- roywigginsDon't love how ChatGPT the readme is, the bullet points under "Why AIsbom?" are very, very ChatGPT.
- that is unfortunately less true that you might think for some students:
https://www.propublica.org/article/garrison-school-illinois-...
https://www.propublica.org/article/shrub-oak-school-autism-n...
https://autisticadvocacy.org/actioncenter/issues/school/clim...
https://www.the74million.org/article/trump-officials-autism-...
"Selected Cases of Death and Abuse at Public and Private Schools and Treatment Centers"
https://www.gao.gov/assets/gao-09-719t.pdf
> Death ruled a homicide but grand jury did not indict teacher. Teacher currently teaches in Virginia
https://www.nbcwashington.com/news/local/area-special-ed-tea...
- The good cops, such as they are, get run out if they try to challenge the institutional problems in police forces. This radically restricts how good a cop can be while still being a cop.
Can good cops speak up about bad cops and keep their job, or do they have to remain silent? How many bad things can you see in your workplace without quitting or whistleblowing while still being a decent person? Can they opt out of illegal but defacto ticket quotas and still have a career? Does writing a few extra tickets so you can stay in the force long enough to maybe change it make you part of the problem?
Many people look at the problems in policing and say that anyone working inside that system simply must have compromised themselves to stay in.
- And when a cop tries to do something about it, this is the sort of thing that happens. This guy seems like he's trying to do the right thing, but the system is designed so he can't:
https://www.theguardian.com/us-news/2023/jun/04/nypd-lawsuit...
> Bianchi claims his superiors retaliated against him for his stance against the “corrupt” cards after he was warned by an official with the Police Benevolent Association, New York City’s largest police union, that he would not be protected by his union if he wrote tickets for people with cards. And if he continued, he’d be reassigned... The lawsuit cites several instances where his NYPD colleagues complained about his ticket-writing, including on Facebook...
> Bianchi’s service as a traffic cop ended last summer when he wrote a ticket to a friend of the NYPD’s highest-ranking uniformed officer, Chief Jeffrey Maddrey, the lawsuit states.
- Possibly it was using WPS keyed to the ferry's WiFi and didn't consider the possibility that the ferry itself could move out of where it was registered (ie, the port)?
- Cells don't survive sequencing, any that you do implant have not been sequenced. At best you can get some confidence that the error rate is small.
- The question is more whether LLMs can accurately report their internal operations, not whether any of that counts as "thinking."
Simple algorithms can, eg, be designed to report whether they hit an exceptional case and activated a different set of operations than usual.
- > In much the same way as any human with modest experience in adding two digit numbers doesn't necessarily sit there and do the full elementary school addition algorithm but jumps to the correct answer in fewer steps by virtue of just having a very trained neural net.
Right, which is strictly worse than humans are at reporting how they solve these sorts of problems. Humans can tell you whether they did the elementary school addition algorithm or not. It seems like Claude actually doesn't know, in the same way humans don't really know how they can balance on two legs, it's just too baked into the structure of their cognition to be able to introspect it. But stuff like "adding two-digit numbers" is usually straightforwardly introspectable for humans, even if it's just "oh, I just vibed it" vs "I mentally added the digits and carried the one"- humans can mostly report which it was.
Here's Anthropic's research:
https://www.anthropic.com/research/tracing-thoughts-language...
- First you need a gene therapy delivery system that doesn't produce any off-target mutations at all, ever.
- Whatever engineered solution could happen, it will almost certainly have more side effects than a diet that includes vitamin C, and even if not, cost way more.
- https://www.anthropic.com/research/tracing-thoughts-language...
> Claude, on occasion, will give a plausible-sounding argument designed to agree with the user rather than to follow logical steps. We show this by asking it for help on a hard math problem while giving it an incorrect hint. We are able to “catch it in the act” as it makes up its fake reasoning, providing a proof of concept that our tools can be useful for flagging concerning mechanisms in models...
> Claude seems to be unaware of the sophisticated "mental math" strategies that it learned during training. If you ask how it figured out that 36+59 is 95, it describes the standard algorithm involving carrying the 1. This may reflect the fact that the model learns to explain math by simulating explanations written by people, but that it has to learn to do math "in its head" directly, without any such hints, and develops its own internal strategies to do so.
- Fond-ish memories of Disney's direct-to-vhs push in the 90s
https://www.businessinsider.com/disney-straight-to-video-seq...
- slashdot hasn't changed in a decade so it's not an unreasonable prediction!
- Next up: hallucinate the contents of the links too!
- .js, short for jecmascript, easy
- >... the standout was a version that came to be called HH internally. Users preferred its responses and were more likely to come back to it daily...
> But there was another test before rolling out HH to all users: what the company calls a “vibe check,” run by Model Behavior, a team responsible for ChatGPT’s tone...
> That team said that HH felt off, according to a member of Model Behavior. It was too eager to keep the conversation going and to validate the user with over-the-top language...
> But when decision time came, performance metrics won out over vibes. HH was released on Friday, April 25.
They ended up having to roll HH back.
- GPT3 was originally just a completion model. You give it some text and it produced some more text, it wasn't tuned for multi-turn conversations.
https://platform.openai.com/docs/api-reference/completions/c...
- > Quickly learned that people are ridiculously sensitive: “Has narcissistic tendencies” - “No I do not!”, had to hide it. Hence this batch of the extreme sycophancy RLHF.
Sorry, but that doesn't seem "ridiculously sensitive" to me at all. Imagine if you went to Amazon.com and there was a button you could press to get it to pseudo-psychoanalyze you based on your purchases. People would rightly hate that! People probably ought to be sensitive to megacorps using buckets of algorithms to psychoanalyze them.