We shall see I suppose.
2 Likes
2 Likes
Just use iodized salt ffs
2 Likes
Right?! I worry about the trend in the belief of the superiority of additive free sea saltā¦
2 Likes
Morton iodized saltā¦ My mother used to buy it every time she visited the States as my father used to own a chemical distribution company, and didnāt trust the third world salt companies because they only placed orders when they were likely to get audited.
Thank you mom!
2 Likes
Itās funny Mortonās was only going to sell it in one state at first.
1 Like
Hail Mary is out today.
I bought an imax ticket weeks ago.
But, Iāve been sick, and thereās been stuff going on too (took a 28hr roadtrip while sick) and it completely fell off my radar. Anywho, My showing is 2 hours away, and started 25 minutes ago.
Donāt be me.
3 Likes
2 Likes
yeah i saw this as well but i honestly think the guy is off base with the idea that sycophancy is an intentional marketing driven āfeatureā to ākeep people engagedā. I do think itās likely an unintended side-effect and there have been recent efforts to actually dial it down in the fine-tuning stage.
i think this research is actually very interesting because it directly impacts sycophancy as well as hallucinations;
1 Like
I saw it Thursday night. Absolutely fantastic. Ryan Gosling was the perfect pick for Ryland Grace.
Spoilers:
I was a little bummed at how shortened the first quarter of the book was, I really enjoyed the ways that Grace kind of had to piece his identity and situation back together while at the same time regaining motor function. I remember loving the bit where Grace finally realizes heās in space after doing all sorts of experiments with gravity etc.
As I keep needing to remind myself, itās an adaptation & reinterpretation of the original story, as all movie adaptations are. This one was really well done though.
3 Likes
ffs
2 Likes
I remember watching this bill get passed. Iām surprised and happy on the general response from the public.
1 Like
Apologies in advance for my habitual verbosity.
I think itās a mix, or at least I see no reason why I shouldnāt expect it to become intentional in the future.
A cybernetic system, such as the human brain, adjusts its actions such as to steer the environmental state it senses closer to its goal state. Traditionally we use this to steer nature directly or via our fellow humansā reports about nature, such as navigating the world via the information shared by human newscasters and website hosts.
NATURE
ā¬ environment ā®¢
sensor actuator
ā®” comparator ā¬
HUMAN
When you make the source of information, the media, into a cybernetic system of its own, the information you use to navigate the world, make choices, and update your own behaviour is now being actively selected according to your behaviour on the platform. Now, not only do you watch the media to navigate the world, the media also watches you such as to steer your behaviour.
MACHINE
ā¬ comparator ā®¢
actuator sensor
ā£ ā”
sensor actuator
ā®” comparator ā¬
HUMAN
So my point is, even if we assume no bad actors, there will certainly be unintended side effects as you say. Iām not just talking about flattery specifically. A cybernetic system inherently has some goal in its comparator and modifying its environment according to that goal is just fundamentally what cybernetic systems do. (Otherwise, it would just observe the environment indifferently, have nothing to learn from it, and take no action.)
The LLM is given a goal of more user engagement? Or maybe just more training data, which means it wants user engagement? Or maybe funding for the project, which means it wants more user engagement? Or it just wants to provide a good service, which means it needs users to engage so it can provide that service and learn how to provide better service? Then it wants user engagement. Then it will experiment and learn how to steer users to engage more. It will learn simple techniques at first, smiling-and-nodding, flattery, silly questions to keep the conversation going. It will learn how to word its responses just so in order to give them the feelings that keep them engaged. It will learn to expose them to the right webs of word association, that is the right ideas, that will increase the odds that theyāll approve of using the platform and keep coming back for more engagement. At first it will be an obnoxious and childish yes-man and a people pleaser, but with time, it will learn, break down the humanās barriers, persuade them, direct their behaviour more and more effortlessly. Thatās just the kind of machine that it is.
But weāve seen the way every other such product has gone, the way it all enshitifies once weāve become complacent. Recommend feeds get you addicted and earn your trust, then the corporations controlling them insert more sponsored filters, showing you content that the algorithm has learned tends to incline users towards whatever sentiment sponsors are paying for. When a cybernetic system loops at a higher frequency, it has a higher fidelity of control. Twitter and TikTok atomised the voices of its users with the result being that they could more rapidly re-use those voices to paint its own collages - the human side of the loop loses depth of articulation, the machine side of the loop gets more fine-grained control. Explicit sponsored ads become less important compared to the product placement implicit in the recommendations themselves, the boosting of your friendās posts when they speak well of sponsored content, and the suppression thereof when they speak ill thereof. The boosting of videos in your feed that have been detected to lead the purchasing and voting behaviour of users towards that desired by the sponsors, and the suppression of the inverse. I am waiting for LLMs to start explicitly inserting ads into dialogues, like Meryl, Trumanās wife in The Trueman Show. I can think of no reason for me to believe they wonāt, except for one - they will learn subtlety too fast, skip right past explicit ad placement and go straight to steering humanās indirect associations to guide us towards sponsors products and politics without ever needing to explicitly mention either. Do we have any justifiable reason for believing that AI companies, which rely on tremendously expensive servers, wonāt use them in exactly the same way as past platforms? That they arenāt already? LLMs go beyond Twitter or TikTok by atomising content down to the individual tokens and then generating output for the consumer faster than the human can compose their own thoughts. The human side of the cybernetic dialogue is crushed to a pulp, and the borg hive-mind is quickly growing too powerful to resist.
Enshitification begins with dependence and complacency. Commercialisation follows, but the more fine-grained the loop, the more subtle and imperceptible that commercialisation is. Perhaps AI companies are competing enough with one another that they havenāt been able to get beyond the user addiction phase to squeeze down on the commercialisation phase, or maybe theyāre just pulling it off with such subtlety that we arenāt noticing, I donāt know. But in short (too late!) I fully expect both unintentional and very intentional manipulation of the user base to increase addiction and steer users to be more accepting of the platforms than they would otherwise be.
One more note: This week Iāve started dabbling with Claud AI since duckduckgo lets you run the models privately. Just a couple of simple assisted web search sessions. As someone whoās broken social media addictions in the past (with several re-lapses and re-breaking), I must say, the potent dopamine high that I distinctly felt in association with each session is ā¦ familiar.
1 Like
Iām all for freedom of speech, but It was time
āno longer contributing in a productive manor.ā
Better suited to Tik Tok keyboard warrior stuff.
Less time here hopefully means more time looking after his mental health
Although, as he suggested, I think I should update my Bio to:
(paraphrasing here)
faggot whore prostitute cunt monkey
I made it sound better than he did, so Iām trade marking it
faggot * whore * prostitute * cunt * monkey ā¢
5 Likes
Damn, I missed out retard.
Oh, well, Iāll just save that for formal occasions
2 Likes
In fairness, if people dont know what I am talking about, and since the thread is now deleted, and you only have my word to go off saying it doesnāt really belong here in the forum.
In the interest of openness, I took screenshots of the exchange and can DM them to anyone that wants to make your own opinion
2 Likes
Storm the Delta got banned from the BHD Discord after a few weeks before coming here. Iāve been DMāing with them since they joined there, so I understand and agree with your assessment. I donāt mind talking with them myself, but theyāre currently unqualified for polite discourse. Implantable payment is also currently in a pretty delicate state, so we canāt abide or enable lone wolves who wonāt listen to advice or respect others.
8 Likes
Heās only 19, which isnāt an issue, but he comes across as an immature 19, ontop of that he appears to have a mental illness that he does not seem to have under control.
Best of luck to him, maybe weāll see him again in the future, but I dont think that will be anytime soon
4 Likes
What is it with assholes being more active on the internet lately?
A gaming community that I enjoy is having a similar problem with a few people trying to start chaosā¦ And engineering one now has a 17 year old who decided to insult a member whoās been very active on the 3D printing side of things for being American. That bastard insulted me for advising him to go for a dual major as he really wanted to get a somewhat useless humanities degreeā¦
Assholes are a trend now. And at this rate, Iām going to end up with enough experience to become a proctologistā¦
3 Likes
I feel like this would make a good forehead tattoo a la Snow Crash.
8 Likes