Should I Apply to a 3.5% Acceptance-Rate Fellowship? A Simple EV Calculator

By Tobias Häberli @ 2025-11-21T10:59 (+16)

Many AI safety programs have acceptance rates in the 1–5% range. That can make it hard to tell when applying is actually worth your time.

In this post, we:


You might have heard about many opportunities in AI safety having low acceptance rates (<5% is very common, we estimate our own base acceptance rate to be roughly 3.5%) and this can feel demotivating.

We think this intuition is often misleading.

First, and most importantly, your probability of being accepted might be higher than the base rate (especially if you’re reading this, hi!).

If you have low opportunity costs (e.g. you have free time or your alternative plan is low-paid), the Expected Value of applying can remain positive even if your probability of acceptance drops below the base rate.

Conversely, if you have high opportunity costs (e.g. a very busy schedule or a high-paying alternative option), the bar for applying is higher. For the math to work out, you generally need either a higher-than-average probability of success or you need to put a high value on the non-monetary outcomes of this fellowship (such as the research impact, network, or other career benefits).

We are interested in candidates avoiding two failure modes:

  1. Highly promising candidates self-selecting out of applying because they anchor on the 3.5% base rate.
  2. Candidates applying when the math clearly suggests they shouldn't (though we think this is rare).

The Model

To formalise this, we can treat the application as a bet. We generally model the decision using the following simplified equation:

Where:

We have implemented this logic into the calculator below.

The Calculator

Application EV Calculator

Notes on the Variables

The Cost of Applying: We estimate the application takes 2–4 hours for most applicants. The general application form will likely take you 1h to complete, and then any mentor you apply to will take an additional 15-90 minutes. To find out more about how much you value your time, check out Clearer Thinking’s Value of Time Calculator.

Probability of Acceptance: We expect this to be the hardest variable to calibrate yourself on.

Intangible Value: This variable captures everything that you expect to get that isn't VFinancial​. We think this is the largest driver of value, but it is also very hard to estimate. We suggest breaking it down into:

Value of Information: As a simplifying assumption, we added a small fixed value (£10) to the model to represent the utility of the application process itself (structuring your thoughts on research) and the signal value of the result. For some people (e.g. those trying multiple adjacent fellowships) this could be considerably higher.

Examples

To illustrate how sensitive the EV is to your specific situation, here are two fictional examples describing common situations.

Case A: The Undergraduate

Even thinking they are worse than the average candidate, the upside is asymmetric enough that the bet is worth taking.

Case B: The Early-Career Professional

Here are some considerations:

For a negative example, someone who thinks they have a 1% chance, values their time at £100/h, is unsure whether an AI safety career is what they want, and has a very strong counterfactual job might get a clearly negative EV from the calculator. That’s probably a good sign not to apply!

Caveats

This calculator is not a definitive decision-making tool. But it might nudge you into the direction of applying, or against it. Please keep in mind:

Final Thoughts
If you find the calculator outputs "Don't Apply," take a moment to stare at that result. Does it feel wrong? If so, listen to your gut and apply.

Does the calculator output "Apply" but you feel unsure? Then, just for once, trust the numbers and apply.

 

  1. ^

    We also don't take into account risk aversion (the stakes of 'losing' ~3 hours are relatively low) or time discounting (the delay between applying and the fellowship is only 2-3 months).


Tristan Katz @ 2025-11-22T06:55 (+3)

As the demotivated person you referenced, I appreciate this! But I think the case for de-motivation is a bit stronger than presented in this calculation, so I'll try to steelman it.

First of all, consider the framing: you've assumed that if I don't get this fellowship I'll continue with the job I already have, and that if I don't apply I'll just have leisure time. In reality many applicants will be looking for a new opportunity, and if they don't get this they'll probably take something else. They might (as I am) be making as many applications as they have energy for, such that the relevant counterfactual is another application, rather than free time.

Here are some more specific points:

This leads to a much much more complicated equation! I asked GPT, to try to link this all together, and well... I'll just link its answer here.

So I remain pretty unsure as to whether it's worth it for myself personally, but I might put in an application anyway :)

Tobias Häberli @ 2025-11-22T09:27 (+2)

Thanks a lot for engaging!

One general point: My rough guess is that acceptance rates have stayed largely constant across AI safety programs over the last ~2 years because capacity has scaled with interest. For example, Pivotal grew from 15 spots in 2024 to 38 in 2025. While the 'tail' likely became more exceptional, my sense is that the bar for the marginal admitted fellow has stayed roughly the same.

They might (as I am) be making as many applications as they have energy for, such that the relevant counterfactual is another application, rather than free time.

The model does assume that most applicants aren't spending 100% of their time/energy on applications. However, even if they were, I feel like a lot of this is captured by how much they value their time. I think that the counterfactual of how they spend their time during the fellowship period (which is >100x more hours than the application process) is the much more important variable to get right.

you also need to consider the intangible value of the counterfactual

This is correct. I assumed most people would take this into account (e.g. subtract their current job's networking value from the fellowship's value), but I might add a note to make this explicit.

you also ought to consider the information value of applying for whatever else you might have spent the time on

I’m less worried about this one. Since we set the fixed Value of Information quite conservatively already, and most people aren't constantly working on applications, I suspect this is usually small enough to be noise in the final calculation.

there is a psychological cost to firing out many low-chance applications

I agree this is real, but I think it's covered in the Value of Your Time. If you earn £50/hr but find applying on the weekend fun/interesting, you might set the Value of Your Time at £5/hr. If you are unemployed but find applying extremely aversive, you  might price your time at e.g., £200/hr.

Tristan Katz @ 2025-11-22T06:58 (+1)

I also have the impression that these fellowships are getting more competitive each year. Do you share that perspective? If so, that would require a further adjustment to the calculation.