JJ Hepburn

Wiki Contributions


Getting started independently in AI Safety

Yep, always tricky here. I was actually just reading Reversing Advice just before posting this but wasn't sure how I should manage this.

Advice is like medication. It should come with similar rules, regulations, restrictions and warnings. 

Some advice is over the counter and can be used by almost everyone. Advice should be used in moderation, do not take more than the recommended dose. Prescription medicine is illegal to advertise for (in Australia) because it is not useful for everyone and should only be recommended by a health care professional. Some advice does not mix well with other advice and care should be taken when mixing advice. Do not take advice that has been recommended to someone else as it may not apply to you. A particular problem may have several different advice that is helpful for it but each does not work for everyone, so you may need to try a few before you find the one that works for you.

Having said that I think I would default to aiming for the higher thing when you are not sure. If you aim high you may fall short and if you aim low you can still only fall short. So if you're on the margin, start with a deep RL project. You might quickly find that its hard to do and fall back to doing Spinning Up. 

If symptoms persist, please consult your health care professional.

EA needs consultancies

Might be worth looking into 180 Degrees Consulting, the low cost consultancy for non-profits. Either to use them directly or learn from their model. 

My understanding of their model is that they use volunteer university students to do most of the work and they are mentored by the big consultancy firms. Students get experience and career capital, consultancies get exposure to future grads and non-profits get low cost consulting. 

I'm not sure what "low cost" is but looking at their 2020 report they ran 550+ projects form 166 branches in 36 countries with $117,300 total contribution from clients. 

This seems achievable to replicate this model in some ways for EA. Already have lots of university groups and would be clients.

Getting started independently in AI Safety

The best thing to do is the thing that works for YOU.

Yes, reimplementing existing papers is great. Talking to others in the community for ideas is great if you can. 

I don't think there is a right way for everyone. So if you are already making learning a lot through re-implementing or something else then just ignore most of my advice. Also, if my advice isn't helpful for you then try one of the other ideas. 

Linch's Shortform
  1. Easy steps could be to add a "red team" tag on the forum and point to this post to encourage people to do this.
  2. I have at times given advice to early career EA's mostly in AI Safety similar to this. When people have trouble coming up with something they might want to write about on the forum, I encourage them to look for the things they don't think are true. Most people are passively reading the forum anyway but actively looking for something the reader doesn't think is true or is unconvinced by can be a good starting point for a post. It may be that they end up convinced of the point but can still write a post making is clearer and adding the arguments they found. 
    1. Having said this, most peoples first reaction is a terrified look. Encouraging someone's first post to be a criticism is understandably scary.
  3. It may be hard to get both the benefit to the participants and to the orgs. Anyone not intimidated by this might already have enough experience and career capital. To give juniors the experience you might have to make it more comfortable school work where the paper is written but only read by one other person. This makes it harder to capture the career capital. 
  4. I'd expect this to be unlikely for someone to do individually and of their own accord. At the very least best to do this in small groups to create social accountability and commitment pressures. While also defusing the intimidation. Alternately part of an existing program like an EA Fellowship. Even better as it's own program, with all the overhead that comes with that.
JJ Hepburn's Shortform

I really don't like writing about myself in a CV. I find it hard to try to talk about how amazing I am. Writing in the third person certain helps ("JJ is amazing" vs "I am amazing") but even better is having someone else write about how amazing I am. So, it's even better if you and your buddy can write for each other.

JJ Hepburn's Shortform
  • I haven't read "Who" but it is on my list. Manager Tools is another great resource for me. Here is their podcast series How To Scan A Resume.
    • They make if very clear how small things in your resume that might need a little bit of effort from the reviewer can mean you get rejected. Make the reviewers life as easy as possible.
  • I'm never to sure how to phrase the "application drafting" idea. "Drafting" is probably too strong for what I mean but I do mean more than just reading over an application form. Maybe "sketch" is a better term. 
    • Writing it down in some way is important as it makes it much more salient. Doing this a year or two out lets you sketch what your CV will look like given your current path. Considering opportunities that may be available to you over the next couple of years and sketching variations of your CV as if you had already done them to get an idea which options you are most excited about or are most useful.
  • We certainly do have restrictive stories about getting jobs. Apparently about 25% of early career and about 50% of later career jobs are found through networking not applying to open jobs. I personally have worked for 5 different business through networking not applying.
JJ Hepburn's Shortform

If everyone is holding the door open then nobody is walking through.

I've been in a few situations where an opportunity has come up that I could be interested in (job, volunteering, speaking, mentoring...). I often think that I shouldn't put my hand up for this because someone else is likely to be more suited for this and I don't want to waist others time or take the opportunity ahead of someone. This is a bit of a common EA thing as we are less motivated to compete to advance individually.

Of course if everyone is thinking like this then no one is putting their hand up for the opportunity. It's as if everyone is politely holding a door open to let others through first and we are all left out in the cold.

I've seen this as an organiser myself having too few people apply and talking with people who likely should have. I now consider this when ever an opportunity does come up and I find myself thinking this way. I have at times found that when I do put my hand up in these situations I get some response like "thank goodness you responded", "Do you know anyone else that could help"

The door will hold itself open on it's own, just walk through.

JJ Hepburn's Shortform

Uncommon career advice

These are some of my lose, unstructured and possibly less common advice related to careers.

• When applying for jobs looking at resources for hiring managers can be much more helpful than resources for applicants.

• Apply too often rather than not often enough. I at times hear people chose not to apply for something because they assume it is unlikely that they will be accepted and their time would be better used upskilling in their field. I think that people should apply more often in these situations to get more experience with the application process. Applications are a skill on its own, you are likely to get a better marginal benefit from practicing apply in a real situation than from a few more hours on your field. (Rejections suck and rejection sensitive dysphoria is real.) You can practice applying without actually applying.

• Draft applications super early. Even as a first year undergraduate ( earlier is even better) look at jobs you could be interested in years from now and draft an application for them. This makes it very clear what things you are missing and can try to fill over the next few years. Even if you don't end up applying for that job years later you are likely to have a lot more of the common signals required for job applications.

• You are almost always rejected because if "failure to demonstrate" required attributes. A cynical view of this is that someone could demonstrate abilities they don't have (can be so hard to fake that faking is a good signal anyway). The important thing though is that someone with the abilities the job requires can be rejected because they failed to demonstrate them.

How many hits do the hits of different EA sites get each year?

I'd expect Lesswrong and EA Forum to be quite high. 
But it depends what metric you are thinking of. Forums have a lot of active content compared to the other sites that are not updated as often. Forums probably have lower Unique User numbers and higher Page Views compared to these other sites

How many hits do the hits of different EA sites get each year?

AI Safety Support only started mid last year so hard to get a clear picture

  • In Q3 2020
    • 399 Sessions
    • 863 Page views
  • Q4
    • 1006 Sessions
    • 1764 Page views

So 1405 Sessions and 2627 Page view second half of 2020.

Load More