Possible, until the costs come home to roost because even nVidia admitted that AI isn't cheaper than people.

I would also argue that AI is less useful than humans. I'm not even going to go into the bizarre circular financing, ridiculous energy costs that are currently being borne by average ratepayers that will come home to roost, the insane backlog of datacenters that are far exceeding any theory of profitability, or even the fact the best AI models get called a "nothingburger" or receive a collective "meh" upon release by professionals.

No let's focus on the grand pronouncements. There shouldn't be ANY PROGRAMMERS left now according to Dario Amodei. Claude should be programming everything. He said, last April, that it would take over in 9-12 months. We're in month 13 and it isn't close.

Where's the AGI? Even Sam's discount version of AGI, something that nobody who thinks of AI in movies would define as AGI, isn't here yet. That means it's two years late.

This reminds me of Tesla Elongelicals one of whom, last April during the runup to the latest super AI buzz, said Tesla had "cracked AGI" and their cars would be "driving everywhere soon without people" and he was giddy. I'm still waiting, and he's silent now, but saying "real soon" to Tesla AGI. To be fair, Elon has been fleecing suckers with this one running for about a decade now.

I'm sorry man, but I've been through this before: Sam and Dario know exactly what they are doing to keep the hype train going. They learned from Elon, keep the target moving so nobody in the press can tie you down. They are burning cash at a rate that is utterly insane. None of it is backed up by any GAAP approved financials, it's all "Trust us bro, it works, it's becoming AGI, it's profitable." They're so desperate now that they're playing more for the press than anyone... and that shows how bankrupt their progress is today.

LLM's are looking to be nothing more than an extremely expensive technological dead end.

Years ago my high school computer lab teacher taught us in our programming portion of the class: "When you don't know if it is secure, you should use the premise it is insecure until proven otherwise."

I think I can see why American corporations fail basic security audits today. They don't allow their programmers to follow basic concepts that were taught in high school.

This is likely a bit facetious but similar to spelling, some people never forget. I remember phone numbers from when I was a child. My kids remember numbers. Why? Because we taught them to remember numbers in case they didn't have access to their phone.

Just saying.

A minor addition?

I know companies that flat out block access to all models outright on all corporate assets. Why? Contractual security requirements. Adding a model, even a theoretically local model, is something that would cause serious heartburn for compliance.

When you make big changes, you don't hide it in the release notes you make an announcement. Don't believe me? You'd better see what happens when Microsoft pulled that crap in the past. They were absolutely roasted.

Let's be brutally honest and say Google is the first, and they expect defense from people like yourself. You make it easier to offer a half-hearted apology.

The simple fact is it appears to be (mostly) bluster and hype. Don't forget that Dario said Claude would be programming EVERYTHING.

AI is quickly becoming the Full Self Driving (NEXT YEAR!) of the computer world, only with an even higher cash burn rate. They (Sam, Dario, et al.) learned well from Elon.

There's a slight difference for farms you might want to consider before climbing on that horse: FOOD FEEDS PEOPLE. Now I'm not one to say flood farming, particularly for almonds, is a great idea. That being said part of the reason farms in California (or even Arizona) use all that water is because of the idiotic way we handle water rights in the US. They stop using that much water... they don't get as much a year later when they might need more water to maintain the crops due to altered weather.

Golf courses aren't my thing, but again, grass a growing thing that humans interact with directly in the physical world. Personally I can think of better uses for human uses, but even the golf course helps with area cooling because it's a green area not blacktop and concrete. So, again, you're climbing on a high horse that is pretty flipping stupid.

I'm not going to get into the whole "b-b-but datacenters totes don't use water like..." because obviously this data center used a fair amount of water. Personally speaking they should be fined 10 times the illegally taken amount. They also shouldn't get away with, "b-b-but we totes thought we were paying..." because as someone who approved bills to go to my accountant for the facility I ran you flipping well can read or you wouldn't be in charge. Either that or the company is completely incompetent and it sucks to be them, have a nice fine and explain your financial mismanagement to shareholders.

Did they add it without notification? Here's a hint snowflake: adding stuff that is security flagged to software without notification at the Enterprise level is not a good thing. Google, of all companies, should know that AI models are verboten on many systems for contractual reasons.

Saying "you can disable this with flags or GPO" after the fact is unacceptable.

Silently adding an AI model is something that is going to get Chrome ripped out by certain employers.

A lot of companies limit AI model access. That means Google doing this in secret is considered a huge InfoSec red flag. At least one company I know will have Chrome ripped off ALL corporate assets (computers / phones / et al.) by the end of this weekend. They will then ask Google for a version that will NEVER install the AI software without central approval. Google says no? No Chrome on corporate assets.

They have contractual requirements that cannot be avoided.

Just like square this is more AI washed separation. I have yet to see a single AI model that truly saves time. The recent case of Oracle eliminating jobs is related to the fact their Special Purpose Vehicle (to move debt slightly off book) is growing to a dangerous point. Right now they have BILLIONS in debt, that can only be serviced if OpenAI can pay them more than 70 Billion on a regular basis. This is insane.

Who honestly believes these tools are actually valuable. I've seen people using Claude and other models... they're at best copy / paste tools.

The entirety of AI is looking like one of the biggest rug pulls in history, and possibly the largest indictment of late stage capitalism. Elon Musk's "negotiation" (do it or I'll find someone who will do what I want) to get SpaceX added to the index far faster than ever, to get index funds buying shares, should be a SCREAMING warning whistle.

I dunno, I felt like the IIe was peak Apple 2.

We're talking about different things. I'm talking about load shifting, you're talking about base load and frequency maintenance. You're not wrong, but that's not what I was talking about.

The point of load shifting is that if I have a task I need to do today that consumes power, I can do it when the sun is shining or when it's not. If the power company lets me do it for free because the sun is shining and there is excess power, that helps them to keep the grid balanced during the day, which is good, and that's what you're talking about.

What I'm talking about is that if you have a task that you were always going to do, and you would have done it at, say, 7pm, when renewables generation is low and load is high, and I incentivize you to do it at 4pm, when demand is high and renewables generation is higher, then you aren't going to do that task at 7pm, which means the total load on the grid at 7pm will be less. If I can shift enough of the load away from 7pm, then I don't have to turn on a coal plant in anticipation of base load need at 7pm. That can bring my cost per mwh down from £500/kwh to £40 per mwh. Load is high at that time, so that can save millions of pounds over the course of an hour.

It's really not nonsensical, actually. Base load can be incredibly expensive. If they can avoid firing up the most expensive plant, they make more money. It's really that simple. Even though it seems "free" to you, what's really going on is that you have become part of the supply side of the equation by using power when it's there, and then _not_ using it when an expensive plant would have to be turned on. This is really a case where everybody wins.

No, it's really inefficient. In order to be useful for power generation, the three square mile circle it illuminates would have to be completely full of solar panels in order to capture all the energy being reflected. And it it's as bright as the moon, that's about one half millionth as bright as the sun. So those solar panels, assuming no cloud cover, will be operating at one millionth the efficiency of daytime.

Meanwhile, battery technology, particularly for terrestrial power storage, keeps getting better and better. This has zero potential to offset CO2. Which is deeply sad for the science fiction geek in us all, but honestly, right now solar generation technology is starting to feel pretty science-fictiony, so maybe that's okay.

The other thing about saving is that if you can depend on UBI, and it's enough to live on, then that takes the pressure off of individuals saving for retirement. Right now the amount of money people have to save for retirement in the U.S. is actually a problem, because there's no safe place to put that much money. And so we wind up with things like private equity and various other forms of securitization a specific group of which led to the 2008 crisis.

All of these securities are just ways of storing value, but you can't actually store value—value is work. "Stored value" is an obligation that someone else will have to work to pay back: I use my wealth to pay you money to do the work that I need done.

So public support for people who need it is actually the same thing as living off savings, except that living off savings is individual, and public support is collective. So public support can take advantage of the law of averages, and private savings can't. Which massively increases the amount you have to save as an individual to be sure you'll be okay in retirement.

And this motivates wealth inequality, which makes things worse and worse for the people who are creating the value you as a person with a decent amount of retirement savings need done. We've already had people saying "no more taxes" because they don't want to work to pay for other peoples' retirements. This is the same thing, and at some point it either turns into runaway inflation, which means your savings loses its value, or else it turns into regime change, which means who knows what? Right now, it means that a bunch of elected people are just raking in money through fraud, which isn't likely to end well for the rest of us.

It's weird how people think of socialism as being somehow expensive in comparison.