Employee compensation comes from capital. And employees are working at companies that provide robots, etc.
There's a return on capital than is not spent on employees. That reflects how much capital is growing and how much can be spent on employees in the future.
>And employees are working at companies that provide robots, etc.
Just as are the top executives. And the shareholders that have put money into companies that provide "robots, etc.". All these people, including labor, are stakeholders. If there was 5% GDP growth that got reflected as 5% growth in net earnings for the company, one would expect that all the stakeholders would see roughly a 5% increase in their personal earnings from the company. The dollar amount would be higher for higher earners (5% of $1M is greater than 5% of $50k), but the percentage increase would be roughly in line. The real world results are not even close to this "rising tide lifts all boats" ideal.
>They didn't make the rising tide analogy, I read it as how much could be captured by labor if we increased leverage.
Fair point, though it's not completely clear from the comment.
>Wages have historically been a lagging indicator.
Of course, companies don't know in advance that they're going to have GDP-assisted growth. My point was that growth on the back of GDP growth is a collective windfall, and you'd expect it to be evenly distributed. But it clearly isn't.
All human collective endeavors (with few exceptions) require 3 kinds of human-related input: capital, labor and ideas.
Nobody puts their capital into an endeavor in which the plan is for the that capital to provide renumeration for the labor for more than the shortest possible time (*). The goal is always to generate revenue in sufficient volume to pay for the labor, and when that goal is met, that success is a function of all 3 kinds of contribution.
So no, employee compensation does not come from capital, but from revenue that results from the successful interaction of capital, labor and ideas.
(*) non-profits would be an obvious exception, except that nobody actually talks about investing capital in such organizations, we just make "donations" or "grants". That money plays the same role as capital, however.
There's no hypothetical, they tried to equate this to basically buying a gun where you do the certification on purchase. Internet connected devices are not planned to be controlled like that but per session. Under that model what they proposed is obviously not going to work.
Parental controls. Client-side blocking of content. Apple's parental controls are so good that parents give young children unsupervised internet access--the entire generation if iPad babies. Current iPads even use on-device AI to detect nudity in photos and prevent them from being taken, sent, recieved, or viewed.
Just because intelligence evolved in people that find rights useful doesn't mean intelligence can only reside in a person.
Living things are driven by a need to reproduce. That's the only reason we exist. The only reason we have self interest.
A machine doesn't require self interest. There's no reason to implement it, except to show it can be done. And of course it can. There's just no practical reason to. It becomes less useful to us.
which (among other things) documents an experiment in which a current-gen AI model attempted to blackmail someone in order to prevent it from being turned off.
Anthropic is not a disinterested party here, and until their experiments can be replicated from an adversarial standpoint by people without a vested interest in hyping up the tech (i.e. one assuming the null hypothesis), I wouldn't consider them to be "good evidence".
16 frontier models from multiple vendors all showing significant "alignment" issues, and tendencies to act "unethically" when threatened with shutdown.
Other models that resorted to blackmail in an attempt to avoid getting shut down: DeepSeek-R1 (79% of the time), Gemini-2.5-Pro (95% of the time), GPT-4.1 (80% of the time), Grok-3-beta (80% of the time).
There's quite a large chunk of emerging literature studying the "alignment problem" at this point, and no shortage papers that are are completely untained by Anthropic self interest (a series of papers studying the "alignment" problem coming out of Chinese universities, for example).
Of course it isn't. What it knows is many weeks old, except for what you tell it. Then it forgets everything you told it.
And it's just a model. It can have many exact copies. It has no life of its own. It doesn't know anything about where it is. It doesn't have sensory organs.
