m left with is a sense that Biden spent all of his administration’s AI energy on preventing China from catching up with the US (time will tell) while the need to prepare American business and labor for the coming wave of disruption has been ignored.

I don’t have confidence in Trump to step up, and I’m expecting the wave to slam into us unprepared, leaving us to scramble and ying-yang to maintain order.

[YttriumNitrate]

It won't be the higher-earning group. Do you think biglaw clients will opt for some dross written by a robot that can't even produce insight? Once you get to the top echelons of law (or most professions), 'value' as a concept loses its meaning and clients are happy to pay for what they perceive to be the best service. Right now Chat GPT puts out content similar to what a smart 8 year old could produce by paraphrasing wikipedia. Maybe if it improves 100 times over it'll be equivalent to what a junior lawyer could output.

You're assuming the false dichotomy that human-made and AI-made are the only options. If AI can double the output of an associate, then the number of associates employed by biglaw could be significantly reduced. I'm sure the Stanford and Yale law grads will be fine (and probably paid more), but things might get a bit dicey further down on the prestige scale.

[twinstudy]

It won't be the higher-earning group. Do you think biglaw clients will opt for some dross written by a robot that can't even produce insight? Once you get to the top echelons of law (or most professions), 'value' as a concept loses its meaning and clients are happy to pay for what they perceive to be the best service. Right now Chat GPT puts out content similar to what a smart 8 year old could produce by paraphrasing wikipedia. Maybe if it improves 100 times over it'll be equivalent to what a junior lawyer could output.

You're assuming the false dichotomy that human-made and AI-made are the only options. If AI can double the output of an associate, then the number of associates employed by biglaw could be significantly reduced. I'm sure the Stanford and Yale law grads will be fine (and probably paid more), but things might get a bit dicey further down on the prestige scale.

In theory, I suppose, and I agree that the non-big law stuff is ripe for the picking, but the practice of law is all about not making mistakes. AI can churn out stuff that 'seems' right to a layperson and, at lower levels of legal complexity, might in fact be 98% right. But it's the 2% it gets wrong that would be a critical problem in law, whether you are working on a litigated case, drafting a will, advising on a property settlement, etc. Indeed you are paying your human lawyer to not make major mistakes.

Humans also make mistakes too - the difference being that human lawyer mistakes will tend to be non-critical ones, since we learn by iteration what things to do or not do to avoid big mistakes. If a junior lawyer does research and the partner signs off on legal strategy, there is a degree of oversight and collaboration that is entirely lost if one of the two is replaced by AI.

[YttriumNitrate]

In theory, I suppose, and I agree that the non-big law stuff is ripe for the picking, but the practice of law is all about not making mistakes. AI can churn out stuff that 'seems' right to a layperson and, at lower levels of legal complexity, might in fact be 98% right. But it's the 2% it gets wrong that would be a critical problem in law, whether you are working on a litigated case, drafting a will, advising on a property settlement, etc. Indeed you are paying your human lawyer to not make major mistakes.

Humans also make mistakes too - the difference being that human lawyer mistakes will tend to be non-critical ones, since we learn by iteration what things to do or not do to avoid big mistakes. If a junior lawyer does research and the partner signs off on legal strategy, there is a degree of oversight and collaboration that is entirely lost if one of the two is replaced by AI.

The way I see it, right now AI is really good at polishing turds. Case and point would be Ron Scott's January 13, 2025, 08:42:27 AM post quoting an AI compilation of corporate marketing bullshit about how the corporations are using AI. It's well written and garbage at the same time.

While it will be a long time before AI can completely replace human lawyers, it won't take very long for the biglaw strategy to change from "junior lawyer does research and the partner signs off on legal strategy" to the much faster  "AI does the research, junior lawyer checks the research, partner signs off on legal strategy."

[twinstudy]

In theory, I suppose, and I agree that the non-big law stuff is ripe for the picking, but the practice of law is all about not making mistakes. AI can churn out stuff that 'seems' right to a layperson and, at lower levels of legal complexity, might in fact be 98% right. But it's the 2% it gets wrong that would be a critical problem in law, whether you are working on a litigated case, drafting a will, advising on a property settlement, etc. Indeed you are paying your human lawyer to not make major mistakes.

Humans also make mistakes too - the difference being that human lawyer mistakes will tend to be non-critical ones, since we learn by iteration what things to do or not do to avoid big mistakes. If a junior lawyer does research and the partner signs off on legal strategy, there is a degree of oversight and collaboration that is entirely lost if one of the two is replaced by AI.

The way I see it, right now AI is really good at polishing turds. Case and point would be Ron Scott's January 13, 2025, 08:42:27 AM post quoting an AI compilation of corporate marketing bullshit about how the corporations are using AI. It's well written and garbage at the same time.

While it will be a long time before AI can completely replace human lawyers, it won't take very long for the biglaw strategy to change from "junior lawyer does research and the partner signs off on legal strategy" to the much faster  "AI does the research, junior lawyer checks the research, partner signs off on legal strategy."

You can't properly check research without retracing the steps. For example, I don't know if a case citation is right without checking the citation. I don't know that a case summary is right without reading the case.  More problematically, I don't know if the AI has properly searched all relevant cases without repeating the search. At which stage, I might as well do it myself. AI is polishing turds and turning them into one big polished turdball, nothing more.

What AI is capable of right now is summarising or imitating, not researching or building from the ground up. And summarising/imitating, while useful for extremely low value output (like drafting a web article designed to bait clicks, or drafting an email or generic CV, or other menial work), has little value in higher-skilled fields where either insight or true accuracy is required.

[MustacheAndaHalf]

I think the rate of legal mistakes made by AI is much higher than 2%.

"A U.S. judge on Thursday imposed sanctions on two New York lawyers who submitted a legal brief that included six fictitious case citations generated by an artificial intelligence chatbot, ChatGPT."
https://www.reuters.com/legal/new-york-lawyers-sanctioned-using-fake-chatgpt-cases-legal-brief-2023-06-22/

[Ron Scott]

Survey of developers:

https://survey.stackoverflow.co/2024/ai

[FrugalToque]

AI will speed up the part where, having figured out how to solve the problem, I go to stack overflow for examples from people who have already solved similar problems.

But the part where I try to understand what the customer wants/needs and how to put that into the code base?  The whole "Problem Definition/Research/Design" phase that's about 50% of Software Engineering?
That part's probably still on me, and will be for quite some time.

[GuitarStv]

Survey of developers:

https://survey.stackoverflow.co/2024/ai

Does this prove much of anything?  I think AI will be integrated in many of the dev tools I use over the next year or two as well.  It doesn't mean that it will really help my work though.  AI is the latest big buzz word.  Like 'cloud' was several years back.  Or 'blockchain'.  Or 'big data'.  We've incorporated some of these buzzwords into functional tools.  Some of them have made differences to the way that we work.  None have been the earth-shattering complete reconfigurations of reality that they were sold to be initially.

[FrugalToque]

Survey of developers:

https://survey.stackoverflow.co/2024/ai

Does this prove much of anything?  I think AI will be integrated in many of the dev tools I use over the next year or two as well.  It doesn't mean that it will really help my work though.  AI is the latest big buzz word.  Like 'cloud' was several years back.  Or 'blockchain'.  Or 'big data'.  We've incorporated some of these buzzwords into functional tools.  Some of them have made differences to the way that we work.  None have been the earth-shattering complete reconfigurations of reality that they were sold to be initially.

It reminds me of that survey done about "Which drug is the most dangerous in each state" and they had a map of the U.S. with the words "marijuana" over most places.
But really, all they'd done is surveyed police officers about "which drug do you spend most of your time on?" or whatever.
This survey is a little better, but it looks like all AI is doing is making the coding part faster, not the designing part.
Maybe someday the A.I. will converse with the customer, read the bug report, figure out what is ACTUALLY wrong, and implement code that isn't garbage, but I'm not worried in the present.

[Ron Scott]

Survey of developers:

https://survey.stackoverflow.co/2024/ai

Does this prove much of anything?  I think AI will be integrated in many of the dev tools I use over the next year or two as well.  It doesn't mean that it will really help my work though.  AI is the latest big buzz word.  Like 'cloud' was several years back.  Or 'blockchain'.  Or 'big data'.  We've incorporated some of these buzzwords into functional tools.  Some of them have made differences to the way that we work.  None have been the earth-shattering complete reconfigurations of reality that they were sold to be initially.

I think it shows that most developers agree AI tools will be more integrated mostly in the ways they are documenting code, testing code, and writing code.

To the extent this integration actually takes place and, perhaps, allows 19 developers to do the work that requires 20 today, that will have a pretty massive improvement in productivity.

IMO you’re absolutely correct to contrast AI to other significant changes in technology, although each change is unique—as are their rates of adoption, improvement, and impact. My best guess is AI will stand out as having a larger impact over a shorter period of time.

Today, AI is merely showing some reasonable signs of utility. Let’s check back in a year or 2 to see where we are.

[Ron Scott]

SoftBank CEO Masayoshi Son, Sam Altman of OpenAI and Larry Ellison of Oracle to place half a Trillion bucks in a new AI superproject. Initial focus on massive improvements in healthcare. Construction already underway.

Project Stargate

https://apnews.com/article/trump-ai-openai-oracle-softbank-son-altman-ellison-be261f8a8ee07a0623d4170397348c41

[Ron Scott]

There is a loose consensus that humans will participate much less in the front line military conflicts of the future, as intelligent machines taking over. The US military is the largest employer on the planet…

Here’s an assessment from a center-left leaning publication on how Russia is viewing these shifts and build up it’s AI military strategy now:
https://www.cnas.org/publications/reports/the-role-of-ai-in-russias-confrontation-with-the-west

[reeshau]

Law is just one example of what I feel is a strong possibility that AI creates a massive productivity boom.

It’s hard for me to envision how (quickly) all the workers will find new good-paying jobs. Time will tell.

But it’s even harder to imagine how the cost of basic essentials would drop pretty fast.

I watched a video the other day in which they predicted we’d be able to lease an AI robot that could do most household chores in a couple years for about $300 a month. The timing would be perfect!

I have some limited experience with robotics, and I find this very hard to believe even from a mechanical productive. I can't imagine robots with the physical and programming capabilities to do most household chores in two years on a mass produced scale for that price.

Although, it would be really impressive!

Google "physical intelligence" to learn more about this.  Folding laundry, for example, is one of the toughest-rated AI tasks there is.  There isn't an internet waiting for you of examples of folded clothes, so robots have to learn the slow way--actually doing it.   A lot.  It more closely mimics the learning of a human baby.  Perhaps someone will link 1,000 robots to one model, and let them learn like an at colony,  but this will bring new challenges and new costs beyond LLM's.

[scottish]

Robots don't have to learn this way.   People know how to fold laundry and can get them off to a good start.   

Some academics are studying how to train a robot to do everything from scratch, but so far this seems impractical.    Of course when you have x million people working on these problems, the solutions change rapidly...

[reeshau]

Robots don't have to learn this way.   People know how to fold laundry and can get them off to a good start.   

Some academics are studying how to train a robot to do everything from scratch, but so far this seems impractical.    Of course when you have x million people working on these problems, the solutions change rapidly...

People know how to do it.  The difficulty is in demonstrating / explaining in a way that can be captured systematically.  And, in the wide variety of clothes / folding preferences.  Language is also like that, of course, but we have four decades of recorded examples, which tech companies can spend truckloads examining.  The experience of folding laundry has nowhere near as much documentation.

[GuitarStv]

I think the key with laundry folding is in realizing that laundry is best stored in unkempt piles near the laundry room.

[Just Joe]

Laundry baskets. One for clean, one for dirty. Don't mix them up except in an emergency. Sniff tests required for validation.

[Ron Scott]

Law is just one example of what I feel is a strong possibility that AI creates a massive productivity boom.

It’s hard for me to envision how (quickly) all the workers will find new good-paying jobs. Time will tell.

But it’s even harder to imagine how the cost of basic essentials would drop pretty fast.

I watched a video the other day in which they predicted we’d be able to lease an AI robot that could do most household chores in a couple years for about $300 a month. The timing would be perfect!

I have some limited experience with robotics, and I find this very hard to believe even from a mechanical productive. I can't imagine robots with the physical and programming capabilities to do most household chores in two years on a mass produced scale for that price.

Although, it would be really impressive!

Google "physical intelligence" to learn more about this.  Folding laundry, for example, is one of the toughest-rated AI tasks there is.  There isn't an internet waiting for you of examples of folded clothes, so robots have to learn the slow way--actually doing it.   A lot.  It more closely mimics the learning of a human baby.  Perhaps someone will link 1,000 robots to one model, and let them learn like an at colony,  but this will bring new challenges and new costs beyond LLM's.

There’s so much talk about AI peaking because it’s used all the internet data—as if there’s nothing more to life than the internet.  LOL—get off the computer and into the real world if you really want to learn something!

BTW…gotta go…

[jrhampt]

I think the key with laundry folding is in realizing that laundry is best stored in unkempt piles near the laundry room.

And also on the pants chair

[reeshau]

I think the key with laundry folding is in realizing that laundry is best stored in unkempt piles near the laundry room.

And also on the pants chair

There is somewhat of a solution you you--the laundry jet.

https://laundryjet.com/

It will *do* your laundry without much more organization than throwing it somewhere.  But it still won't *fold* your laundry.

[Wolfpack Mustachian]

I think the key with laundry folding is in realizing that laundry is best stored in unkempt piles near the laundry room.

And also on the pants chair

There is somewhat of a solution you you--the laundry jet.

https://laundryjet.com/

It will *do* your laundry without much more organization than throwing it somewhere.  But it still won't *fold* your laundry.

Wow... That house is massive