anishazaveri
Posts 1
Comments15
Can I ask why you picked MSI as an example? If I take your argument seriously, is MSI the family planning charity you recommend I donate to?
I work at a startup designing synthetic proteins using deep learning: https://www.evozyne.com/. Even though the products my company works on are impactful, due to counterfactuality, I think my impact is through ETG.
You don't need a bio background to work in bio-related ML. Getting a CS degree with some bio-related courses/self-study the side seems enough. Also bioinformatics != bio-ML.
As a person who was a biologist and now does ML:
My impression is EAs (especially 80k) think you will make an impact through research only if you are in the top few percent of researchers in the world. I think that is especially hard to achieve in biology (especially wet-lab biology) because:
- Success in biology is incredibly resource constrained. So getting into a rich lab is key
- Success in biology is much more luck-dependent than other fields. Intelligence is secondary.
Other reasons to not do biology:
-
Biology postdocs/PhDs work longer and are paid lesser than CS
-
Feedback cycles in biology have long time windows. This means it can take years to know your project failed. Personally, I found this incredibly demotivating but people’s tolerance for this can differ
-
Option value for other jobs is worse. If you have a CS degree and decide to leave academia it’s easier to get an industry job than it’s for bio
I think a stronger case may be made for substituting fish with bivalves, though this is again anecdotal.
I second that this is a problem exacerbated by 80,000 hours. For example, I used to work in biomedical research, and 80,000 hours recommends a career path that involves getting a PhD at a top school. I did my PhD in India, which severely limits my career capital. Eventually, I decided to leave research and move to data science to ETG. To be clear, there were other factors involved and I think it's likely that 80,000 hours is correct that it's only worth being in academic research if you are in the top 0.1%. But it is strangely discouraging nonetheless
Can you elaborate more on characteristics that predict successful founders. How easy is it to identify these before the applicants go through the program?