Yes, but that's a different issue. That would be illegal whether it was being used for AI training or not.

Copyright isn't even an issue. The word "use" has been thrown around so many times that many people have come to believe copyright law lets copyright owners control the use of their works. It doesn't. The law only applies to copying, distributing, and public performances. It says nothing about AI training. Maybe it SHOULD cover that. But Congress hasn't passed that law yet. This doesn't even require a Fair Use exemption. The works might have been illegally copied and distributed in order to assemble a training set. But that's a separate issue.

Brown paper bags can be recycled. When I asked by the at the recycling center if I could toss a huge stack of paper bags in with the boxes he chuckled and said, "You know what brown paper bags are? Just really thin cardboard!"

Here's a Computerphile video on what they mean by 'watermarking text', (probably): https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3F... But this only works if someone using a well-known LLM. Anyone who is using this tech for something nasty could just make their own specialized system (much smaller, cheaper) or modify an existing opensource model with additional training.

Whenever someone says they have a new radical engine design the first thing to look at is he surface area to volume ratio. Heat loss to the walls of the combustion chamber is a MAJOR source of inefficiency. It should be kept as low as possible.

After a quick glimpse of their renderings I'm very skeptical of their 60% efficiency claim.

Nuclear is quite capable of throttling up and down to match variations in load. But they usually aren't used in this way because reducing production reduces revenue, but doesn't reduce expenses. Any fiscally sensible utility will run their nuke plants balls-to-the-walls 24/365 and then match the load variation will variable-costs energy types.

...the renewables don't have enough storage to act as regular reliable and dispatchable power plants. The cost if fuel is a rounding error in the total costs of running a nuke plant. So saving a little U-235 while the sun is shining or wind is blowing doesn't reduce the amount that customers will ultimately pay... one way or another. And no one wants to pay for both.

Suppose you have a nuclear plant that can produce all the energy you need for 10c/kwh. Then a wind farm gets built that can displace 50% of the nuclear plant's production and do it for for 8c/kwh. What is the rate that customers pay? If you're thinking 9c/kwh, the proportionate average, then you're wrong. The nuke plant still needs the same revenue to stay in the black. So they sell half the electricity for twice as much per kwh. Thanks to the addition of "CHEAP!" renewables to the mix, the customers in this hypothetical scenario are now paying 14c/kwh. This will not be politically popular.

Fuel costs are a larger proportion of the total for coal and natural-gas powerplants. So the effect there isn't as bad. It's possible for exceptionally cheap wind and solar to match the fuel-only cost for these plants and actually not raise rates for customers. But no renewables operator would ever sell their electricity for less than the going rate.

Government should insist that weather-dependent renewables have some minimum amount of storage integral to their production.

Because shitty round-trip efficiency.

It's objectively true that the greenhouse effect is real and that with more CO2 you'll get more greenhouse effect and warmer temperatures. But whether this is at the level of "crisis" is entirely subjective. How can there be objective scientific evidence for or against that?

.

This is a nonsense article about a nonsense paper

I look up at the subject line and see 'thegaurdian.com' and think "Yep, sounds about right".

Switching to diet worked for me.

The anti-aspartame people have been linked to the sugar and HFCS industries. Seems only fair.

Aspartame has to be the most thoroughly studied compound there is. If there was a big enough problem to be significant on an individual level then it would be unambiguous by now.

Okay, but only if they enable short selling. (TFA article is paywalled. So I don't know if it mentions that)

If everyone involved only makes profit when the line goes up then the line will mostly go up. Short selling, sleazy as it is widely perceived, is necessary for good price discovery. The lack of good price discovery is why BTC is > 0.

At least not without scaling up by many orders of magnitude. Just look at how much memory even basic models need (gigabytes) and then look at how many qbits QC researchers are struggling to get working. Just... no.

Why are we even talking about these as part of the same conversation? Michio Kaku must have bills to pay.

At a given moment, you need as much as you need. You have the capacity or you don't. How does Dynamic Line Rating help you? The only use I can see for this is to allow highly variable sources (ie: renewables) export a little more and curtail a little less during favorable condition to eke out a little bit more revenue. This is not a solution to anything. Maybe it could be factored in with some sort of geographically fine-grained dynamic time-of-use pricing scheme to discourage consumption when the grid is nearing max capacity.

The grid is going to need huge upgrades anyway. Just building a system to proper size is the correct solution.

This is a bad idea for all the reasons already mentioned. A better idea is to request/require plain cardboard direct-shipable packaging from the factory. Specify the location and size of the identifying product labeling and barcodes such that they can be completely hidden by the shipping label.