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.

---

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 wonder if you could load the correct answer document as a source into NotebookLM, upload each student's tests as sources, and then compare each student to the answer key in one prompt and output one document.

I am not familiar with NotebookLM but Claude actually recommended I upload the zip file with all the student's answers and it would provide me with an excel spreadsheet with their names and answers

bthotugigem05 said:

---

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

---

Sounds cool and I would have loved this when I was working. For those in the MS ecosystem, they have an equivalent product called CoPilot Notebook.

I've used neither; just adding the latter for those ingrained in MS.

FatZilla said:

---

Basic code building and or code checking for vulnerabilities. My devs use it all the time.

I personally use it for creating basic 3d print model files that i then edit as needed but chatgpt can do hours of basic work in seconds with the right prompts

---

Do you use Claude? Or just pass code through ChatGPT?

All I do is Nguyen said:

---

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, 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

---

Having grown up in a different country, it's wild to see students get multiple choice on a SQL question. But great use !!

To be totally transparent I HATE having to manually grade that many short answer questions lol

bthotugigem05 said:

---

Just had a sales guy resign. He took copious notes and added to them every week, dating back to 2022. I uploaded them as a source into NotebookLM and can now have conversations with it about specific customer engagements covering that timespan. Absolutely unreal.

---

id like to obtain a copy of the Texags master database with all banned/deleted threads, load them into notebookLM and conversationally have it diagnose some of the head cases that have posted here for the last 26 years post crash.

There have truly been some weirdos drawn to this site who've used it for very twisted and cathartic purposes.

AggieChemE09 said:

---

I uploaded the layout of my local HEB. Then told it the path I typically walk through it.

Now I give it a few recipe links and it splits the ingredients out, and then sorts the list so that I check things off in the order that I come across them in the store. It will also put a tag on each item letting me know what recipe the item is associated with.

---

Finally bit the bullet and got a Claude subscription, it's been an enormous game changer for me

Same. I'm so glad I did it.

bam02 said:

---

We recently lost our house to a fire and are early in the claim process. I have been using ChatGPT to help me find things in my policy declarations. It's been very helpful for creating our contents lists and assigning values to items (this is an unbelievably daunting process and AI has really sped it up).

---

Could AI be fed a video of the home and create an itemized list of contents and fair market value from Marketplace listings?

bthotugigem05 said:

---

Finally bit the bullet and got a Claude subscription, it's been an enormous game changer for me

---

I am a heavy ChatGPT user and starting to experiment with Grok more. What do you like about Claude compared to these? I might have to check it out.

Wife has used grok for home design images. Basically cutting out the interior design industry.

Not sure but I'll try that. We still haven't really begun the contents part of the claim. It's so daunting.

YouBet said:

---

Mega Lops said:

---

bthotugigem05 said:

---

Just had a sales guy resign. He took copious notes and added to them every week, dating back to 2022. I uploaded them as a source into NotebookLM and can now have conversations with it about specific customer engagements covering that timespan. Absolutely unreal.

---

id like to obtain a copy of the Texags master database with all banned/deleted threads, load them into notebookLM and conversationally have it diagnose some of the head cases that have posted here for the last 26 years post crash.

There have truly been some weirdos drawn to this site who've used it for very twisted and cathartic purposes.

---

Dude, if we did this we could likely pinpoint whatever user name MFBarnes is now operating under although pretty sure I already know who that is.

---

While I support this use, it would also remove the thrill of the hunt.

I was gathering all my tax data before preparing my tax return. I made a donation to Goodwill of clothes my kids had outgrown over several years. I have three kids, two boys and a daughter. I just wrote down that I donated 4, 30-gallon trash bags of kids clothes. I guess at the time I thought I'd remember what was inside of those bags. Wrong.

I prompted Gemini asking it to generate a possible list of the contents of the bags of clothes outgrown by two boys and a girl of N ages over several years, including fair market values. It came out with a pretty good list. I tweaked it so that it more accurately represented what my kids actually wear. When I was done, I asked it to verify that the list could be reasonably be contained within that number of bags. I ended up with a donation of almost $800.

I'm at CPA and the list was pretty reasonable. My clients used to send me Goodwill list and they'd write down what they donated but not give any values. I can't tell you how much time I spent going out to Goodwill's website to try and determine fair market values for the items on the list. Gemini made this so much easier and faster.

Make sure to double check, triple check, quadruple check everything it outputs. These models hallucinate. And sometimes hallucinate bigly. They make it look perfect on paper but it's Dunning-Kruger on steroids. Incredibly useful and a big time saver for grunt work but absolutely not intelligent and not a replacement for a person going through everything.

Yep. I've gone through everything with a fine-tooth comb and fixed a few things. It's tight and ready for presentation.