News for hackers. Stuff that matters.
Re:A simulation of me: fine-tuning an LLM on 240k text messages by vunderba on Thursday January 04, 2024 @08:40PM (#38874683)
I’m far from the first to think of this. Several people — perhaps inspired by creepy Black Mirror episodes — have tried to fine-tune an LLM on their SMS or WhatsApp history in an effort to create a simulation of themselves.

It's a much older concept than Black Mirror. Ever since Markov chain IRC bots got popularized in the late 90s and early 2000s, people have been trying to train their virtual doppelgängers. I'm sure it won't be long before you can pay to have a post-Mortem ghost of yourself duplicated in a cemetery seance style MMO to perpetually live on and interact with other people who have passed on.

Re:A simulation of me: fine-tuning an LLM on 240k text messages by noduerme on Thursday January 04, 2024 @10:05PM (#38875232)
On the casino I used to run, I started a pilot program with a homemade poker bot (labeled as such, and only deployed on poker tables labeled as "bot friendly"). The bot had no set model of its own. It was designed to mimic specific players on the casino, regulars who had played 10,000+ hands and who agreed to have their history cloned, by ingesting their entire hand/betting history and looking for what they had done in similar scenarios as far as hole card value, probability, position, pot odds, table "heat" (a rolling average of over- or under-betting from other players in the previous hands based on what they showed if they showed down), etc.

I got a few good players to agree to let me take all their data and make poker clones of them. What was really funny was watching them get baffled and often beaten in heads-up play against their doppelgangers. The bot never had more information about the game than any other player. But its betting style varied wildly depending on who it was based on.

As a side, this also helped me build a real-time system to monitor potential collusion based on betting patterns that deviated from a player's normal behavior at times when they were sitting at the same table as someone else whose patterns deviated in their presence.

Re:A simulation of me: fine-tuning an LLM on 240k text messages by digdugdirk on Thursday January 04, 2024 @11:13PM (#38875620)
"On the casino I used to run"? That might be the most casual intro to what sounds like a fascinating corner of the internet I never experienced.

Do you have any other interesting stories or references to that time of your life?

Re:A simulation of me: fine-tuning an LLM on 240k text messages by noduerme on Friday January 05, 2024 @03:28AM (#38876960)
I do, and I don't mind showing you them if you're interested, but why do you care?

ah fuckit. SO yeah, I lived outside the US and ran a bitcoin casino for some years for non-US players, which was blocked to US IP ranges and required IDs to eliminate US customers (even though Bitcoin gambling still wasn't officially illegal at the time). My general idea was to make a casino for smart people who liked puzzles, so to that end I built games that would let people have an even-money chance if they played the game perfectly.

The coolest thing that came out of this was one of my players went on to build an escape room in the Netherlands that used one of my games as a puzzle to solve to get out of the room. Here's some stuff from that casino (it was active from about 2010 to 2013).

https://www.youtube.com/watch?v=lJV40hu9J38 https://youtu.be/rkRfSAtCYAw https://www.youtube.com/watch?v=gD4GqCAAsmo

Re:A simulation of me: fine-tuning an LLM on 240k text messages by lagniappe on Friday January 05, 2024 @11:13AM (#38880653)
Doug?
Re:A simulation of me: fine-tuning an LLM on 240k text messages by noduerme on Saturday January 06, 2024 @10:40PM (#38898200)
I am not named Doug. There was a player named Doug, if I recall correctly, who with his wife became the focus of a collusion investigation which he denied, turning into a major war where he was ultimately banned... don't know if that's what you're referring to, but that's what jumped to my mind.
Re:A simulation of me: fine-tuning an LLM on 240k text messages by lagniappe on Sunday January 07, 2024 @12:27AM (#38898744)
Small world :)
Re:A simulation of me: fine-tuning an LLM on 240k text messages by noduerme on Friday January 05, 2024 @07:21AM (#38878357)
I was the lone coder and artist for all those games, btw. And the whole platform. As well as running the casino daily and doing all tech and customer support by myself for three years, 24/7 while rolling out updates and new games. And it's not easy to roll out bug fixes when you're dealing with real money and lots of online players arguing in your chat rooms!

My advice to anyone is not to get into that industry, even though you can make any game you want and people love the games and the business model sounds way better on paper than selling your games through an app store... the civilized world has ganged up to make sure you won't succeed, or that if you do you'll either kick all your money back through licensing schemes or else be arrested. It's basically too good to be true and too profitable. You can't make money at it unless you're in the club.

Re:A simulation of me: fine-tuning an LLM on 240k text messages by quickthrower2 on Thursday January 04, 2024 @11:59PM (#38875865)
Half of the work of running a gambling site is making sure all your customers are losers. If they are consistently winning (cheating or not) you want to get rid of them. Unless you are purely running a "pool" of some kind (think Betfair).
Re:A simulation of me: fine-tuning an LLM on 240k text messages by qup on Friday January 05, 2024 @12:06AM (#38875903)
In poker, all the players are profitable to the house. I think it's up for debate whether it's good for poker rooms to get rid of winners, but if there was a benefit, it would be an indirect benefit, not a direct one.
Re:A simulation of me: fine-tuning an LLM on 240k text messages by quickthrower2 on Friday January 05, 2024 @12:12AM (#38875939)
It depends how prolific they are. Someone running a winning bot farm is taking more money from the losers than the house would be if the losers kept winning/losing against each other. This assumes that the house wants losers to win alot to stay addicted. If they just get beaten all the time they might quit sooner, they may also run out of money sooner.

A bit like how lottery tickets costing $1 will make you win $1, $2 etc. prizes so you buy another ticket and so increase the revenue (a percentage of which is profit), the same for the house. 2 losers winning money off each other all night means they both lose and gave lots of money to the house. A winner taking the 2 losers money in a fell swoop means a lot less money for the house.

Re:A simulation of me: fine-tuning an LLM on 240k text messages by noduerme on Friday January 05, 2024 @03:57AM (#38877095)
er, every hand played in poker is profitable for the house (except for the ones where all players fold before the flop, prior to which no rake is taken). Bots, etc. don't matter. The only reason to prevent bots and collusion and other forms of mischief is to make your poker room a good place for people to play.
Re:A simulation of me: fine-tuning an LLM on 240k text messages by kawzeg on Friday January 05, 2024 @06:20AM (#38877937)
I think the point the poster you're responding to was making is this: If people lose their money quickly to bots/collusion/cheating (or simply skilled players), they will quickly run out of money. If they play each other, they can play more hands, so more profit for the casino.

To exemplify the two extremes, assuming 10% rake: Player A has $10, and loses $1 every hand: after 10 hands, he's broke and $1 went to the casino. Player B and Player C are equally skilled and equally lucky. They keep playing $1 pots, trading back and forth. After twenty hands, both have still have $9 while the casino already earned $2. They could essentially keep going until they have given all their money to the casino.

Re:A simulation of me: fine-tuning an LLM on 240k text messages by noduerme on Friday January 05, 2024 @07:12AM (#38878304)
Well that is how it works. The casino provides the service of collecting players together to form a game, serving a fair platform for them to play on, ingesting and disbursing funds (which is no small matter), and then attempting to prevent them from cheating each other. That is the business model for poker, and it's extremely hard to make it profitable. If you factor in the time it takes to do all those things, as well as managing the software 24/7, it's a wonder anyone bothers to host it at all these days.

[edit] As an individual player, obviously, whether IRL casino or online your goal is to outpace the rake. You factor the rake into your expected losses and find a casino with a low rake (ours was actually 3%) and try to get away as a winner. But of course as a casino owner I love grinders and they're who you want day in and day out.

Re:A simulation of me: fine-tuning an LLM on 240k text messages by qup on Friday January 05, 2024 @04:33PM (#38885267)
3% rake is amazing. Where's that?
Re:A simulation of me: fine-tuning an LLM on 240k text messages by noduerme on Friday January 05, 2024 @03:21AM (#38876931)
If your games are transparent and verifiably fair, you don't need to "make sure" that half your customers lose. You just post the payout odds, and have a fat enough wallet to cover the volatility. There is no "work" involved in deciding who wins or loses.

That said: If you're making a book (for sportsbetting) then yes, you are trying to balance the difference rather than predict a winner. That's a totally different thing from running a casino.

Re:A simulation of me: fine-tuning an LLM on 240k text messages by dontupvoteme on Thursday January 04, 2024 @10:33PM (#38875408)
As a coarse bare minimum this is from Neuromancer.

I am not sure if Gibson found inspiration elsewear but this is pure flatline dixie.

Fundamentally I suppose it's naught but whispers from the beyond.

Re:A simulation of me: fine-tuning an LLM on 240k text messages by LightBug1 on Saturday January 06, 2024 @05:57PM (#38896437)
“Well, it feels like I am, kid, but I’m really just a bunch of ROM. It’s one of them, ah, philosophical questions, I guess...” The ugly laughter sensation rattled down Case’s spine. “But I ain’t likely to write you no poem, if you follow me. Your AI, it just might. But it ain’t no way human.”
Re:A simulation of me: fine-tuning an LLM on 240k text messages by skybrian on Thursday January 04, 2024 @10:00PM (#38875202)
If you want to include fictional antecedents, there is Dixie Flatline from Neuromancer.
Re:A simulation of me: fine-tuning an LLM on 240k text messages by layman51 on Thursday January 04, 2024 @10:16PM (#38875304)
There’s also Ubik by Philip K. Dick where the recently deceased are able to communicate with the living AND aside from that also advertise a product.
Re:A simulation of me: fine-tuning an LLM on 240k text messages by fouc on Thursday January 04, 2024 @11:45PM (#38875764)
I looked it up. Ubik was written earlier in 1969, than Neuromancer, which was written in 1984
Re:A simulation of me: fine-tuning an LLM on 240k text messages by skybrian on Friday January 05, 2024 @12:35PM (#38881899)
Haven't read Ubik. I'm under the impression it essentially happens by magic (fantasy)? Or is some fictional technology involved?
Re:A simulation of me: fine-tuning an LLM on 240k text messages by layman51 on Friday January 05, 2024 @07:14PM (#38886855)
It is a kind of cryogenic technology where the person is “half-alive”. But still I guess the book itself does have magical elements in it such as psychic powers.
Re:A simulation of me: fine-tuning an LLM on 240k text messages by HPMOR on Thursday January 04, 2024 @08:58PM (#38874814)
This actually seems like a really good startup idea.
Re:A simulation of me: fine-tuning an LLM on 240k text messages by spiritplumber on Thursday January 04, 2024 @10:00PM (#38875203)
It's literally an episode of Max Headroom. https://www.maxheadroom.com/index.php?title=Episode_ABC.2.2:...
Re:A simulation of me: fine-tuning an LLM on 240k text messages by fnorder on Thursday January 04, 2024 @09:30PM (#38875014)
I'm petty sure that Ray Kurzweil was working on this very thing when Google brought him on.
Re:A simulation of me: fine-tuning an LLM on 240k text messages by mojomark on Thursday January 04, 2024 @09:40PM (#38875074)
The documentary with him saving his father's writings and such was one of the saddest things I've seen - that moment has stuck with me for years.

I suppose the people who will be paying for such a service will be the loved ones left behind, not the one's dying. I wonder if the doppelgangers will start to talk to each other in the graveyard in the middle of the night?

Re:A simulation of me: fine-tuning an LLM on 240k text messages by kevindamm on Thursday January 04, 2024 @10:29PM (#38875381)
If two digitized ghosts are talking but no air gets vibrated, do they make a sound?
Re:A simulation of me: fine-tuning an LLM on 240k text messages by krainboltgreene on Thursday January 04, 2024 @09:19PM (#38874934)
This is definitely what silicon valley would consider a good startup idea.
Re:A simulation of me: fine-tuning an LLM on 240k text messages by Alpha3031 on Thursday January 04, 2024 @09:54PM (#38875164)
cripes does anybody remember Google People
Re:A simulation of me: fine-tuning an LLM on 240k text messages by Filligree on Friday January 05, 2024 @06:52AM (#38878176)
How could I possibly forget?
Re:A simulation of me: fine-tuning an LLM on 240k text messages by teitoklien on Thursday January 04, 2024 @09:00PM (#38874826)
> Pay to have a ghost of yourself duplicated in a cemetery.

I mean, they already do it at funerals [1]. I don’t see why not, would make cemetery a fun hangout place to talk to people from history.

[1](https://m.youtube.com/watch?v=s2FJRbRsBBY)

Re:A simulation of me: fine-tuning an LLM on 240k text messages by Minor49er on Thursday January 04, 2024 @09:24PM (#38874962)
Apparently this works by asking questions to someone while they are still alive. Then after the individual has passed, friends and family can ask questions to a video screen which will pick out the response that it feels is the most appropriate and play it. Though I'm sure we're not far off from a service that will do this through deepfake technology

Makes me think of the 2004 movie The Final Cut where people get memory implants that record their entire lives for the purpose of creating a highlight reel of their lives after they're gone

Re:A simulation of me: fine-tuning an LLM on 240k text messages by anjel on Thursday January 04, 2024 @09:33PM (#38875039)
See also: Tupak; Coachella
Re:A simulation of me: fine-tuning an LLM on 240k text messages by jacquesm on Friday January 05, 2024 @12:59AM (#38876185)
"Do me a favor, boy. This scam of yours, when it's over, you erase this god-damned thing.".
Re:A simulation of me: fine-tuning an LLM on 240k text messages by a_bonobo on Friday January 05, 2024 @02:26AM (#38876635)
There's this recent cool SF novel by the Polish (and undertranslated!!!) SF author Jacek Dukaj called 'The Old Axolotl'. In it, people used gaming systems to copy their 'beings' into machines before a planetwide wipeout event of biologic life. The machines, with the human simulacra within them, survive for hundreds of years. What's interesting about the books is that the machines feel human and have all the human emotions, but somewhat foreshadowing LLMs, they cannot learn like humans learn. The robot society is essentially stale, forever, nobody develops. (at least that's my interpretation as to why the robots fail in the book)
Re:A simulation of me: fine-tuning an LLM on 240k text messages by MattPalmer1086 on Friday January 05, 2024 @04:51AM (#38877354)
It's a great book, and totally agree the author is under translated!
Re:A simulation of me: fine-tuning an LLM on 240k text messages by kolinko on Friday January 05, 2024 @06:19AM (#38877929)
There were two Netflix miniseries related to the book, btw. Exploring the surviving humans on the planet.
Re:A simulation of me: fine-tuning an LLM on 240k text messages by MattPalmer1086 on Friday January 05, 2024 @07:54AM (#38878578)
Ah, if you mean Into The Night, I enjoyed that, although the link to the old axolotl wasnt entirely obvious!
Re:A simulation of me: fine-tuning an LLM on 240k text messages by kolinko on Wednesday January 10, 2024 @05:39PM (#38944462)
Yes! And there was a second mini series connected to this one!

Yakamoz S-245

Re:A simulation of me: fine-tuning an LLM on 240k text messages by bhoops on Friday January 05, 2024 @03:08AM (#38876854)
there was a post on linked in around this https://www.linkedin.com/pulse/one-needs-die-anymore-nishant...
Re:A simulation of me: fine-tuning an LLM on 240k text messages by skywal_l on Friday January 05, 2024 @04:04AM (#38877124)
See also `Revelation Space` series of sci-fi books by Alastair Reynolds. People are being simulated so their knowledge and wisdom can be tapped after their death.

[0] https://revelationspace.fandom.com/wiki/Gamma-level_simulati...

[1] https://revelationspace.fandom.com/wiki/Beta-level_simulatio...

[2] https://revelationspace.fandom.com/wiki/Alpha-level_simulati...

Re:A simulation of me: fine-tuning an LLM on 240k text messages by maxbond on Friday January 05, 2024 @04:09AM (#38877155)
A great short story in this universe on this topic is "Monkey Suit."