robotelvis

I wrote an interactive tutorial for how modern AI works. Every chapter is explained with interactive playgrounds and every chapter has been proof read by my 11 year old son to confirm that it is accessible.

This was on hacker news a while back, but it now has 4 new chapters - taking the reader from next token prediction to a real AI agent with post-training, conversations, reasoning, and tool use.

The goal is that this tutorial shouldn't require much math knowledge. If there is anything in there that is too hard to understand then let me know.

My 11 year old son seemed to get it all, so I think it should be pretty approachable, but I'm sure there is stuff I can improve.

I also looked at getting a flip phone but decided it wasn't really viable. Realistically you need to have an iOS or Android phone, because even if you don't need it /most/ of the time, you do need it sometimes.

The trick is to find ways to carry it with you less.

I played with disabling Safari but it's hard to sustain that because you'll often find yourself needing to open a random QR code to do something.

But I have buried it on the third page of a sub-folder, so I'm not tempted to randomly open it.

I think the most important thing is to just not /usually/ have the phone in your pocket, so you break the habit of reaching for it.

Cool that you were able to do that.

There are definitely lots of things you need a smartphone for today, but fortunately few enough that it's practical to not have a smartphone with you most of the time, even if you own one.

Hopefully if enough people go phoneless, society will adapt to require smartphones less often.

New Public (https://newpublic.org) | Senior+ Software Engineer | REMOTE (US, Canada) | Full-time | 160-175k

We are working with a consortium of public broadcasters around the world to create a publicly-owned public-interest online space that helps people with different viewpoints understand each other and relate to each other. This project has a lot of momentum behind it, particularly as recent political changes in the US are getting countries interested in the idea of digital sovereignty.

Our partners have a combined regular audience that collectively includes a substantial fraction of the western world, so if we get the product to a good enough state that they can promote it aggressively, it has the potential to be one of the biggest and most important platforms in the world.

The New Public dev team currently consists of me, Rob Ennals (ex Google, Meta, Quora), Blaine Cook (created Twitter), and Daniel Bachhuber (ran Wordpress.com), so it's a fun team to be on. We also have very strong experienced Designers and PMs.

Particularly interested in people with experience working on social products, experience being a core developer of a large product with lots of external contributors, or a deep interest in helping people with different viewpoints relate to each other.

https://newpublic.org/jobs/software-engineer-psi

I was asked to give a talk on how to make the most out of OpenAI's new Codex coding agent, and it turned into a blog post.

Codex works best when you have a really clean code base, with thorough tests, and well documented best practices. Getting your code base into that kind of shape might seem like a lot of work, but Codex is already great at the kind of mundane tasks needed to get your code into a state where Codex can work well.

Codex has also changed how I think about software engineering. With previous AI tools, I used them to help me write the code, but now my job is to mentor a team of AI agents as they write the code.

Even if you don't write code, it's important to understand what it's like to use Codex, because this kind of "manage a team of AI agents to do your job" is likely the future of all human work.

Best to apply to the jobs you think are interesting using the form.

If we think you could be a good match, then we'll reach out to you.

New Public (https://newpublic.org) | Open Source Developer | REMOTE | Full time $140-$160k

We are working with a consortium of public broadcasters around the world to create a publicly-owned public-interest online space. This project has a lot of momentum behind it, particularly as recent political changes in the US are getting countries interested in the idea of digital sovereignty.

We have lots of people wanting to get involved in the project, and so we plan to open the project up to anyone to contribute, whether they be public media, private media, institutions, universities, or enthusiasts. To make that work well, we want to hire an open source engineer who can manage contributions, create training materials, do refactoring, manage forums, and do all the other things necessary to nurture a healthy open source community.

Particularly interested in people with experience maintaining Open Source projects and working with outside contributors.

https://newpublic.org/jobs/open-source-dev

If your advisor is one of the stars of their field, then probably more than one of their grad students will be professors. If not, then probably none of them will.

Otherwise this point is indeed correct.

In my experience the best approach is to first try to solve the problem without having read the prior work, then read the prior work, then improve your approach based on the prior work.

If you read the prior work too early to you get locked into existing mindsets. If you never read it then you miss important things you didn’t thought of.

Even if your approach is less good than the prior work (the normal case) you gain important insights into why the state of the art approach is better by comparing it with what you came up with.

I think part of the issue is that the people who want simple non-SUVs are the people who mostly buy used cars.

The majority of people buy their cars used, but the car makers make the cars they can sell to the minority who buy new cars - and they have very different tastes. New car buyers like gadgets and they like SUVs.

Demis was a legend at Cambridge. I was in CS a few years below him. He would have had very strong recommendations from professors.

The flip side is that if punishments are eye watering harsh then people will be even less willing to inflict them.

Bear in mind also that the vast majority of academic fraud isn’t cut and dried easily proved. It’s p hacking or “accidental” flaws in an analysis, or forgetting to mention some important detail.

Calling out bad work is career suicide. You are defecting on your tribe. That’s half of the problem.

Another reason for doing this is that if the people doing replication also do original research then calling out someone’s work as bad incentivizes them to sabotage your work when they inevitably review your papers.

You can avoid that to some extent by having replication and original work be separate specialities - and making sure that replication gets prestige so good people do it.

The tricky thing with randomizing is that science gets very specialized, both with equipment required and knowledge. So there may only be a handful of people whose work you can competently replicate.

And those same people are reviewing the papers you publish and will not hesitate to sabotage your career if you have made them look bad by failing to replicate their papers.

Well you aren’t going to get tenure unless you distort your results and it’s hard to change established habits.

That, and you select for the kind of people who are willing to fake results to further their own careers.

To some extent the Elks is a healthy community. Our lodge is definitely a place where kids can hang out together free from interfering adults - that’s what our kids do there with their friends all the time.

We show up, they run off and do their thing with the other kids. We find them when it’s time to go home.

It’s a place with a fence so they can’t wander of, where you know everyone is safe to be around, and where you know if something bad happened another Elk would see and sort it and and come tell you.

They do indeed that rule. It’s somewhat controversial internally and I can imagine it eventually going away, but is there now.

Fortunately they are pretty relaxed about what you mean by God. I’m a pantheist (everything is God and individual consciousness is an illusion) and AFAICT that’s fine. Nobody dug into my beliefs you find out exactly what I thought.

I think the key intention is that you can’t be a hedonist who cares only about yourself. You need to care about living a life of service to something greater than yourself. I’m on board with that.