gerikson

joined 2 years ago
[–] gerikson@awful.systems 8 points 1 day ago

Wait until they find out it's not all iambic pentameter and Doric columns...

[–] gerikson@awful.systems 5 points 1 day ago

Translation is a good fit because generally the input is "bounded" and stays on the path of the original input. I'd much rather trust an ML system that translates a sentence or a paragraph than something that tries to summarize a longer text.

[–] gerikson@awful.systems 5 points 1 day ago* (last edited 1 day ago)

I enjoy the work for the 3 Macs from the British Isles:

  • Ken McLeod - Scotland: Fall Revolution series, Newton's Wake, Learning the World
  • Ian McDonald - Northern Ireland: Luna series, Brasyl. I'm currently on Hopeland
  • Paul McAuley - England: Quiet War series, Fairyland

In general I prefer UK English SF, because it's a bit less infected by the pernicious frontier mentality of US mainstream SF. Note that there are very good American authors too who kinda push back on that, but my impression was formed when Christopher Priest and Jerry Pournelle were active and could be contrasted.

[–] gerikson@awful.systems 5 points 1 day ago (2 children)

Hey, no kink-shaming.

[–] gerikson@awful.systems 5 points 2 days ago (2 children)

Ken McLeod’s The Cassini Division tells the fate of all uploaded superhumans - blasted to plasma by bombardment of comet nuclei

[–] gerikson@awful.systems 14 points 5 days ago (3 children)

This is classic labor busting. If the relatively expensive, hard-to-train and hard-to-recruit software engineers can be replaced by cheaper labor, of course employers will do so.

[–] gerikson@awful.systems 21 points 5 days ago* (last edited 5 days ago) (2 children)

A hackernews doesn't think that LLMs will replace software engineers, but they will replace structural engineers:

https://news.ycombinator.com/item?id=43317725

The irony is that most structural engineers are actually de jure professionals, and an easy way for them to both protect their jobs and ensure future buildings don't crumble to dust or are constructed without sprinkler systems is to simply ban LLMs from being used. No such protection exists for software engineers.

Edit the LW post under discussion makes a ton of good points, to the level of being worthy of posting to this forum, and then nails its colors to the mast with this idiocy

At some unknown point – probably in 2030s, possibly tomorrow (but likely not tomorrow) – someone will figure out a different approach to AI. Maybe a slight tweak to the LLM architecture, maybe a completely novel neurosymbolic approach. Maybe it will happen in a major AGI lab, maybe in some new startup. By default, everyone will die in <1 year after that.

Gotta reaffirm the dogma!

[–] gerikson@awful.systems 4 points 6 days ago

Why put in the work when you can ask Claude to summarize them for you and reap those sweet sweet internet points?

[–] gerikson@awful.systems 9 points 1 week ago (9 children)

A classic example of the "AI can't be dumb because humans are dumb too" trope, Pokemon Red edition:

https://www.lesswrong.com/posts/HyD3khBjnBhvsp8Gb/so-how-well-is-claude-playing-pokemon?commentId=muqLXFGXJLbdStaNm

[–] gerikson@awful.systems 6 points 1 week ago

Interesting to see if he gets one.

I believe Trump is A-OK with pay for play for pardons, but what's the actual price? Something flew by where people were buying a one-to-one with him for $5M, but that's basically "private". A pardon of someone as high-profile as SBF has to be worth the reputation hit. Can SBF and/or his family swing it? Would SBF be a good ally/toady of Trump?

Somehow I don't see it. Unlike UIrich, a lot of people lost real money when FTX imploded. There wasn't that much sympathy for him from crapto huggers. And let's not forget he's an autistic Jew, not a clear hero for the people who have Trump's ear.

[–] gerikson@awful.systems 9 points 1 week ago (2 children)

Worrying about a woke nanny AGI, and not the woke wirehead AGI (wireheading being a lot scarier).

This is very much the right-wing mainstream fear - not being able to generate Nazi memes with OpenAI

 

current difficulties

  1. Day 21 - Keypad Conundrum: 01h01m23s
  2. Day 17 - Chronospatial Computer: 44m39s
  3. Day 15 - Warehouse Woes: 30m00s
  4. Day 12 - Garden Groups: 17m42s
  5. Day 20 - Race Condition: 15m58s
  6. Day 14 - Restroom Redoubt: 15m48s
  7. Day 09 - Disk Fragmenter: 14m05s
  8. Day 16 - Reindeer Maze: 13m47s
  9. Day 22 - Monkey Market: 12m15s
  10. Day 13 - Claw Contraption: 11m04s
  11. Day 06 - Guard Gallivant: 08m53s
  12. Day 08 - Resonant Collinearity: 07m12s
  13. Day 11 - Plutonian Pebbles: 06m24s
  14. Day 18 - RAM Run: 05m55s
  15. Day 04 - Ceres Search: 05m41s
  16. Day 23 - LAN Party: 05m07s
  17. Day 02 - Red Nosed Reports: 04m42s
  18. Day 10 - Hoof It: 04m14s
  19. Day 07 - Bridge Repair: 03m47s
  20. Day 05 - Print Queue: 03m43s
  21. Day 03 - Mull It Over: 03m22s
  22. Day 19 - Linen Layout: 03m16s
  23. Day 01 - Historian Hysteria: 02m31s
 

Problem difficulty so far (up to day 16)

  1. Day 15 - Warehouse Woes: 30m00s
  2. Day 12 - Garden Groups: 17m42s
  3. Day 14 - Restroom Redoubt: 15m48s
  4. Day 09 - Disk Fragmenter: 14m05s
  5. Day 16 - Reindeer Maze: 13m47s
  6. Day 13 - Claw Contraption: 11m04s
  7. Day 06 - Guard Gallivant: 08m53s
  8. Day 08 - Resonant Collinearity: 07m12s
  9. Day 11 - Plutonian Pebbles: 06m24s
  10. Day 04 - Ceres Search: 05m41s
  11. Day 02 - Red Nosed Reports: 04m42s
  12. Day 10 - Hoof It: 04m14s
  13. Day 07 - Bridge Repair: 03m47s
  14. Day 05 - Print Queue: 03m43s
  15. Day 03 - Mull It Over: 03m22s
  16. Day 01 - Historian Hysteria: 02m31s
 

The previous thread has fallen off the front page, feel free to use this for discussions on current problems

Rules: no spoilers, use the handy dandy spoiler preset to mark discussions as spoilers

 

This season's showrunners are so lazy, just re-using the same old plots and antagonists.

 

“It is soulless. There is no personality to it. There is no voice. Read a bunch of dialogue in an AI generated story and all the dialogue reads the same. No character personality comes through,” she said. Generated text also tends to lack a strong sense of place, she’s observed; the settings of the stories are either overly-detailed for popular locations, or too vague, because large language models can’t imagine new worlds and can only draw from existing works that have been scraped into its training data.

 

The grifters in question:

Jeremie and Edouard Harris, the CEO and CTO of Gladstone respectively, have been briefing the U.S. government on the risks of AI since 2021. The duo, who are brothers [...]

Edouard's website: https://www.eharr.is/, and on LessWrong: https://www.lesswrong.com/users/edouard-harris

Jeremie's LinkedIn: https://www.linkedin.com/in/jeremieharris/

The company website: https://www.gladstone.ai/

 

HN reacts to a New Yorker piece on the "obscene energy demands of AI" with exactly the same arguments coiners use when confronted with the energy cost of blockchain - the product is valuable in of itself, demands for more energy will spur investment in energy generation, and what about the energy costs of painting oil on canvas, hmmmmmm??????

Maybe it's just my newness antennae needing calibrating, but I do feel the extreme energy requirements for what's arguably just a frivolous toy is gonna cause AI boosters big problems, especially as energy demands ramp up in the US in the warmer months. Expect the narrative to adjust to counter it.

view more: next ›