Not nearly as much of a resource burden as you imagine. Resources and energy are more limited by labor than availability. 100 billion robots is not substantially different than the current resources applied to cars.

TXTransplant said:

---

You can spend your entire academic career coming up with new theories to explain physics, chemistry, biology, thermodynamics, etc. You can get research money to do this (and it's relatively cheap to execute) and you can generate tons of papers. A lot of academic research moved in this direction the last 20-30 years...all you needed was a computer. No expensive lab equipment and supplies.

But every theory, if ever to be put into real practice, has to be tested in the real world. In chemistry and biology, this means doing experiments in the lab. With respect to engineering, it means building small, demonstration or pilot scale plants/equipment.

---

I think that you are using the word "theory" in a popular way, not a scientific way.

I think that you mean "conjecture" or "hypothesis".

In science, a theory is a rigorous explanation of observations and experimental results. It is based on evidence and makes predictions that are testable.

A conjecture or hypothesis is more of an idea that may eventually be the foundation for a theory, but to become a theory it is a very long road and may not even happen in the lifetime of whoever had the conjecture, It is more of a guess than anything else.

TXTransplant said:

---

This thread has gotten off-track, but you've not addressed the natural resources (energy, metals, chemicals, plastics, etc) on Earth that will be required to create 100 billion robots and the electricity needed to power them.

And even if AI computation is in space, you still have to build the computers/data centers (with resources here on Earth), not to mention the rockets and the fuel needed to launch them into orbit.

I know Musk wants to mine other planets for natural resources and has even suggested moving manufacturing there, but again, I don't think that is something I will see in my lifetime.

---

Definitely something we won't see in any of our lifetimes.

Think what it would take to set up mines on other planets. We would have to create entire new enterprises to launch the products back to Earth. It would be very expensive to set up and would be a very long term project.

For example, supposed that we found a huge load of tungsten on Mars. Even if we had the capability, it would cost a fortune to ship it back to Earth. The shipping costs per ton of tungsten would be enormous.

It is very likely, even, that humans sent to explore Mars would ever be able to to return.

It might make for great science fiction, though.

Can we take that discussion to a different thread? Would love to see this one distilled to how people are using it, regardless of whether or not they're nuking the environment or whatever

I just instructed it to "tell me a story from a rattlesnake's point of view of a rattlesnake hunting for a horse".

Here is part of the resulting story:

Quote:

---

With a burst of kinetic energy stored in my muscular spine, I launch. My head snaps forward, fangs extending toward the horse's lower leg, aiming for the softer skin just above the ankle joint. The impact is swift; my venomous bite pierces the hide, delivering a potent dose of neurotoxin directly into the bloodstream.

The horse shudders, a sudden gasp escaping its nostrils. Its eyes widen, and it lifts its hind leg, trying to shake off the pain. The venom spreads quickly, numbing the nerves and causing the muscle tissue to relax. The animal stumbles, its powerful legs faltering as the toxin takes hold. It rears briefly, snorting, then collapses back onto the dusty ground, breathing heavily.

I retreat a short distance, keeping low, watching the horse's struggle. The venom does its work, slowing the animal's heart and weakening its muscles. Within minutes, the horse's breathing becomes shallow, its massive chest rising and falling with increasing difficulty. The once mighty beast now lies vulnerable, its great bulk turning into a still, heavy mound on the prairie floor.

I approach cautiously, aware that the horse's size could attract scavengerscoyotes, ravens, perhaps even a curious human passing by. I coil around the animal's flank, positioning myself to protect my hardwon prize from other predators. My own metabolism is slow, but the sheer amount of protein and fat in a horse will sustain me for weeks, perhaps months, as I slowly digest the massive meal.

---

OP...

Your landscape redesign is a great example of how certain working groups will become obsolete. My son, senior in HS, was heavily focused on becoming an Architect. However, the 7-9 years it takes to get licensed and the potential of AI taking work away from the human Architects, resulted in him taking a dfierent approach...Civil Engineering with a Masters in Construction Management.

Anyone using NotebookLM at scale? We just got access to the enterprise version yesterday and it's blowing my mind.

I use it a lot for:

Cooking -- both recipes and helpful tips. Things like "when do I add the potatoes in to a roast in a slow cooker" or "how long should I saute asparagus tips for a pasta dish."

Itineraries -- Really useful here especially when navigating open/close times for places you want to visit and limiting the amount of driving.

Editing -- I use it every day to edit things I've written.

Answers -- to the many questions I have every day.

Spreadsheets -- I used it to create a couple of spreadsheets for tracking client projects.

I will say that the most basic tools -- Copilot and ChatGPT -- definitely still need human review and intervention.

Mucho austin said:

---

I have used it to do complex formulas in spreadsheets

---

The new Claude app integration is nice for Excel. Beats the hell outta Copilot.

At Christmas, I tried it for generating gift ideas for my wife, who is hard as hell to shop for. I had prompt it based on her persona and uploaded a bunch of pics of stuff she already owned to help build a style profile. It came back with a few helpful suggestion but mostly a bunch of stuff she already owned.

I'll chime in here. I am a high school teacher in VA, I teach programming (AP comp sci Principles, HTML, CSS, SQL, Python. Naturally all teachers warn away from AI for cheating purposes, but after reading this article (https://x.com/mattshumer_/status/2021256989876109403?s=46), it is a long read but well worth it, I am leaning INTO using AI.

Last week I had it create for me an assessment for my students based on everything we've learned so far (see attached screenshots for what it gave me). I have settled on Claude for my AI and outside of it not doing pictures or videos, it's exactly what I need.

The best part is after my students upload their tests to canvas, I'll download the zip file of their responses and have Claude grade it for me. This is going to be such a time saver for me; writing/grading tests, creating unit and lesson plans, etc

AI grading AI.

The MCQ isn't the issue, canvas checks that. It's the short answer questions I'm going to have it grade.

Hey I've done the grading for 8+ years now, I wanna see how AI handles it. How accurate is it? Are there any glitches or issues?

Earlier I was asking medical questions and it misspelled Walgreens to Walgarur