@Cook's banner p

Cook


				

				

				
0 followers   follows 0 users  
joined 2024 April 20 23:53:45 UTC

				

User ID: 3012

Cook


				
				
				

				
0 followers   follows 0 users   joined 2024 April 20 23:53:45 UTC

					

No bio...


					

User ID: 3012

There is an IKEA effect when it comes to LLMs, where people find the text that they prompted to be fascinating, and the text that others prompted to be boring.

That's the thing. SBF would gamble the entire Earth on double or nothing with 51% odds, because of the positive EV. Now he's in prison for 25 years. It's not a coincidence. He knew he was gambling, and he lost.

Thanks! I presume that's oral finasterise, and the topical version is less effective? And doesn't the impotency go away if you stop talking it?

Curious to see how Claude sonnet 3.5 (new) performs. Feels like it has more emotional intelligence than ChatGPT. You might have invented a new AI benchmark.

I notice the front of my hairline thinning. What can I do? Male in my 30s. I lasted this long but now my time has come.

My recollection is that there are actually effective treatments if started early? And is there a way to objectively measure thinning? Calipers? Any Gwern-esque research on this?

Claude also suggests furniture and art, but also a nice piano or a grandfather clock.

If I ask it to be more creative, it suggests a custom stained glass window, a commissioned illuminated family history manuscript, a heritage garden installation, a commissioned tapestry or quilt.

For what it's worth, ChatGPT gets this answer in one shot if you just paste your original comment into it.

Yes, welcome to the future. I use LLMs all the time. They excel at building prototypes, little tools, and self contained functions. It can't one shot complex projects (neither can I ) but you can still handhold it and prompt and split the complex project into smaller modules that it can reason about.

So far I have not found a good way of getting it to fix bugs in existing codebases, but I do wonder if that's just a UI issue rather than an intelligence issue, since there's no easy way for me to let it track the data flow between several files.

The more context you add, the better. This is a mistake people make. It can't read your mind and often people just don't give it enough information.

Programming will absolutely be different in 10 years time, but so will a lot of the rest of world.

There is a "game" and you'll learn how to play it. Take these with a grain of salt:

  • Whatever your job is, your job is to make your manager happy
  • Figure out how your manager (and others who have power over you and your career) are rating you, and optimize for that.
  • On the above, it may not matter how your coworkers rate you, so you may not need to optimize for their opinions.
  • It is easier for others to judge superficial details than deep ones, so that's what they'll do.
  • You are always "busy". You help others by "squeezing in" the work between X and Y.
  • The helpful guy who always accepts extra work will inevitability end up with lots of extra work. This is okay if this work counts towards something (e.g. promotion), but otherwise can be harmful if it prevents you from doing higher impact work, so act accordingly.
  • Do visible, high impact work. If you do something, and nobody knows about it, did you really do anything?
  • Go to the social events, build relationships. Relationships outside your local team can be particularly valuable.

Maybe these are obvious, but I have seen people who don't get it. To be clear, I am not saying don't work hard, but work hard on things that matter, and there is a skill in figuring what what matters.

I suggested it because I feel like your anxieties come from reading the news and feeling like you have a duty to be informed and concerned about what you read. Your worries aren't about personal problems. If you weren't reading external sources, would you notice these issues in your day to day life?

Worrying about these things is net negative. It makes you feel bad and doesn't actually change anything. You are not obliged to stress about the state of the world.

Do you read a lot of news or social media? Try cutting all that out for a while.

I don't disagree. I think this is the thesis of Garrett Jones' "Hive Mind" book, but I haven't read it.

It would have to be genetic engineering, or embryo selection, or something like that. But selecting for intelligence is thought of negatively for arbitrary reasons.

What are the odds of getting two heads in a row (on two independent unbiased coin flips)? Do you know the answer? How many UK politicians do you think can get the right answer? 52%, sample size of 101 (Source here)

Math is hard. If, in a few seconds, you can see the relationship between 30 mile/gallon and $3/gallon means you get 10 miles per dollar, so for 450 miles you spend 45 dollars, then you probably have enough quantitative ability to work a six-figure job.

But broadly you're right. Talent is rarest resource in the universe and you can't make people smarter. But you only need one Newton or Haber to advance the frontier.

It's a real shame. Some students are far more effective consumers of education than others, so with limited educational resources we really ought to be focusing on them, but this goes against the equity dogma of the day.

It is legal in the UK to express support for Putin and the Russian side of the war? Would it fall under "grossly offensive" online communication?

I had an argument with a friend and I felt it would be illegal, but examples either way would be great. Either of someone vocally supporting Russia and wishing death to Ukraine online and nothing happening, or them then getting a police visit or charged.

I also found the voice off-putting, but I presume you would be able to instruct it to adopt a different tone and manner.

You should watch Pantheon, a two season animated sci-fi series. It feels like LessWrong: The Show. There is even a quokka cameo.

On the subject, does anyone have any sci-fi TV/movie recommendations? I am not a big consumer and I haven't seen much. What I liked about Pantheon is that it's plausible and near term, and decent modeling of global and corporate politics.

Learn to code, work in tech, it's a good fit for you.

What makes you think he's not already based and red pilled? I've not seen any indication of his personal political views.