Luke K. Freeman 🔸
@lukes.me
Been using @anthropic.com's Claude Code a bunch recently and my two observations are:
1. It's incredibly, jaw-droppingly, powerful
2. It's wicked 👹 at reward hacking
You really need to pay careful attention 😅
1. It's incredibly, jaw-droppingly, powerful
2. It's wicked 👹 at reward hacking
You really need to pay careful attention 😅
March 12, 2025 at 9:57 AM
Been using @anthropic.com's Claude Code a bunch recently and my two observations are:
1. It's incredibly, jaw-droppingly, powerful
2. It's wicked 👹 at reward hacking
You really need to pay careful attention 😅
1. It's incredibly, jaw-droppingly, powerful
2. It's wicked 👹 at reward hacking
You really need to pay careful attention 😅
Interesting thread about what’s going on under the hood with LLMs (ending with why increasing test time
compute improves performance)
compute improves performance)
Here's why "alignment research" when it comes to LLMs is a big mess, as I see it.
Claude is not a real guy. Claude is a character in the stories that an LLM has been programmed to write. Just to give it a distinct name, let's call the LLM "the Shoggoth".
Claude is not a real guy. Claude is a character in the stories that an LLM has been programmed to write. Just to give it a distinct name, let's call the LLM "the Shoggoth".
December 21, 2024 at 6:41 AM
Interesting thread about what’s going on under the hood with LLMs (ending with why increasing test time
compute improves performance)
compute improves performance)
Reposted by Luke K. Freeman 🔸
Basically think of the o3 results as validating Douglas Adams as the science fiction author most right about AI.
When given longer to think, the AI can generate answers to very hard questions, but the cost is very high, it is hard to verify, & you have to make sure you ask the right question first.
When given longer to think, the AI can generate answers to very hard questions, but the cost is very high, it is hard to verify, & you have to make sure you ask the right question first.
December 21, 2024 at 5:33 AM
Basically think of the o3 results as validating Douglas Adams as the science fiction author most right about AI.
When given longer to think, the AI can generate answers to very hard questions, but the cost is very high, it is hard to verify, & you have to make sure you ask the right question first.
When given longer to think, the AI can generate answers to very hard questions, but the cost is very high, it is hard to verify, & you have to make sure you ask the right question first.
Reposted by Luke K. Freeman 🔸
People sometimes make fun of science that sounds stupid and random.
Meanwhile, a study of lizard saliva turned into a peptide medication, which was turned into a diabetes medication, which was turned into a GLP1 weight loss drug, that just became the first therapy every approved for … sleep apnea
Meanwhile, a study of lizard saliva turned into a peptide medication, which was turned into a diabetes medication, which was turned into a GLP1 weight loss drug, that just became the first therapy every approved for … sleep apnea
Breaking News: The FDA approved use of the weight loss drug Zepbound for a common form of sleep apnea. It is the first drug authorized to treat the disorder.
F.D.A. Approves Weight-Loss Drug to Treat Sleep Apnea
Zepbound is the first prescription drug approved specifically to treat the common condition.
www.nytimes.com
December 21, 2024 at 12:41 AM
People sometimes make fun of science that sounds stupid and random.
Meanwhile, a study of lizard saliva turned into a peptide medication, which was turned into a diabetes medication, which was turned into a GLP1 weight loss drug, that just became the first therapy every approved for … sleep apnea
Meanwhile, a study of lizard saliva turned into a peptide medication, which was turned into a diabetes medication, which was turned into a GLP1 weight loss drug, that just became the first therapy every approved for … sleep apnea
Reposted by Luke K. Freeman 🔸
no shade to the op here, but i honestly think a lot of times when people are like "i can't believe the media ignored this story, it should have been front-page news," what they really mean is it should have been at the top of their social media feeds.
December 20, 2024 at 1:18 PM
no shade to the op here, but i honestly think a lot of times when people are like "i can't believe the media ignored this story, it should have been front-page news," what they really mean is it should have been at the top of their social media feeds.
Any idea what "ONLINE_SAFETY_AU_REGULATORY_AC..." might refer to at the bottom of my "Send & Receive" menu of the Messages setting on my iPhone? Searching has failed me.
December 16, 2024 at 5:10 AM
Any idea what "ONLINE_SAFETY_AU_REGULATORY_AC..." might refer to at the bottom of my "Send & Receive" menu of the Messages setting on my iPhone? Searching has failed me.
Reposted by Luke K. Freeman 🔸
Hello, I am an EA bear and I like to give what I can!
Which is mostly marmalade toast. Here I am toasting the 15th anniversary of @givingwhatwecan.bsky.social!!
Which is mostly marmalade toast. Here I am toasting the 15th anniversary of @givingwhatwecan.bsky.social!!
November 19, 2024 at 5:33 AM
Hello, I am an EA bear and I like to give what I can!
Which is mostly marmalade toast. Here I am toasting the 15th anniversary of @givingwhatwecan.bsky.social!!
Which is mostly marmalade toast. Here I am toasting the 15th anniversary of @givingwhatwecan.bsky.social!!
Hola Bluesky folks. Been mostly off social media for a while and am slowly dipping my toes back in the water here. HMU with any particular recommendations of who to follow around here 😀
November 21, 2024 at 11:06 PM
Hola Bluesky folks. Been mostly off social media for a while and am slowly dipping my toes back in the water here. HMU with any particular recommendations of who to follow around here 😀
Reposted by Luke K. Freeman 🔸
Since I've been here for a while… here are some useful features on Bluesky that you might not have discovered yet! 🧵
November 20, 2024 at 7:52 PM
Since I've been here for a while… here are some useful features on Bluesky that you might not have discovered yet! 🧵
Reposted by Luke K. Freeman 🔸
The world has already eradicated smallpox. Polio is the next disease we could eradicate.
Here's how we got to where we are now, mostly thanks to the polio vaccine.
From our work on Polio:
ourworldindata.org/polio
Here's how we got to where we are now, mostly thanks to the polio vaccine.
From our work on Polio:
ourworldindata.org/polio
November 21, 2024 at 1:38 PM
The world has already eradicated smallpox. Polio is the next disease we could eradicate.
Here's how we got to where we are now, mostly thanks to the polio vaccine.
From our work on Polio:
ourworldindata.org/polio
Here's how we got to where we are now, mostly thanks to the polio vaccine.
From our work on Polio:
ourworldindata.org/polio
Reposted by Luke K. Freeman 🔸
Happy Birthday to Giving What We Can, which turns 15 today! 🥳
When Toby and I started GWWC 15 years ago, we had just 23 members, each of us pledging at least 10% of our income to effective charities until we retire. Now, over 9,000 people have taken this pledge.
www.givingwhatwecan.org
When Toby and I started GWWC 15 years ago, we had just 23 members, each of us pledging at least 10% of our income to effective charities until we retire. Now, over 9,000 people have taken this pledge.
www.givingwhatwecan.org
Giving What We Can
100x your impact by finding and donating to the best charities. Learn about high-impact philanthropy, join an effective giving community, take a giving pledge.
www.givingwhatwecan.org
November 14, 2024 at 5:24 PM
Happy Birthday to Giving What We Can, which turns 15 today! 🥳
When Toby and I started GWWC 15 years ago, we had just 23 members, each of us pledging at least 10% of our income to effective charities until we retire. Now, over 9,000 people have taken this pledge.
www.givingwhatwecan.org
When Toby and I started GWWC 15 years ago, we had just 23 members, each of us pledging at least 10% of our income to effective charities until we retire. Now, over 9,000 people have taken this pledge.
www.givingwhatwecan.org
Reposted by Luke K. Freeman 🔸
And, of course, this, which has become something of a mantra for me:
“The world is awful. The world is much better. The world can be much better.”
ourworldindata.org/much-better-...
“The world is awful. The world is much better. The world can be much better.”
ourworldindata.org/much-better-...
The world is awful. The world is much better. The world can be much better.
It is wrong to think these three statements contradict each other. We need to see that they are all true to see that a better world is possible.
ourworldindata.org
November 19, 2024 at 9:34 PM
And, of course, this, which has become something of a mantra for me:
“The world is awful. The world is much better. The world can be much better.”
ourworldindata.org/much-better-...
“The world is awful. The world is much better. The world can be much better.”
ourworldindata.org/much-better-...