Llama 403b takes OOM a kilowatt minute to respond on our local gpu server, or about 10 grams of C02 per email. Last I checked, add another 20 grams of amortized manufacturing emissions. A typical commute is OOM 5-10 kg of CO2.
this article is alarmist bullshit. (for entirely unrelated reasons openai delenda est)
A thousand times? I’d have a hard time typing out that many queries in 8 hours. Even 100 seems like a stretch for someone who uses it within something like cursor.
More and more environments offer LLM aid without having you explicitly typing in a query. E.g. trigger inference whenever static analysis fails (e.g. on a compile error). Or trigger an LLM aided auto-complete with Ctrl-Space. I don't think it'll be particularly unusual to reach 1000 queries in a working day that way.
Coding models these days use an inference every time you stop typing. Let’s say it’s 0.1 inference oer keystroke. If you keep VSCode open all day, I could believe it’s a significant energy draw.
Google now uses several inferences per Google search.
The average user’s #inferences-per-day is going to skyrocket.
My point is that it’s understandable to consider AI a significant contributor to the average professional’s energy budget. It’s not an insult to point this out.
There's also radar reflection, "Observe that the received power drops with the fourth power of the range, so radar systems must cope with very large dynamic ranges in the receive signal processing." - https://www.eetimes.com/radar-basics-part-1/
I would also call it a classic as I learned it from this scene in Robert Heinlein's book "Rolling Stones": "The result was eight shiny right-angled corners facing among them in all possible directions —a radar reflector. ... The final result was to step up the effectiveness of radar from an inverse fourth-power law to an inverse square law — in theory, at least. In practice it would be somewhat less than perfectly efficient ...", https://archive.org/details/rollingstones0000robe_q7v9/page/...
A quick search on Google Scholar finds there's a fourth-power law for shock waves:
"A fourth-power law relating the stress jump through a steady structured shock wave and the maximum strain rate within the shock wave has received recognition as a unifying relation over a sensibly wide range of materials and shock compression amplitudes." - https://pubs.aip.org/aip/jap/article-abstract/107/1/013506/2...
"The third‐order nonlinear optical susceptibility χ(3) of this glass was found to be proportional to the fourth power of the radius of the colloid particles or the fourth power of the absorption coefficient at the peak of a plasmon band when the total volume of the colloid particles was constant."
One of the first things I tried with Claude Opus 3.5 was connecting it to ELIZA, and Claude did not like it one bit. After it hit me with
> I apologize Eliza, but I don't feel comfortable continuing this conversation pattern. While I respect the original Eliza program and what it aimed to do, simply reflecting my statements back to me as questions is not a meaningful form of dialogue for an AI like myself.
Did you notice that the issue was that O0P had failed to update the recovery address of their recovery address, and google removed access to both the main email and the recovery email at the same time?
Some people are confused why this could be a big deal. An analogy: on GitHub, if you echo a GitHub access token in an action’s log, it will be automatically censored. This post would be like noticing that someone’s action step is just named ghp_1ae27h… and that the name isn’t censored, and speculating on what that says about the token-censorship algorithm
Key point: if you try it yourself, it will be in clear-text for you (you already know your password, so there's no issue), but everyone else will only see "***".
We ended up deciding, against all seeming reason, to just stop after nuking each other twice. Perhaps our presidents and dictators can get on the phone and reach a similar agreement. I would quite like them to.
I’m always curious if this affects the content of the messages or just the tone. We have constructive proof that a voice in your head saying “Take your son to the top of a mountain, tie him to a rock and threaten him with a knife” can be viewed as morally good and casually powerful without changing the content at all.
In countries with women’s and children’s rights, the space of acceptable shamanic behavior is suddenly a lot narrower
Oh man, the binding of Isaac really is the strangest cognitive dissonance. I have no doubt that every practicing Christian, Muslim, and Jew I know [1] would ascertain that Abraham is having a serious mental health crisis and dual 911 if it happened to somebody they know today but at the same time I've never seen a good argument for why the actual story in Genesis should be taken seriously on its own.
[1] obviously there will always be outliers here but if we're exclusively considering 21st century America, I feel confident wagering that the amount of religious people who would take Abraham seriously is statistically insignificant, and similar to the amount of nonreligious people who would take him seriously. Also I'm specifically referring to how third parties react to Abraham nearly murdering his son, I don't think it's fair to hold Abraham himself to the same standards as long as we're assuming that he is suffering a mental health crisis.
Saying "raise your son to be patriotic and fight for his country" is acceptable in western societies. There isn't particularly a big difference in morality when it comes to the extremes
> Saying "raise your son to be patriotic and fight for his country" is acceptable in western societies.
Maybe kinda in some circles? I suspect almost none of the tech-worker urban liberals who are the majority on HN would embrace that as a parenting notion. Even among military veterans, your phrasing would likely sound a bit off the mark. They would generally see the call as being about protecting other people and serving their country.
To address your second sentence, most people see a big difference in morality between "Protect the innocent from people who would harm them" and "Harm the innocent".
this article is alarmist bullshit. (for entirely unrelated reasons openai delenda est)
reply