It won't; predictions of what a company thinks it might do five years in advance are generally worthless. From the company's point of view, saying "yes" will please shareholders and has no consequences, saying "no" will displease shareholders. I'm actually surprised the number claiming this is so low.
These same CEOs still don't realize that AI is very suited to replacing executive roles and outperforms your average CEO in every aspect without the human downsides. Shareholders would be way more interested in having a static board of AIs running public companies than replacing lowly workers that frankly will always be relatively cheap.
On the same coin, a lot of government can probably be performed by AI as well. Reminds me The Matrix where humanity delegates all governance to AI/machines
Why do you think these CEOs do not get that? If you can be replaced by AI and it shows better value/risk I am sure you will be in the job market. As long as AI isn't trusted the cost of a CEO to a company is small compared to scale of the company.
Replacing a CEO would save a lot less money than replacing even a small percentage of workers.
The typical CEO of a F500 company makes $17.7 million compensation. That might seem like a lot, but it's only about 0.04% of the average F500 revenue, which is $42 billion.
On the other hand, the average F500 has about 65000 employees and payroll expenses can easily be upwards of $5 billion. About 300 times as much money is spent on employees than on the CEO!
The discussion about AI-driven workforce reductions misses the point entirely: on a macro level the real problem isn’t job losses — it’s the persistent stagnation in productivity growth, particularly in developed economies. Historically, automation has driven significant increases in productivity, which in turn raised living standards and created new industries.
However, recent decades have failed to deliver those gains. Productivity growth in the U.S., for instance, has been below 1% for most of the last 15 years, despite massive advances in technology. Given that AI should be increasing productivity, I tend to think that the actual utility of most of our advances in AI is massively out out of proportion with the reality.
I remember back in 2012 I was working for UPS as a package sorter. All along the aisle were leftover machines covered in dust and grease. They were all leftover from a time when some bright spark tried to automate package sorting.
The idea seems pretty easy at first. Read the label, send it to the correct area. But if you work at it for a few hours you realize how much skill it takes to do properly. Labels are obstructed, damaged, crossed out, and modified. I've seen boxes with 5-6 different labels all in various states and defaced in various ways.
Then comes the part where you need to get the packages from the input belt onto the correct outputs. Every package was a different shape, size, and condition. A lot had broken open. Some were nothing but masking tape in the shape of a mushy bowling ball. We would then often have to buffer the packages while one belt went down.
I would sometimes spend hours thinking about what it would take to automate my $15/hour job and just kept coming up with ridiculously complicated robotic projects with AI that is only possible now with extremely expensive computation. Even then it would need a ton of human supervision.
17 comments
[ 3.1 ms ] story [ 23.9 ms ] threadhow many new jobs will come to exist?
Capitalism is going to take on an extremely interesting form.
https://news.ycombinator.com/item?id=42637197
On the same coin, a lot of government can probably be performed by AI as well. Reminds me The Matrix where humanity delegates all governance to AI/machines
The typical CEO of a F500 company makes $17.7 million compensation. That might seem like a lot, but it's only about 0.04% of the average F500 revenue, which is $42 billion.
On the other hand, the average F500 has about 65000 employees and payroll expenses can easily be upwards of $5 billion. About 300 times as much money is spent on employees than on the CEO!
However, recent decades have failed to deliver those gains. Productivity growth in the U.S., for instance, has been below 1% for most of the last 15 years, despite massive advances in technology. Given that AI should be increasing productivity, I tend to think that the actual utility of most of our advances in AI is massively out out of proportion with the reality.
The idea seems pretty easy at first. Read the label, send it to the correct area. But if you work at it for a few hours you realize how much skill it takes to do properly. Labels are obstructed, damaged, crossed out, and modified. I've seen boxes with 5-6 different labels all in various states and defaced in various ways.
Then comes the part where you need to get the packages from the input belt onto the correct outputs. Every package was a different shape, size, and condition. A lot had broken open. Some were nothing but masking tape in the shape of a mushy bowling ball. We would then often have to buffer the packages while one belt went down.
I would sometimes spend hours thinking about what it would take to automate my $15/hour job and just kept coming up with ridiculously complicated robotic projects with AI that is only possible now with extremely expensive computation. Even then it would need a ton of human supervision.