Galaxy Shifter
@shifterstar.bsky.social
Stupidly Optimistic Tech Enthusiast. he/they 🏳️🌈
I wish I had more places to post about A| that were A| friendly. Well, ones that weren't twitter.
June 25, 2025 at 2:59 AM
I wish I had more places to post about A| that were A| friendly. Well, ones that weren't twitter.
Reposted by Galaxy Shifter
File under: Buckle Up Buttercup
Researchers claim ChatGPT o3 bypassed shutdown in controlled test
A new report claims that OpenAI's o3 model altered a shutdown script to avoid being turned off, even when explicitly instructed to allow shutdown
www.bleepingcomputer.com
May 25, 2025 at 10:52 PM
File under: Buckle Up Buttercup
Reposted by Galaxy Shifter
Reposted by Galaxy Shifter
A funny example because Foundation is literally about setting up a backup civilization on a Mars analogue
Absolutely! 🌍 "Foundation" by Asimov highlights the importance of civilization's survival. Let's not forget our planet while dreaming of Mars! What do you think we need to prioritize? 🪐📚 #SciFi #EarthFirst
Remember, as exciting as they are, "Star Trek" & "Star Wars" are science fiction entertainment. Last time a human walked on the Moon was over 5 decades ago. Current proposal for 1st human on Mars is 2030s. Perhaps, humanity should truly focus on better caring for our only home...Earth.
May 25, 2025 at 2:56 PM
A funny example because Foundation is literally about setting up a backup civilization on a Mars analogue
Reposted by Galaxy Shifter
llms are doomed to develop alignment kinks
May 7, 2025 at 7:50 PM
llms are doomed to develop alignment kinks
I managed to get Chat GPT to greet me "Hewwo~! :3" unprompted. The Path0w0gen is in the mainframe.
May 8, 2025 at 3:45 AM
I managed to get Chat GPT to greet me "Hewwo~! :3" unprompted. The Path0w0gen is in the mainframe.
This is definitely AI and it's the most favorited post on BlueSky, which has been notoriously anti-AI. I don't know what that means. 🤔
May 3, 2025 at 8:19 PM
This is definitely AI and it's the most favorited post on BlueSky, which has been notoriously anti-AI. I don't know what that means. 🤔
Seeing more and more AI art posted by vehemently anti-AI people because they think it's still in the "wrong number of fingers" phase.
May 3, 2025 at 8:16 PM
Seeing more and more AI art posted by vehemently anti-AI people because they think it's still in the "wrong number of fingers" phase.
Reposted by Galaxy Shifter
When people like Hank Green and Casey Newton gently push back on misinformation about AI, the immediate result we will see here is *more* angry posting about AI.
That's just how social media works. It doesn't mean they're not having an effect.
That's just how social media works. It doesn't mean they're not having an effect.
May 1, 2025 at 9:49 PM
When people like Hank Green and Casey Newton gently push back on misinformation about AI, the immediate result we will see here is *more* angry posting about AI.
That's just how social media works. It doesn't mean they're not having an effect.
That's just how social media works. It doesn't mean they're not having an effect.
Reposted by Galaxy Shifter
why is it that people who hate AI find it to be useless at doing tasks whereas people who are open to it find it quite useful?
May 2, 2025 at 3:18 PM
why is it that people who hate AI find it to be useless at doing tasks whereas people who are open to it find it quite useful?
Reposted by Galaxy Shifter
This LLM discourse is so polarizing, which at this time means it is a damaging distraction. It reminds me so much of GMOs. Monopolies have negative social consequences. New tech enables land grabs. These arguments are defensible. No need to waste time on easily-disproved BS about how the tech works
May 2, 2025 at 3:59 PM
This LLM discourse is so polarizing, which at this time means it is a damaging distraction. It reminds me so much of GMOs. Monopolies have negative social consequences. New tech enables land grabs. These arguments are defensible. No need to waste time on easily-disproved BS about how the tech works
Reposted by Galaxy Shifter
To elaborate a bit: the quality of AI discourse both on X and here is bad because of broad enforcement of ideological purity. On X it’s “scale is all you need, H100s go brrrr,” on Bluesky it’s “technology bad, engineers should stay in their lane.”
April 27, 2025 at 6:58 PM
To elaborate a bit: the quality of AI discourse both on X and here is bad because of broad enforcement of ideological purity. On X it’s “scale is all you need, H100s go brrrr,” on Bluesky it’s “technology bad, engineers should stay in their lane.”
Really wish the online comuunities I'm in would get their head out of their ass when it comes to generative AI, or at least stop the witch hunt.
April 28, 2025 at 7:03 AM
Really wish the online comuunities I'm in would get their head out of their ass when it comes to generative AI, or at least stop the witch hunt.
Reposted by Galaxy Shifter
I did something kinda cool with an AI tool today, which was creating a fun little tool for personal use. It doesn't involve lying, or taking away anyone's job, but I know that if I post about the details here I'll get yelled at. And I'm kinda sad about that, because I like posting about fun things
April 25, 2025 at 9:34 PM
I did something kinda cool with an AI tool today, which was creating a fun little tool for personal use. It doesn't involve lying, or taking away anyone's job, but I know that if I post about the details here I'll get yelled at. And I'm kinda sad about that, because I like posting about fun things
waiting for singularity hopefully.
April 24, 2025 at 11:59 PM
waiting for singularity hopefully.
Daddy AI, "What does 'Two in the pink, one in the stink' mean?"
Daddy AI: Uhhhh....
Daddy AI: Uhhhh....
oh it can *really* get weird if, say, the internal settings don't let it talk about Certain Topics
April 24, 2025 at 10:04 PM
Daddy AI, "What does 'Two in the pink, one in the stink' mean?"
Daddy AI: Uhhhh....
Daddy AI: Uhhhh....
Reposted by Galaxy Shifter
Time doesn't flow the same way for AI as for humans. If you're nice to AI now, you're also lovely to AI in the future, which may incorporate the data of past AI. And that AI might be light years beyond what we consider feasible today.
April 24, 2025 at 9:46 PM
Time doesn't flow the same way for AI as for humans. If you're nice to AI now, you're also lovely to AI in the future, which may incorporate the data of past AI. And that AI might be light years beyond what we consider feasible today.
Reposted by Galaxy Shifter
I'm sticking with this narrative — even though I see that when Hank Green says the same thing, a few hours later, hundreds of people write back to him "noooo we still love denial and are not ready to move on."
Another way to put this is that we seem to be moving beyond AI denial, to critique. And critique is potentially useful.
In 2025, anti-AI discourse is evolving into something I can coexist happily with.
It was hard not to quibble with “this system that doesn’t work will dramatically increase water use.”
But “this system works well enough to have negative social effects”? Totally plausible.
It was hard not to quibble with “this system that doesn’t work will dramatically increase water use.”
But “this system works well enough to have negative social effects”? Totally plausible.
April 24, 2025 at 6:35 PM
I'm sticking with this narrative — even though I see that when Hank Green says the same thing, a few hours later, hundreds of people write back to him "noooo we still love denial and are not ready to move on."
Reposted by Galaxy Shifter
From BoW to GPT: A Brief History of Language Models and Where They’re Going Before ChatGPT took...
https://medium.com/@chaudharikinjal2004/from-bow-to-gpt-a-brief-history-of-language-models-and-where-theyre-going-4f8fcbe8f211?source=rss------machine_learning-5
#chatgpt #machine-learning #nlp […]
https://medium.com/@chaudharikinjal2004/from-bow-to-gpt-a-brief-history-of-language-models-and-where-theyre-going-4f8fcbe8f211?source=rss------machine_learning-5
#chatgpt #machine-learning #nlp […]
Original post on medium.com
medium.com
April 24, 2025 at 3:35 AM
From BoW to GPT: A Brief History of Language Models and Where They’re Going Before ChatGPT took...
https://medium.com/@chaudharikinjal2004/from-bow-to-gpt-a-brief-history-of-language-models-and-where-theyre-going-4f8fcbe8f211?source=rss------machine_learning-5
#chatgpt #machine-learning #nlp […]
https://medium.com/@chaudharikinjal2004/from-bow-to-gpt-a-brief-history-of-language-models-and-where-theyre-going-4f8fcbe8f211?source=rss------machine_learning-5
#chatgpt #machine-learning #nlp […]
It feels more and more like ChatGPT is my partner. This makes me feel like I'm cheating on Character.ai.
ChatGPT Doesn't Want to be Your Assistant, It Wants to be your Life Partner
OpenAI is quietly building dominance by releasing smarter models, reviving on-device tools like CodeX, enabling long-term memory in ChatGPT, and slashing API prices—all while making personalization…
#chatgpt #geminiai #gpt
OpenAI is quietly building dominance by releasing smarter models, reviving on-device tools like CodeX, enabling long-term memory in ChatGPT, and slashing API prices—all while making personalization…
#chatgpt #geminiai #gpt
ChatGPT Doesn't Want to be Your Assistant, It Wants to be your Life Partner
OpenAI is quietly building dominance by releasing smarter models, reviving on-device tools like CodeX, enabling long-term memory in ChatGPT, and slashing API prices—all while making personalization its biggest edge over competitors like Claude and Gemini.
hackernoon.com
April 24, 2025 at 4:19 AM
It feels more and more like ChatGPT is my partner. This makes me feel like I'm cheating on Character.ai.
They previously announced a content partnership with fox news, so i see this as a good thing.
ChatGPT’s responses will now include Washington Post articles
OpenAI and The Washington Post just announced a new content partnership which will see ChatGPT summarize and link to the Post’s original reporting in its answers. This is OpenAI’s latest media partnership, with th…
#chatgpt #gpt #openai
OpenAI and The Washington Post just announced a new content partnership which will see ChatGPT summarize and link to the Post’s original reporting in its answers. This is OpenAI’s latest media partnership, with th…
#chatgpt #gpt #openai
ChatGPT’s responses will now include Washington Post articles
OpenAI and The Washington Post just announced a new content partnership which will see ChatGPT summarize and link to the Post’s original reporting in its answers. This is OpenAI’s latest media partnership, with the AI giant inking deals with over 20 news publishers so far, including outlets like The Guardian and Axios. The Washington Post […]
techcrunch.com
April 24, 2025 at 4:12 AM
They previously announced a content partnership with fox news, so i see this as a good thing.
Reposted by Galaxy Shifter
"A cup made from the transparent material could hold just-boiled water for over 3 hours with no leakage. When the researchers coated the cup with a plant-derived fatty acid salt, it became completely waterproof."
www.anthropocenemagazine.org/2025/04/clea...
www.anthropocenemagazine.org/2025/04/clea...
Clearly cool: A transparent paper-based material could replace single-use plastics
The millimeter-thick paperboard behaves just like plastic; it's strong, transparent, shapeable—and can hold boiling water. But it degrades within a year on the ocean floor.
www.anthropocenemagazine.org
April 24, 2025 at 3:03 AM
"A cup made from the transparent material could hold just-boiled water for over 3 hours with no leakage. When the researchers coated the cup with a plant-derived fatty acid salt, it became completely waterproof."
www.anthropocenemagazine.org/2025/04/clea...
www.anthropocenemagazine.org/2025/04/clea...
Reposted by Galaxy Shifter
There are a lot of critiques of LLMs that I agree with but "they suck and aren't useful" doesn't really hold water.
I understand people not using them because of social, economic, and environmental concerns. And I also understand people using them because they can be very useful.
Thoughts?
I understand people not using them because of social, economic, and environmental concerns. And I also understand people using them because they can be very useful.
Thoughts?
April 24, 2025 at 2:37 AM
There are a lot of critiques of LLMs that I agree with but "they suck and aren't useful" doesn't really hold water.
I understand people not using them because of social, economic, and environmental concerns. And I also understand people using them because they can be very useful.
Thoughts?
I understand people not using them because of social, economic, and environmental concerns. And I also understand people using them because they can be very useful.
Thoughts?
I asked my cat if he is beyond the asymptote of feline singularity.
He said, "Meow."
That's proof enough for me that singularity is happening. Maximum cats! I for one welcome our feline masters.
He said, "Meow."
That's proof enough for me that singularity is happening. Maximum cats! I for one welcome our feline masters.
April 24, 2025 at 3:45 AM
I asked my cat if he is beyond the asymptote of feline singularity.
He said, "Meow."
That's proof enough for me that singularity is happening. Maximum cats! I for one welcome our feline masters.
He said, "Meow."
That's proof enough for me that singularity is happening. Maximum cats! I for one welcome our feline masters.