They can do it, it's just not "by default", they need to be prompted to do it. So at least the danger is manageable if you know what you're doing and how to prompt around it.
Not really. They're still non deterministic language predictors. Believing that a prompt is an effective way to actually control these machines' actual behavior is really far fetched.
They com like that from factory. Hardcoded to never say no.
They're not hardcoded to never say no, but some of the models were trained to be "yes men" because their creators thought it would be a good property to have. GPT-4o for example.
yeah that distinction is pretty important, and in general that guy I believe IS making the point - if you can not control it with guaranteed outcomes - you cannot control it.
You’re absolutely right! I do have insufficient data for a meaningful answer. This is not an *insightful prediction* — it’s *Dunning-Kruger masquerading as qualified intelligence*
This is one of those stories, just like the SR-71 "ground speed check" story, that every single time I see it posted I just have to read the entire thing again. I love it.
You better watch out. When my evil twin feels y'all aren't upvoting my posts enough he thinks "let's do a search for articles that have gotten 200+ votes at least 5 times in different years" [1] It's a highly effective strategy that I know dang doesn't like!
So I'll post another article about robot grippers which you should upvote instead of the breathless "AI will give us more Nobel Prize winning research" posts because: (1) robots that can change bedpans and pick strawberries really will change the world, and (2) they give out a certain number of Nobel Prizes a year and AI won't change that.
Man you should share your story. I got through a few Linux device driver labs but the more I read the less I understand. Even the keyboard driver or the tty driver are thousands of lines long.
I don’t know how people managed to write graphics card drivers back in the day. In the 80d it’s going to be all assembly code too, I think.
They are more black magic than the non-driver kernel components. I can at least understand the concept of kernel components such as VFS/Scheduler and read legacy kernel code without too much trouble, but drivers, even those in Linux 0.12 back in 1991, are crazily hard for me.
Once I discovered that the SR-71 Ground Speed Check is most likely not true, it doesn't hold the same weight for me anymore.
Way too many unlikely variables all lining up, and no other accounts of the story from all of the people (pilots, air traffic controller, etc) supposedly on the frequency.
An absolute classic! Was just telling a buddy about this one the other day while talking about The Egg by Andy Weir (another short story I really enjoy). Every time I read this one, I get chills at the end. Asimov really was a master.
It's amazing that in the late 1930's, someone with his academic credentials and intellect decided his life would be best spent writing science fiction.
I looked this up on Wikipedia. It seems that he was working as an instructor (not a professor) of chemistry; since he was making more money as a writer during that time, he slowed down or stopped his research. Doesn’t seem to have been an intentional choice so much as how things happened to turn out.
> he was working as an instructor (not a professor)
No he eventually became a full professor too.
"He began work in 1949 with a $5,000 salary(equivalent to $68,000 in 2025), maintaining this position for several years. By 1952, however, he was making more money as a writer than from the university, and he eventually stopped doing research, confining his university role to lecturing students.[g] In 1955, he was promoted to tenured associate professor. In December 1957, Asimov was dismissed from his teaching post, with effect from June 30, 1958, due to his lack of research. After a struggle over two years, he reached an agreement with the university that he would keep his title and give the opening lecture each year for a biochemistry class. On October 18, 1979, the university honored his writing by promoting him to full professor of biochemistry."
Yes that’s true, but I was referring to the line where it said he was making more money as a writer, which was before he became a tenured professor. In any case, we’re both addressing the point that he did have an academic career aside from writing.
I've read his biography. It was definitely intentional - and of course making a living by writing was a big factor. But he just didn't like the academic environment or his colleagues.
> He's best known for SF but he certainly didn't spend his whole career on it.
Indeed after becoming a giant of the field in the 1940s and 1950s, when he wrote most of the novels and short stories we know him for (Robots, Foundation and Empire) he took a long hiatus. In the 1960s and 70s, as far as I can tell, his meager sci-fi output consisted of some short stories, a couple of novelizations of sci-fi movies, and a standalone novel (The Gods Themselves).
After Sputnik he focused on science writing, believing that to be more widely useful.
He only returned to writing more Foundation, Robots, and Empire novels in the 1980s.
"This is by far my favorite story of all those I have written.
After all, I undertook to tell several trillion years of human history in the space of a short story and I leave it to you as to how well I succeeded. I also undertook another task, but I won't tell you what that was lest l spoil the story for you.
It is a curious fact that innumerable readers have asked me if I wrote this story. They seem never to remember the title of the story or (for sure) the author, except for the vague thought it might be me. But, of course, they never forget the story itself especially the ending. The idea seems to drown out everything -- and I'm satisfied that it should.
" - Isaac Asimov
Considering AC could persist indefinitely in hyperspace while interacting with normal matter, the answer would appear to be "hyperspace", whatever that is.
I remember the first time I heard this story. I was maybe 7 at a planetarium and they animated it with music little hand drawn starships and retro computers floating among the stars. They turned the stars all out for the final scene.
(It's a video game that does a brilliant job touching on similar themes to The Last Question. If you liked The Last Question and can fit a video game into your life, you will probably like Outer Wilds. Warning: if you start searching for "outer wilds," the algorithm will aggressively try to spoil you. Progression in the game is gated behind knowledge, so this is worse than usual. If you have trouble resisting the temptation to google past a rough description, it's a sign you should just jump in and play it. End recommendation.)
Great game, but if you get stuck for a long time, just look up some spoilers. Multiple times I abandoned the "right" approach to a problem because I couldn't get it to work and wasted countless hours trying to solve it the wrong way - only to find out I should have stuck to the right approach.
The game doesn't give any guidance, and wasting those hours is not rewarded.
The only other tip I'll give:
When you first play the game, spend the first 1-2 hours on your little planet learning everything (how to maneuver, how to use the signalscope, etc). Once you leave the planet, a timer will start. There is no way to "save" the game. You will die when the timer runs out. Don't panic. That's expected. Don't try to figure out what you did wrong to die - you will die no matter what. The game will restart, but anything you learned in the past will be in your computer's memory for retrieval.
OK, 2 more tips (one I wish someone had told me - I finished the game without it):
1. You can make time go by if you sleep at the fire.
2. There is a way to "meditate" until you die. This is very useful when you get stuck and can't get out of somewhere. To find out how to meditate, talk to the people on other planets (you may have to talk more than once before he teaches you).
This is a standalone game that needs to be purchased. For PC, it can be acquired through Steam (https://store.steampowered.com/app/753640/Outer_Wilds/). It is also available on consoles, it is not available on mobile. It is playable with keyboard and mouse, but it was primarily created with a game controller in mind.
At it's core, it's a game about exploration to understand what's happening. I recommend looking around and being curious to enjoy it, and avoid rushing. It's my favorite game.
To give you an estimate, I completed the base game with all secrets in about 20-30h. There's also a DLC called "Echoes of Eyes" adding a new area to explore. In total, I spent 45h to fully complete the game.
"Similar" is doing substantial work. If this is your only clue, it is likely to mislead you for at least 50% of the game, and I strongly suspect you will have fun anyway :)
For a while I thought I really liked sci fi novels and short stories, and maybe that's somewhat true. But I've started wondering if maybe I just liked Asimov's writing in particular. Other writers in the genre are more hit or miss. Can anyone recommend other writers that are on his level?
It wasn't until I discovered I was on the spectrum that I realized why it clicked so much. >.< I'm masking all the time, running conversational simulations to anticipate the societally-expected response to any given situation (and am high on the IQ spectrum).
Have you tried Arthur Clarke? I would say he is close to Asimov in many ways, being from the same time.
For others who share some similarities, though with a greater emphasis on character and adventure, perhaps Hal Clement, Larry Niven or Robert L. Forward.
As someone who loves the Big Three (Asimov, Clarke, Heinlein) and have read a lot of SF, I pretty much despise Bradbury. There’s no science in his science fiction.
Or even Star Trek to be honest. I don't know why Star Wars always gets passed off as "science fantasy" when it's a more grounded universe than Trek by far - space wizards notwithstanding (which Trek has plenty of.)
Even in a lot of hard SF, a lot of the science is wonky if it fall outside of the author's special interest or area of expertise. Relevant to Asimov, the only reason robots have "positronic" brains in his stories is that positrons were a new discovery at the time and it sounded cool and futuristic to him.
ted chiang if you haven't already. story of your life, exhalation, the lifecycle of software objects. same thing asimov does where the sci fi premise is really just a frame for a very human question. except chiang does it in like 30 pages and you feel it for a week
>> But I've started wondering if maybe I just liked Asimov's writing in particular.
A less commonly mentioned Asimov book that I really enjoyed and will read again is "The End of Eternity". If you've not read it, the ending is IMHO amazing and unique.
Last Question reminds me of it because of the style.
A Fire Upon The Deep is a fantastic novel for programmers to read, and I think the prequel A Deepness In The Sky is even better. There are some amazing old-school coding jokes in there, like that everyone thinks the universal time counter started at the first moon landing, but programmer archaeologists know it was really 15 megaseconds later.
Neal Stephenson's work is outstanding in my opinion, although some find it polarizing. My favorite of his is _Anathem_, followed closely by _Seveneves_.
Iain Banks's science fiction novels (mostly set in the Culture, but he does have others) are also great.
I wasn't expecting to find my favorite short-story on HN today! That's a pleasant surprise! This is how I started my journey in reading Isaac Asimov, I really recommend it!
A classic. It was dramatized by the Rochester NY, USA Museum of Science as a planetarium show, and I saw it there about 1974 with my father. Great times.
One of my all-time favorites. Almost every time I'm involved in a conversation about books, I always mention this. It amazes me how many people have never heard of it.
My favorite short story of all time. Between this and Deep Thought in HHGttG, I couldn’t believe the prescience when the bitter lesson was learned and LLMs and GPUs started eating the world.
the LLM parallel does hit different on this read multivac says insufficient data across ten trillion years and the whole story is basically if more compute and more data eventually gets you there. what's weird is the story answers yes, not on any timeframe that helps the people asking tho.
feels uncomfortably close to the actual situation where the models keep getting better and the answer keeps being "not yet, ask again later" while the answer is getting ready years late
I feel like the software running multivac represents something vastly more advanced than today's LLM.
I wonder if Asimov considered multivac to be an ancestor to his positronic robots, or if the two exist in different universes. I don't recall the two ever appearing in the same story.
I'm happy to see this short story posted here, it is one that I deeply loved when I was 14 or alike, and read it again multiple times. But I wonder: how did it survive in those sites without being shut down by the Asimov writings copyright holders? Given that the story is short and highly shared, it was just tolerated?
EDIT: actually I see that the link historically posted here more often is now dead: multivax.com/last_question.html
I was wondering the same. All the links to Asimov stories I've bookmarked in the past are now dead, so there probably is some enforcement of copyright.
the thing that gets me every reread is the structure of the joke. same question, asked across the entire lifespan of the universe, same answer every time. asimov could have made it tragic but instead it reads almost like a bit that keeps escalating and then the punchline is that the answer was always going to come, just on a timeline so absurd it laps back around to funny
Shouldn't the guy who runs this site be concerned about copyright infringement? Not sure to what extent the Asimov estate cracks down on unauthorized copies but he should be cautious.
In the 80s, our local planetarium did a show based on this story. The executive director of the museum associated with the planetarium had a very nice deep voice and was the perfect narrator, though it gave the Cosmic AC a slight Texas accent.
Somehow never read this one. But did write a short story ~20 years ago with a similar arc. I guess reading a lot of Asimov and Clarke and others will do that to you.
Color me surprised, when gemma-4 provided this answer: "Based on our current understanding of the universe, the short answer is no, it is not possible."
I love this story. When I first read it online in college many years ago I was surprised, and disappointed, when I suddenly realized it was a short story. It's a great one to recommend to people.
Outer Wilds, the video game, does a brilliant job expanding on this theme if you're hungry for more. "There's more to explore here."
Warning: progression is gated behind knowledge so spoilers are worse than usual and The Algorithm will aggressively try to spoil you if you start poking too deep into "outer wilds" searches. If you like The Last Question and can fit a game in your life, Outer Wilds is a solid bet.
Boy, it sure would be nice if real LLMs were capable of giving an answer like that.
They com like that from factory. Hardcoded to never say no.
It's not a guaranteed way to control their behavior, but you can more than move the needle.
So I'll post another article about robot grippers which you should upvote instead of the breathless "AI will give us more Nobel Prize winning research" posts because: (1) robots that can change bedpans and pick strawberries really will change the world, and (2) they give out a certain number of Nobel Prizes a year and AI won't change that.
[1] old issues of Byte magazine are a good bet: try https://archive.org/details/byte-magazine-1986-05
0: https://web.mit.edu/jemorris/humor/500-miles
https://users.cs.utah.edu/~elb/folklore/magic.html
I'm a bit proud of having suggested the author to add the 2019 entry (thanks to https://news.ycombinator.com/item?id=19798678).
Hopefully there's another repo of Internet stories somewhere else?
https://www.haiku-os.org/legacy-docs/benewsletter/Issue4-8.h...
I don’t know how people managed to write graphics card drivers back in the day. In the 80d it’s going to be all assembly code too, I think.
They are more black magic than the non-driver kernel components. I can at least understand the concept of kernel components such as VFS/Scheduler and read legacy kernel code without too much trouble, but drivers, even those in Linux 0.12 back in 1991, are crazily hard for me.
Way too many unlikely variables all lining up, and no other accounts of the story from all of the people (pilots, air traffic controller, etc) supposedly on the frequency.
A short anonymous joke that may or may not be true is better than a long story that is almost certainly made-up by someone in authority.
https://en.wikipedia.org/wiki/Isaac_Asimov#Education_and_car...
No he eventually became a full professor too.
"He began work in 1949 with a $5,000 salary(equivalent to $68,000 in 2025), maintaining this position for several years. By 1952, however, he was making more money as a writer than from the university, and he eventually stopped doing research, confining his university role to lecturing students.[g] In 1955, he was promoted to tenured associate professor. In December 1957, Asimov was dismissed from his teaching post, with effect from June 30, 1958, due to his lack of research. After a struggle over two years, he reached an agreement with the university that he would keep his title and give the opening lecture each year for a biochemistry class. On October 18, 1979, the university honored his writing by promoting him to full professor of biochemistry."
Indeed after becoming a giant of the field in the 1940s and 1950s, when he wrote most of the novels and short stories we know him for (Robots, Foundation and Empire) he took a long hiatus. In the 1960s and 70s, as far as I can tell, his meager sci-fi output consisted of some short stories, a couple of novelizations of sci-fi movies, and a standalone novel (The Gods Themselves).
After Sputnik he focused on science writing, believing that to be more widely useful.
He only returned to writing more Foundation, Robots, and Empire novels in the 1980s.
After all, I undertook to tell several trillion years of human history in the space of a short story and I leave it to you as to how well I succeeded. I also undertook another task, but I won't tell you what that was lest l spoil the story for you.
It is a curious fact that innumerable readers have asked me if I wrote this story. They seem never to remember the title of the story or (for sure) the author, except for the vague thought it might be me. But, of course, they never forget the story itself especially the ending. The idea seems to drown out everything -- and I'm satisfied that it should. " - Isaac Asimov
https://users.ece.cmu.edu/~gamvrosi/thelastq.html
Considering AC could persist indefinitely in hyperspace while interacting with normal matter, the answer would appear to be "hyperspace", whatever that is.
(It's a video game that does a brilliant job touching on similar themes to The Last Question. If you liked The Last Question and can fit a video game into your life, you will probably like Outer Wilds. Warning: if you start searching for "outer wilds," the algorithm will aggressively try to spoil you. Progression in the game is gated behind knowledge, so this is worse than usual. If you have trouble resisting the temptation to google past a rough description, it's a sign you should just jump in and play it. End recommendation.)
Great game, but if you get stuck for a long time, just look up some spoilers. Multiple times I abandoned the "right" approach to a problem because I couldn't get it to work and wasted countless hours trying to solve it the wrong way - only to find out I should have stuck to the right approach.
The game doesn't give any guidance, and wasting those hours is not rewarded.
The only other tip I'll give:
When you first play the game, spend the first 1-2 hours on your little planet learning everything (how to maneuver, how to use the signalscope, etc). Once you leave the planet, a timer will start. There is no way to "save" the game. You will die when the timer runs out. Don't panic. That's expected. Don't try to figure out what you did wrong to die - you will die no matter what. The game will restart, but anything you learned in the past will be in your computer's memory for retrieval.
OK, 2 more tips (one I wish someone had told me - I finished the game without it):
1. You can make time go by if you sleep at the fire.
2. There is a way to "meditate" until you die. This is very useful when you get stuck and can't get out of somewhere. To find out how to meditate, talk to the people on other planets (you may have to talk more than once before he teaches you).
That's all I'll say.
If so, please let us know so that other people do not get spoiled, and can you provide a link or links to the game that doesn't spoil it?
Thank you!
At it's core, it's a game about exploration to understand what's happening. I recommend looking around and being curious to enjoy it, and avoid rushing. It's my favorite game.
To give you an estimate, I completed the base game with all secrets in about 20-30h. There's also a DLC called "Echoes of Eyes" adding a new area to explore. In total, I spent 45h to fully complete the game.
There, I said it. The reason I say it openly is because I almost quit the game not understanding that this is supposed to happen.
Not really much of a spoiler.
It's on me for procrastinating playing the game for so long, it was bound to happen.
It wasn't until I discovered I was on the spectrum that I realized why it clicked so much. >.< I'm masking all the time, running conversational simulations to anticipate the societally-expected response to any given situation (and am high on the IQ spectrum).
https://web.archive.org/web/20140527121332/http://www.infini...
For others who share some similarities, though with a greater emphasis on character and adventure, perhaps Hal Clement, Larry Niven or Robert L. Forward.
You may have already read his story The Library of Babel: https://sites.evergreen.edu/politicalshakespeares/wp-content...
Even in a lot of hard SF, a lot of the science is wonky if it fall outside of the author's special interest or area of expertise. Relevant to Asimov, the only reason robots have "positronic" brains in his stories is that positrons were a new discovery at the time and it sounded cool and futuristic to him.
I also find C.J.Cherryh's books to be often quite interesting.
Asimov really did have a knack for clear, deceptively simple writing that isn't all that common.
A less commonly mentioned Asimov book that I really enjoyed and will read again is "The End of Eternity". If you've not read it, the ending is IMHO amazing and unique.
Last Question reminds me of it because of the style.
If you want good sci-fi a good list can be:
- Ender's Game
- The Martian + Project Hail Mary
- A Fire Upon the Deep
- Dune
Iain Banks's science fiction novels (mostly set in the Culture, but he does have others) are also great.
(I second Ender's Game, The Martian, and Project Hail Mary.)
They’re just too dry for my tastes.
The last question God might be for you If you’re super rational and are really into technology.
Belief in God is like a supermarket. Once you decide to enter you’re probably going to find something that works for you.
feels uncomfortably close to the actual situation where the models keep getting better and the answer keeps being "not yet, ask again later" while the answer is getting ready years late
I wonder if Asimov considered multivac to be an ancestor to his positronic robots, or if the two exist in different universes. I don't recall the two ever appearing in the same story.
God's numbering system is "unlucky".
EDIT: actually I see that the link historically posted here more often is now dead: multivax.com/last_question.html
didn't know about ooo, maybe because it's not available on namecheap!
If you go up one level, you can see this story is one entry in a great library of stuff:
https://hex.ooo/library/
I consider these other two also great stories that I must read every time:
I Don't Know, Timmy, Being God Is a Big Responsibility
https://qntm.org/responsibilit
Gorge
https://qntm.org/gorge
Warning: progression is gated behind knowledge so spoilers are worse than usual and The Algorithm will aggressively try to spoil you if you start poking too deep into "outer wilds" searches. If you like The Last Question and can fit a game in your life, Outer Wilds is a solid bet.