deegles

I guess someone with more knowledge about electric grids and cars will have to chime in.

I think the TCO would be impacted if enough people get electric in a city to require electric grid upgrades. That money will come from increased electric rates, higher rents, or in taxes of some form. Of course it's nearly impossible to predict ahead of time but it's a looming cost.

Imagine a 100 car parking lot for an apartment building, if everyone drives 20 miles per day average at an efficiency of 3.5kwh/mile, that's 7 megawatts that needs to be delivered every night or people can't get to work (assuming load balancing chargers). That seems like a lot. Multiply that by several buildings in a block plus the charging equipment for each parking spot... The money needs to come from somewhere.

the interaction time between the UV and virus particles is too short to deactivate them.

I think actually following through over time is the hard part. Wanting to do something is easy.

Would it be more efficient to use solar thermal for some of the desalination instead of $10b worth or regular solar?

A billion dollars worth of gold is about 15,000 kg. Though I wonder if the price would collapse as soon as it became clear large amounts of gold can be mined from space.

The paper just says "an estimated energy density of >1 kW/m^2"

The paper says a cubic meter of the material could generate more than a kilowatt (per hour?)... sounds promising if it actually gets commercialized.

I assume it would only work well in places where there are significant humidity swings, I didn't read the paper very carefully though.

https://onlinelibrary.wiley.com/doi/epdf/10.1002/adma.202300...

I got downvoted in another thread for suggesting that a company might do exactly this

They probably added a clause to their terms of service retroactively granting them permission to use your images for this purpose.

I think few people would disagree with the statement, but the real questions are how many grams of veggies per day? How many different veggie types per week? How do I know if it's enough to actually reduce my risk? etc.

But the issue remains where your questions have to have easy "retrieval" aspects. You can't ask general questions that a human would have no trouble with. A book report is the example I used because you need to have read the entire book. essentially a book-length context window.

Using RAG feels like asking an acquaintance to write a book report by giving them semi-randomly cut out paragraphs from the book.

Finetuning seems to be a solution, but there is still the issue of the LLM not really "learning" facts.

What's the latest on actually adding knowledge to an LLM?

Or you could just take a picture of the screen with your phone... employees don't need fancy new tools to exfiltrate data.

I installed a reverse osmosis water filter, switched to cast iron or stainless steel cookware, and use glass for food storage as much as I can. I wish there was an easy test for my level of exposure but I'm definitely aware.

They could get dump trucks full of cash within a week if they wanted it.

I just drove by a new giant highway overpass being built in Texas and I was wondering to myself "how many houses could have been built with all the money and resources sunk into this?"

All of Texas (30m) could live in a high density city the size of Tokyo. Which is huge, but not as huge as the rest of Texas.

This is a working memory exercise called "n-back". For true torture try "dual n-back" where you also have to match a tone, shape or color.

https://en.m.wikipedia.org/wiki/N-back

can this use speaker diarization?

The best nootropic is really high quality, regular sleep, regular exercise, and good diet. If you don't have the basics down, no supplement will make up for it.

Wish I could use this to convince my family to mask.

There was a device called the Dreem headband that would use EEG and specially timed tones to increase the duration of deep sleep (look up "sleep spindles"). Unfortunately that specific feature was blocked in the US and they took it off the market. But I hope that something like that will be launched again.

What's the best way to use an LLM with a large codebase that isn't RAG? Ideally we could have the full source in the context or already trained into the model... I was thinking I could set something to fine tune a model overnight and every morning I'd have a fresh one ready. Any ideas?

You could do it with a 4 channel mixer and USB audio devices hooked up to one Bluetooth transmitter... I don't know about latency though and good luck with the cable management!

Rolls MX51S Mini Mix 2 Four-Channel Mixer https://a.co/d/48XLqLi

USB to RCA https://a.co/d/iam01Xj

Bluetooth 5.0 Transmitter 3-in-1 https://a.co/d/47TuAMn

My gut feeling is that nothing will change meaningfully as long as employee promotions are dependent on increasing engagement (either directly or indirectly).

I believe that no single person sits down and says "let's make kids get addicted to our product," but they do run A/B tests or new features and keep the better performing variant. The end result of many of these cycles is the same.

Ultimately it's promotion-driven development.

I just watched Oppenheimer, and I remember thinking how this would be much better as an HBO style multi season drama. But there's no way they'd be able to sign actors consistently for the many seasons it would take to cover all the material... using AI actors however, they would be able to keep consistent characters even as the actors switch out. Of course it's a race to the bottom, but think of all the talented actors who can't work because they don't have the right "look" and didn't get a lucky break.

What about the impact of fast charging on power grids? Where I live there are power outages on hot days because of people running the AC, I can't imagine how it would hold up with a large percentage of houses charging a car. Sure, they could install a secondary battery pack to charge the car and avoid overloading the grid, but batteries would have to be really affordable.

I (naively) think a restaurant would have an easier time detecting people who are doing the bare minimum. The issue becomes how to "punish" freeloaders who are also owners? Imagine the nightmare scenario of a restaurant where every employee is a part of the LLC that owns it. "Firing" someone becomes an onerous legal process.

Is gumroad a good platform for selling software like this? How is licensing handled?