This interview was conducted in December 2024. It has been transcribed and edited for clarity.

Naomi Bashkansky is currently a senior pursuing a concurrent master’s and bachelor’s degree in
computer science. Naomi’s background in software engineering has transformed into crucial work in the
AI sphere — both in AI safety and AI policy. Naomi serves as the Technical Director of Harvard’s AI
Safety Student Team. This interview was conducted in November 2024.

HTR: Your peers consider you an innovator, but what does being an innovator mean to you and
how does this title manifest itself in your daily life?

Naomi: I think an innovator consists of two steps: the first is to notice important problems and the second
is to hold yourself accountable for working on them even when no one else will hold you accountable to
work on them. Two stories here that I think represent this quite well come from the mathematician
Richard Hamming. So, Hamming used to go around and ask scientists from different fields “What are the
most important problems in your field?” and then he would ask them “Why aren’t you working on them?”
Although I think this is partially a troll, it’s partially, I think an important question to get asked, in order to
work on things that matter.
Another story from Hamming that I thought was quite fun that goes towards the holding yourself
accountable part is from him working on the Manhattan Project. There was some concern that detonating
the first nuclear bomb would cause a chain reaction and then ignite the atmosphere… which is not good.
Hamming was asked to check a physicist’s equations but didn’t know the equations, which caused him
lots of stress. The physicist upon seeing this, replied that no one would ever blame Hamming if he didn’t
know the equation and the atmosphere did end up igniting. I think that that’s the wrong attitude to
have…and I think Hamming realized this was a problem and thought very hard about it.
So how do these stories relate to my daily life? I think this is how I actually ended up working in AI
safety. I was not interested in AI at all, a few years ago. I think I became somewhat more interested when
I realized GPT 3 and Dall-E 2 existed as like “Oh these AIs are actually getting quite good these days!”
But I only became very passionate about it when I started thinking very hard, starting two years ago,
about the trajectory of AI and what subproblems in AI are the most important to work on. And now I do
that!

HTR: What sparked your interest in computer science and AI? Did you come to college knowing that
was something you wanted to work on or did you figure that out throughout your trajectory here?

Naomi: I knew I wanted to work on computer science for a while. I initially wanted to do software
engineering like my father before me. And I think I always knew since ninth grade when I first took a
Python course that software engineering is just a nice, cushy job in an interesting subject area … so yeah,
why not? I think that was my initial rationale for going to computer science. I was like “Okay, well, by
default, I’ll have a nice job in software engineering or maybe quant finance if I’m lucky.” But I would say
I wasn’t super passionate about it — I would say that I only became very passionate about any intellectual
subject when I got into AI. I outlined how this sort of happened in the previous question, and I would say

in large part, I became very passionate about this when I encountered other students in the AI safety
student team when it was starting up two years ago, and having many debates with them — late night
debates — about how the future is going to go.

HTR: So you mentioned the student team. Do you have any clubs, classes, or experiences at Harvard that
have shaped your interest during your time here?

Naomi: Yeah, I would say, in addition to the amazing team, the other thing that perhaps has shaped me
the most is a Gen Ed I took in my freshman spring. This is Joshua Green’s Evolving Morality Gen Ed,
because it is part of this class that I learned about GPT3’s existence — so this is the early days,
pre-ChatGPT. I learned about Dall-E 2’s existence, and I read super intelligence. I think those three things
did quite a bit to jump-start my interest in AI.

HTR: So as a senior now, you’ve had your fair share of internships and work in the field. What insights
can you share from your experiences and what have you learned from working in industry?

Naomi: There are just way too many things to share. I think that you learn a lot by being in industry and
being immersed in a particular subject. I think this is at least in part, why I wanted to work in policy,
because it was so new to me. I learned so much this past summer from being immersed in a completely
different field and figuring out how policymaking actually happens. How do bills actually get passed? All
this sort of passive knowledge that you just don’t learn from being outside. I think this applies especially
to a field like policy where Senators aren’t going on Reddit and explaining “Oh yeah, here are the deeds of
how to convince me of whatever such and such thing.” And it may be certainly less so for areas like
software engineering, where you actually can find quite a lot of information online about how getting
stuff done in software engineering actually works. I think you just learn a lot of passive knowledge. I can’t
describe all the things that I have learned, but I do recommend having extensive conversations with
seniors, professors, TAs, or grad students who do have industry experience to learn these things if you’re a
freshman and you’re reading this.

HTR: As a follow-up question, how do you reconcile the policy aspects of your work with the technical
aspects? Do you see the two working together?

Naomi: So a few things come to mind here. The first is concrete advice for younger years. For people
who actually want to work in DC and have a technical background, there are various fellowships that help
significantly with this. The ones that come to mind are Tech Congress and the Horizon Fellowship. These
do sometimes require higher degrees, which I think is why getting a concurrent Masters is sometimes
useful. But many people don’t know about these fellowships, and I think they’re quite important if you
have a technical background.
Setting that aside, in terms of how I can use my policy experience for technical stuff: even though I had a
fantastic time in DC, I think I don’t immediately want to return to policy making. I do quite enjoy
technical work, but perhaps one day I will return, in which case, my policy experience will obviously be
invaluable. I think there are some types of roles that I might be interested in that are mostly technical in
nature that also benefit from policy. So there’s some roles that are explicitly benefited from policy, such as

a role at a frontier AI lab like OpenAI, Anthropic, and DeepMind, focusing on risk management. There, I
think having policy experience and knowledge is directly quite useful, whereas I think just learning more
about policy has improved my model of the world of how decisions happen and how the future is likely to
go. And this somewhat improves what research I decide to work on, even if I’m working on a purely
technical role, or even just whether I choose to work on a technical role at all.

HTR: That’s very insightful. Sort of switching gears now, I see that you are a World School Chess
Champion, which is really cool! How have your hobbies and interests played a role in where you are
today? Do you like doing stuff for fun that relates to what you’re doing in the field, or are these separate
things for you?

Naomi: Yeah, I did play quite a bit of chess growing up. Specifically, I played competitively from ages
five to fifteen, and then I sort of quit in early high school because I wanted to do more useful things with
my life, if I’m honest, so I learned to code instead.
I would say I very rarely play chess these days if I’m honest. I do occasionally play for fun, but it’s usually
more as a social activity to get to know people or reconnect with friends, rather than chess for its own
sake. I don’t know if I have really many hobbies these days. I think I’m a bit busy for that.
How has chess shaped me though? I think there are many ways that people sometimes assume chess has
shaped me, such as that it made me more patient, or it made me “smarter” or something. I think both of
these are false. I think that I was an unusually patient five-year-old, and this is why I was good at chess as
a five-year-old…so I don’t think it made me more patient. I do think one thing chess taught me, in
particular, what achieving success in chess taught me, is that I am capable of achieving a lot of success,
and I think this made me much more ambitious with what I do. And if I’m honest, I still think I wasn’t
nearly ambitious enough as a kid. Like, I think I actually would advise my younger self to be much more
ambitious, and I think a lot of young teenagers can actually do quite a lot, can learn very useful skills, and
can have a big impact in the world, even before they enter college. And I think I just did not realize this at
all as a teenager. And there are some people, some of my friends at Harvard, who have done this, and I
think this is extremely cool, but nonetheless, chess did teach me to be ambitious, and then Harvard taught
me even more so.

HTR: Sort of along those lines, how have you found community at Harvard or in the things that you do?
How have you found people to hang out with and engage in shared interests?

Naomi: I think in my freshman year, my strategy for this was to take a bunch of difficult computer
science classes. And I think this was an excellent strategy. I met many interesting people, some of whom I
am still extremely close to these days.
Then after that, it was mostly via the AI safety team, where I met many wonderful college students and
law school students and CS PhDs who I became quite close to.
And I think now, this year, or at least this semester, I realized that all of these people that I met in
freshman year have become really cool people, now that I’m a senior and I am trying to meet many more

of them and to, grab lunch with many more of them — people who are interested in not just AI, but also
startups or physics.

HTR: Are there any anecdotes that you would like to share about any challenges or successes that you’ve
encountered or how to overcome them?

Naomi: So I’m doing the concurrent master’s and bachelor’s. And for some time, I was considering, do I
want to actually finish up the master’s and graduate in May? Or do I want to just get the bachelor’s, finish
up in December, and start full-time research in January? I was thinking very hard about this over the
summer, and then I decided, why not do both? So I am currently enrolled in seven classes, so that next
semester, I’ll have time for research. In addition to taking seven classes this semester, I’m also helping run
the AI safety student team as a technical director, and also doing a part-time job.
This is kind of a lot. I have somehow found the time for everything and still get enough sleep. I think in
previous semesters, I just didn’t work that hard or maybe didn’t work that efficiently, or something. I think
I could have done this in the past as well. Yeah, it’s not really a story of “I had a challenge and then I
succeeded.” It kind of just worked out.

HTR: One final question. Looking beyond graduation, what do you see yourself doing? And how would
you like to influence your field? Where do you see the rest of your life after Harvard going?

Naomi: Starting in January 2025, I’ll be taking a leave of absence to join OpenAI’s Alignment Team as a
Resident.
With regards to what I want to do a few years after college, I think it honestly depends a lot on the state of
the world and what happens with AI. I think it’s just very unclear to me how quickly AI capabilities will
progress, and it’s plausible they will progress quite substantially in the next few years. And depending on
how the world reacts to that, I might change my career decisions accordingly.

HTR: Thank you for taking the time to interview with us, Naomi!