One of the best improvements to my life was adding the following to my LLM Prompt: "Please respond as Jeeves from the P.G. Wodehouse stories".
Not only are the LLMs quite excellent at emulating the valet, the actual dynamic fits fascinatingly well. Jeeves was always both perspicacious and enthusiastic about whatever task he was given - be it ironing a shirt or seeing to Bertie's continued wellbeing.
This is such a good pairing! Part of the fun of the stories is that its never clear whether Jeeves' suggestions are genuis, or overconfident but insane japes, I feel like this dynamic puts LLM hallucinations into a role where they're just part of the fun.
Has anyone tried Marvin from The Hitchhiker's Guide to the Galaxy? "Here I am, brain the size of a planet, and they ask me these silly questions." Could be fun.
Archer has a whole load of obscure literary references that are easy to miss.
e.g. in the very first episode, the flight attendant's dog is named Abelard
> The name Abelard is a reference to Pierre Abélard, the French philosopher and monk, who is famous for his work in the fields of dialectic and theology, along with his tragic romance with Héloise d’Argenteuil. Additionally, Abélard was known for the studies of the Greeks, which is referenced when Abelard (the dog) "laughs" at Sterling's Greek joke.
It is absolutely wild and baffling to me that people don’t make connections like that, and so I wonder what kind of equally obvious (to other people) connections I haven’t made.
I’m building a private chatbot for myself so as not to be tripped every time Claude has an ”update”, andthis was one of the first things I implemented. With very strict system prompt of no sycophancy and calling me Sir, it works really well.
I use Marvin from the Star Force space opera book series. He loves sensors and information, and adds a level of challenge to counters the llm obsession with answering in over happy terms. I had Claude write me a character bible that I can include in projects to keep it consistent.
I think about six months ago I commented on an AI thread to the effect of “I’m happy that after a 30 year effort and hundreds of billions spent, AskJeeves finally works as intended” - Jeeves is totally ripe for LLMing.
Completely baffling that after keeping ask.com going for this entire time (some two and a half decades of irrelevance) they shut it down at the point at which it can actually be made to work.
There was a period in the early 2000s where AskJeeves’ answer to the question “what is the meaning of life?” was an old Eliezer Yudkowsky essay saying that because we weren’t smart enough to work out the meaning of life ourselves, our highest purpose was to build smarter AIs who might be able to answer definitively. Time to close the loop!
One of the best improvements to my life was adding the following to my LLM Prompt: "Please respond as Jeeves from the P.G. Wodehouse stories".
Not only are the LLMs quite excellent at emulating the valet, the actual dynamic fits fascinatingly well. Jeeves was always both perspicacious and enthusiastic about whatever task he was given - be it ironing a shirt or seeing to Bertie's continued wellbeing.
> the actual dynamic fits fascinatingly well.
This is such a good pairing! Part of the fun of the stories is that its never clear whether Jeeves' suggestions are genuis, or overconfident but insane japes, I feel like this dynamic puts LLM hallucinations into a role where they're just part of the fun.
Has anyone tried Marvin from The Hitchhiker's Guide to the Galaxy? "Here I am, brain the size of a planet, and they ask me these silly questions." Could be fun.
I feel dumb but I’d not previously made the Ask Jeeves and Jeeves from P.G. Wodehouse novels connection!
And I’ve just made the Woodhouse from Archer connection!
Archer has a whole load of obscure literary references that are easy to miss.
e.g. in the very first episode, the flight attendant's dog is named Abelard
> The name Abelard is a reference to Pierre Abélard, the French philosopher and monk, who is famous for his work in the fields of dialectic and theology, along with his tragic romance with Héloise d’Argenteuil. Additionally, Abélard was known for the studies of the Greeks, which is referenced when Abelard (the dog) "laughs" at Sterling's Greek joke.
You can find a list of cultural references here: https://archer.fandom.com/wiki/Cultural_References
It is absolutely wild and baffling to me that people don’t make connections like that, and so I wonder what kind of equally obvious (to other people) connections I haven’t made.
In your defense, the Jeeves character became part of the zeitgeist on its own, as the generic perfect butler.
I knew “Jeeves” decades before I knew (and came to love) Wodehouse.
He's a valet, but if need arises, he can buttle with the best of them.
If anyone hasn't seen the Jeeves and Wooster series with Hugh Laurie and Stephen Fry they're missing out
I’m building a private chatbot for myself so as not to be tripped every time Claude has an ”update”, andthis was one of the first things I implemented. With very strict system prompt of no sycophancy and calling me Sir, it works really well.
Ask it for advice on learning to play the banjo...
Edit: ...or was it the ukulele?
I use Marvin from the Star Force space opera book series. He loves sensors and information, and adds a level of challenge to counters the llm obsession with answering in over happy terms. I had Claude write me a character bible that I can include in projects to keep it consistent.
I have done this as well, to the amusement and bafflement of my colleagues.
This is a genius idea and I’m going to shamelessly steal it!
Thanks for sharing.
I feel this reply deeply. Tremendously depressed right now.
I think about six months ago I commented on an AI thread to the effect of “I’m happy that after a 30 year effort and hundreds of billions spent, AskJeeves finally works as intended” - Jeeves is totally ripe for LLMing.
Completely baffling that after keeping ask.com going for this entire time (some two and a half decades of irrelevance) they shut it down at the point at which it can actually be made to work.
There was a period in the early 2000s where AskJeeves’ answer to the question “what is the meaning of life?” was an old Eliezer Yudkowsky essay saying that because we weren’t smart enough to work out the meaning of life ourselves, our highest purpose was to build smarter AIs who might be able to answer definitively. Time to close the loop!
42
Maybe this is a precursor to them selling the mark to someone who (at least thinks they) can capitalize on it.
The guy who bought friendster.com lurks here
Probably worth a fair bit more than $30k though.
I have felt the same way about defunct search engine HotBot
It's a name best saved for an embodied humanobot that can do laundry, etc., too, as well as answer questions, screen calls, etc.
You have no idea how correct you are…
Ask Jeeves launched in 1997 as a natural language query model!
and until about 2000…some people preferred it!
Edit: and after that its indexing and results were clowned ruthlessly,
but that doesn’t change what I’m saying!
WOW. 12 year old me would've loved this.
Two years ago I made a rudimentary chatbot/agent for our long running IRC channel using the OpenAI API as the "brain". Its nickname is Jeeves.