I also really donāt enjoy AI boom.
GPT-3 is a large language model that was released in 2020 by OpenAI and is capable of generating high-quality human-like text. [ā¦] An upgraded version called GPT-3.5 was used in ChatGPT, which later garnered attention for its detailed responses and articulate answers across many domains of knowledge.
Who wrote this? OpenAI marketing?
Do these people realise that itās a self-fulfilling prophecy? Social media posts are in the training data, so the more they write their spicy autocorrect fanfics, the higher the chances that such replies are generated by the slop machine.
He canāt seriously expert anyone to believe this at this point.
Iāve been wondering about this for a while. Do they really believe in this stuff or are they just so thoroughly out of ideas for āthe next thing that results in exponentially growing profitā that they just cling to it, while deep down knowing itās not actually real?
It doesnāt say anywhere in the article whether the memo also mentions why the workers would want thatā¦
Also,
ā60 hours a week is the sweet spot of productivity,ā
The fuck? That statement is so disconnected from my perceived reality that I have to wonder whether āproductivityā even means the same thing to these people as what it means to me.
Yep, the clarification doesnāt really clarify anything. If theyāre unable to write their terms of service in a way that a layperson in legal matters can understand the intended meaning, thatās a problem. And itās impossible for me to know whether their āclarificationā is true or not. Sorry, Mozilla, youāve made too many bad decisions already in the recent years, I donāt simply trust your word anymore. And, why didnāt they clarify it in the terms of service text itself?
That they published the ToS like that and nobody vetoed it internally, thatās a big problem too. I mean, did they expect people to not be shocked by what it says? Or did they expect nobody would read it?
Anyway, switching to LibreWolf on all machines now.
Sigh. Not long ago I switched from Vivaldi back to Firefox because it has better privacy-related add-ons. Since a while ago, on one machine as a test, Iāve been using LibreWolf, after I went down the rabbit hole of āhow do I configure Firefox for privacy, including that it doesnāt send stuff to Mozillaā and was appalled how difficult that is. Now with this latest bullshit from Mozillaā¦ guess Iāll switch everything over to LibreWolf now, or go back to Vivaldiā¦
Really hope theyāll leave Thunderbird alone with such crapā¦
I often wish I could just give up on web browsers entirely, but unfortunately thatās not practical.
What kind of total vampire would finance this ā¦ oh, itās YC. Yeah, makes sense.
The AI guys are really playing with the exact same cheat every time, arenāt they? Thanks to pivot-to-ai for continuing to shine a light on thisā¦ I hope the wider press eventually learns about it, too.
Yeah, thatās also something I found oddly missing (i.e. that replacing crypto systems world wide, if it becomes necessary, will take a very long time).
Some people act as if āI canāt run it now therefore itās garbageā which is just such a nonsense approach to any kind of theoretical work.
Agreed ā and I hope my parent post, where I said the presentation is interesting, was not interpreted as thinking that way. In a sibling post I pointed out the theme in there which I found insightful, but I certainly didnāt want to imply that theoretical work, even when purely theoretical, is bad or worthless.
Before clicking the link I thought you were going for aluminium, i.e. a variation of
Wow this is some real science, they even have graphs.
Thank you!
Oh wow, thank you for taking the time! :)
Just one question:
None of the other assorted proposals (loop quantum gravity, asymptotic safety, ā¦) got lucky like that.
Is this because the alternate proposals appeared unpromising, or have they simply not been explored enough yet?
Thanks for adding the extra context! As I said, I donāt have the necessary level of knowledge in physics (and also in cryptography) to have an informed opinion on these matters, so this is helpful. (Iāve wanted to get deeper in both topics for a long time, but life and everything has so far not allowed for it.)
About your last paragraph, do you by chance have any interesting links on ācriticism of the criticism of string theoryā? I wonder, because I have heard the argument āstring theory is non-falsifiable and weird, but itās pushed over competing theories by entrenched peopleā several times already over the years. Now I wonder, is that actually a serious position or just conspiracy/crank stuff?
Comparing quantum computing to time machines or faster-than-light travel is unfair.
I didnāt interpret the slides as an attack on quantum computing per se, but rather an attack on over-enthusiastic assertions of its near-future implications. If the likelihood of near-future QC breaking real-world cryptography is so extremely low, itās IMO okay to make a point by comparing it to things which are (probably) impossible. Itās an exaggeration of course, and as you point out the analogy isnāt correct in that way, but I still think it makes a good point.
What I find insightful about the comparison is that it puts the finger on a particular brain worm of the tech world: the unshakeable belief that every technical development will grow exponentially in its capabilities. So as soon as the most basic version of something is possible, it is believed that the most advanced forms of it will follow soon after. I think this belief was created because itās what actually happened with semiconductors, and of course the bold (in its day) prediction that was Mooreās law, and then later again, the growth of the internet.
And now this thinking is applied to everything all the time, including quantum computers (and, as I pointed to in my earlier post, AI), driven by hype, by FOMO, by the fear of āthis time I donāt want to be among those who didnāt recognize it earlyā. But there is no inherent reason why a development should necessarily follow such a trajectory. That doesnāt mean of course that itās impossible or wonāt get there eventually, just that it may take much more time.
So in that line of thought, I think itās ok to say āhey look everyone, we have very real actual problems in cryptography that need solving right now, and on the other hand hereās the actual state and development of QC which youāre all worrying about, but that stuff is so far away you might just as well worry about time machines, so please letās focus more on the actual problems of today.ā (thatās at least how I interpret the presentation).
Interesting slides: Peter Gutmann - Why Quantum Cryptanalysis is Bollocks
Since quantum computers are far outside my expertise, I didnāt realize how far-fetched it currently is to factor large numbers with quantum computers. I already knew itās not near-future stuff for practical attacks on e.g. real-world RSA keys, but I didnāt know itās still that theoretical. (Although of course I lack the knowledge to assess whether that presentation is correct in its claims.)
But also, while reading it, I kept thinking how many of the broader points it makes also apply to the AI hypeā¦ (for example, the unfounded belief that game-changing breakthroughs will happen soon).
Fun question to think about: when was the last time HP made a product that you would actually recommend to anyone?
On the left side within the text box thereās a sparkle emojiā¦ so I guess that means AI slop machine confirmed
More seriously though, Google Translate had odd and weird translation hiccups for a long time, even before the LLM hype. Very possible though that these days they have verschlimmbessert1 it with LLMs.
1 Just tried it, google translate doesnāt have a useful translation for the word, neither does DeepL. Disappointing. Luckily, there are always good old human-created dictionaries.