Just as human accountants still have a job after the "paper spreadsheet" was replaced the with the "computer spreadsheet" in the 70s-80s, as more work is completed, then more work will be given.
I figure with AI will be the same.
Humans do a good job at thinking of more work to do.
As the more math problems that are solved, may lead to future questions to work on.
This is somewhat related to Jevons paradox.
TL;DR, with proper regulations in place, the datacenter build out can be beneficial to residential electrical customers and tax payers, however, without proper regulations, there's significant risks, especially if the AI datacenter bubble pops.
The historical data shows in most cases on the long term, datacenter's slow down residential price increases, with some exceptions where they may receive extreme subsidies, or in places with limited electrical power to begin with.
However, one big concern is if the AI datacenter boom pops, and many AI datacenters go out of business, then a lot of infrastructure could go unpaid, and the cost be forced on tax payers and residential customers.
For the water issues, fresh water may be cheap in the short term, until it dry's up, then the price sky rockets. This needs to be kept in mind, as fresh ground water isn't unlimited, and can be used faster than it is replinished.
Direct-to-chip liquid cooling seems promising, but is only a part of the solution.
"The Naming Convention: The "0.7 nm" label is an industry generation name (or "process node") rather than an exact physical measurement of any single component on the chip.
The Vertical Solution: Because physically shrinking individual parts any further causes extreme electricity leakage and quantum interference, IBM stopped trying to make the individual components thinner laterally. Instead, they achieved this density by stacking the nanosheets vertically across two layers using their 3D NanoStack architecture."
-LLM
"The 0.7nm (7 Angstrom) label refers to a marketing process node representing equivalent density, while the physical nanosheet layers in IBM's new chip are roughly 5 nanometers thick. By adopting a 3D "Nanostack" architecture to build upward rather than shrinking components, IBM achieved 100 billion transistors on a single chip. For more details, visit Memeburn."
Alan Kay's Views on Computers in Schools
Key points summarizing Alan Kay's views on computers in schools, drawn from his interviews, writings (e.g., "Powerful Ideas Need Love Too!" in 1995), and talks over the decades:
Computers are tools, not magic solutions
The computer is simply an instrument whose music is ideas. Knowledge and real intellectual growth aren't stored in the machine itself; it's a medium for powerful thinking, but only when used with deep ideas and skilled guidance.
The piano analogy
Putting a piano in every classroom won't create a developed music culture, because the music lives in people (musicians), not the instrument. Similarly, placing computers in every school doesn't automatically improve education or foster deep learning — schools often lack ideas for meaningful use.
Tokenism and superficial adoption
Schools frequently have few real plans for computers beyond access. This leads to "tokenism" where tech is present but underused, much like having abundant books or pencils without improving reading or math skills.
"Chopsticks culture" risk
Without proper support (e.g., retraining teachers as "musicians" in ideas/computing), kids might just tinker playfully ("piano by bricolage" or chopsticks-style playing), creating superficial engagement rather than genuine growth or new ways of thinking.
Sad observations in classrooms
Kay has described visiting "computerized classrooms" where everyone seems happy and engaged, but closer inspection shows little that's intellectually stimulating or truly growth-promoting — more like "junk food" technology: enjoyable but lacking nutrition.
The real barrier is adults' imaginations
The biggest obstacle to better education with computers is the completely impoverished imaginations of most adults (teachers, administrators, policymakers), who fail to envision or implement profound, idea-centered uses inspired by thinkers like Seymour Papert.
No revolution yet
Despite decades of promises (from the 1970s Dynabook vision onward), the educational technology "revolution" hasn't truly happened because focus stays on hardware/access rather than cultivating powerful ideas, contexts for growth, and teacher expertise in using computers as a new medium for thinking.
These critiques remain relevant, as Kay has consistently argued that computers could enable revolutionary learning (e.g., simulating, modeling, inventing knowledge), but only with visionary, human-centered approaches — not just more devices.
The only time I've had to do phone activation is for activating Terminal Service Licenses for Citrix clusters in enterprise. I wonder if that will also be affected or not.
My dad's a retired photographer. He told me that, back in the day, for a wedding the photographer got paid well, and the DJ barely got paid anything. But today, it's switched, the photographer barely gets paid anything, and the DJ demands a huge fee.
Also he told me that back in the day a live band was paid well, but now live bands barely find any work for weddings, and if they do barely get paid anything.
My dad's a retired photographer. He told me that, back in the day, for a wedding the photographer got paid well, and the DJ barely got paid anything. But today, it's switched, the photographer barely gets paid anything, and the DJ demands a huge fee.
It's important to know every "high level behavior" of what your code is doing.
For example, even if you write a simple "Hello World" app, there's many layers of processing. Even if you write your "Hello World" app in assembler, there's a ton of various OS subsystems and hardware firmwares and microcontrollers involved. But at the end of the day, you trust your OS and hardware, and understand that your "Hello World" app will most likely just output "Hello World", but there's a always possibility of rare quirks. For example if a solar flare happens when you run your Hello World app, and the solar flare interrupts a boolean logic operation, then your "Hello World" app crashes, which is generally accepted as it's rare.
But if you don't understand the "high level behavior", then that's a problem.
Claude AI Prompt:
For the attached code, justify and explain the code.
Remove unnecessary code, inconsistent code style, imaginary API usage, comments serving as LLM prompts, or other indications of AI-generated output.
(I teach CS in college and grad school.)
You can't practically give unlimited time. An unproctored test will see massive cheating on the test. So if you want the output of the test to somewhat resemble the skill of the student, you need a proctor. And that proctor needs to go home eventually.
By the constraint of my university, most of my test are 75 minutes long. The way my tests are built, you probably should not need more than 45 minutes to answer everything. Yet, I always have students staying the entire time that is available.
During exam week, we have to give a full 150 minutes. Usually A students leave within 30-50 minutes. B students tend to leave within 40-60 minutes. C students will stay about an hour and a half. And D and F student will stay the entire time staring at their exam without having written anything on it for the last hour and a half; but still they will stay another 30 minute.
So unlimited testing time is just adding expenses and inconvenience without benefiting students at all.
For me, as an adult, I found that taking vitamin D, and magnesium threonate really helps with my memory and speed. I remember a CS teacher of mine said that for me, I was the opposite of most other students. I would be the first student to quickly grasp algorithms, and make connections and such, but I needed to improve on my code organization and syntax. One job I had a coworker that was the other way around, as he had a dual major in CS and English. We worked great together. I would come up with elegant ideas, and he would help organize them.
I always struggled with timed tests. I had a lot of childhood trauma living in section 8 housing and my family being a victim of many various crimes at a very young age, so I've always had a ton of anxiety. I was that kid in the hallway still doing my advanced math test in the hallway as I needed 3 times the time to finish, but always got an A. I also was afraid of submitting any answer unless I checked it forwards, backwards, and thought of multiple ways of solving it. In my real job I'm an IT architect\engineer. I have worked on systems that affect 100s of millions of people. I also worked on hospital systems, where messing up IT could affect real people lives. You want someone like me who takes time to triple check before risking peoples lives, or causing millions of dollars in outages or data loss.
For example, the original version of Lil Nas X - Old Town Road, without Billy Ray Cyrus, is completely missing from the internet.
If connected to the internet, wouldn't the below stop 99% of these issues:
Auth1: restrict connections by source IPv4 or IPv6 of the vendor (if vendor managed). The customer could add their own HQ IPv4 or IPv6 source addresses.
Auth2: require a legit client certificate when connecting to them that's signed by the vendor (if vendor managed). The customer could also add their own client cert from their internal CA.
Auth3: require username\password
Only if all 3 auth layers pass can you connect.
I research various human cellular pathways and treatments as a hobby.
AI seems to not "piece" ideas together.
For example, let's say:
* paper #1 suggests that compound X activates pathway A
* paper #2 suggests that activation of pathway A will then also activate pathway B.
If I ask AI, what compounds activate pathway B, it is very unlikely to tell me compound X as a possibility.
(bringing together research from both papers)
Polymer physicists are into chains.
