En Route to AI PhD: NSF Fellowship Weeks 0–5
Jesse Parent: A free-form set of thoughts on my experience as part of the inaugural cohort of the National Science Foundation’s CSGrad4US Fellowship. In this post, background and stage setting.
October 4, 2021: This post is a bit of a rush-up recap of the last month or so since this period of my life began. Finding out about the fellowship was a late-August surprise, after an unassuming email notification. I still remember Orthogonal Research and Education Lab PI, Dr Bradly Alicea for his letter of recommendation months ago. August was also my first month after moving to Boston, following a whirlwind of a summer abroad on a bit of travel, new jobs, research, and maybe “a week of vacation in London surrounding a business trip to Oxford.” It was also exciting to be lead organizer and co-host for the Discussion Group at CogSci 2021 “Trajectories in Cognitive Science”, as well as participate in a slew of other connference activity at OREL. Seeing Boston at the end of summer, and finding I know many colleagues, academic, general nerds, and friends here already, was a treat.
September, though, was a clear shift away from hustle and bustle going late into the evening, Back Bay full of life; the shrinking golden hour fitting the season’s narrative perfectly. I admit, I am enjoying being close to a body of water through the seasons — it’s been a long time since that was the case for me. There is something about the spirit of Boston that makes me feel at home, even if ultimately I feel like one of its many academic tourists. I have a feeling I will not be in my current flat for its whole lease, let’s say, one way or another. I will earnestly hope I move due to a PhD program.
The fellowship itself has gotten off to peaceful start, relatively. We’ve only technically had two “weeks of meetings” for the training portion of the program so far; in essence, fellows get 2 years of mentoring and cohort group meetings during the application and (ideally) first-year of studies.
I appreciated the background and intent of the program within the first week; in essence, the driver of this program is to help steer domestic PhD prospects to actually applying in and completing their training here in the USA, and help maintain the academic pipeline within the nation. It wasn’t so explicitly mentioned, but many PhD or graduate-research level students are swept away into industry, or into other trajectories of career in general.
The fellowship is also designed “for people just like me”, at least, somewhat. In my case, I am a non-traditional student who went back to school after changing careers somewhat and have spent some time in industry. From some conversations, I (fortunately?) seem to have a solid amount of research relative to many industry-based folks, and have quite a good sense of the academic shuffle therein. The grade-out involved some elements of diversity and as I like to say, creating on-ramps for others, as part of being someone the US Government and NSF considers for investment. To that end, I’ve become a bit more intentional about using its platform to speak for inclusivity and accessibility — although I’m glad that I’ve been given freedom to use space in OREL for it, too (see the Society Ethics Technology team, and the commitment to Open Science more broadly.)
The second week of training was about the application process itself. There were some solid tips that I hadn’t actually heard at many of the other gradschool-prep workshops I’ve been a part of over the last few years — so respect to the organizers on that end. I mentioned some of them in our lab meetings, as part of our work on a professional development-like segment, as well. I also got to “meet” my mentor, Dr Stephen J Guy, of the University of Minnesota. He’s working in robotics and spatial perception, and unintentionally or not did a great job of showing off the cool work done in and around his lab at UMinn.
So what do you want to specialize in, though?
One of my biggest choices right now, outside of schools and advisors of course, is this need to investigate spatial computing, VR/AR/XR, robotics, and potentially multi-modal machine learning. Those seeds have been buried a long time ago but are now popping through the soil, and I’m left wondering if I ought to tailor some of my research interests to those arenas. I have had a major impetus in learning about the Metacognition workshop at NeurIPS 2021, but not quite enough time to finish a short-paper for it, disappointingly. (I may still submit something to Neuromatch 4.0 however!)
Visions of Research
Sometime in the last two weeks, I had a particular vision of a macro-research project involving expanding the capacity of AI systems towards more generalizable intelligence (of course) — and specifically towards higher-level capacities involving representing other agent’s needs. Ideally, other human’s needs, in terms of both generic physical requirements (HealthTech), but more so inner world-needs. This is significantly lofty, and yet, I seem to b drawn to it nevertheless. So much of my life has been about understanding boundaries, trauma, trans-genrational influence on norms, self-image, understanding, and the like; it’s easy to put it into agent-environment relations, yes. But if there is a way I can transfer my insights into something a AI system could understand — or at least help other humans understand — that seems worthwhile. An “AI”, (or I might prefer the use of an AI Agent), that could properly understand everything going on in a DnD game, for example, may be a concrete, more communicable target.
I will go on the record here, however, saying I’m quite agnostic about whether or not AI can become AGI; whether or not ‘a machine’ or a disembodied AI agent of any kind, can possess the kind of consciousness that humans or other “living things” do. I’m disinclined to strongly take sides on that altogether; I enjoy the grounded-ness of biology, and understand the engineering limitations of working with what we can do now, what we can functionalize — and this is generally my baseline. But I’m not averse to breakthroughs coming to pass. Rather, it’s that it will take something rather like the “robust AI” that AI gadfly Gary Marcus frequently alludes to, and challenges others who are focused only on big data and deep learning, to pursue. I agree with him that there needs to be more cognitive structures known and in operation to get to higher level and transferable, generalizable intelligence and related capacities. But, like everyone else it seems, the means and ways to get there feel limited.
I would love to pursue this big questions further in my PhD studies, and the continue to be the background of much of my current and ongoing research as well.
Jesse Parent is a NSF CSGrad4US Fellow seeking 2022 PhD openings in Computer Science with ties to Cognition Studies. He is also an Assistant Scientist and Lab Manager at Orthogonal Research and Education Lab. Follow him on twitter at @JesParent, on LinkedIn, or get in touch at www.jesparent.com
On the world as it is right now
I think I would be remiss to conclude this letter without one section on what’s actually going on, in 2021. I can’t help but think in many ways the world is in a techno-stupor, and a pandemic hangover. Perhaps the better word is “shell-shocked”, maybe mixed with “jaded”, and honestly, a limited sense of hope, and of the future. It’s somewhat masked in laid back music and a hesitancy about asserting anything beyond the need for not-too-strong restrictions, but the zeitgeist to me feels challenging. So many of my friends are, frankly, burdened with the world, with shots/jabs, with planning how to distance, masks, lack of social engagement, lack of social support. Is this part of the inevitable turn into living a more virtual world as our social and physical environment decays (ala, Dan Fagella)? Maybe.
I confess, I’ve wondered how to share my journey on social media: IG feels far too product oriented, either commodifying yourself, or others in general; FB feels empty ; tiktok? Twitter is my bastion of a very well-tailored follow list. I follow thousands of academics in neuroscience, AI, cogsci, philosophy, and general those pursuing understanding and inclusivity — and that’s wonderful. It is truly a blessing, for me. But this lends to partly why I’m writing here, which in its original format, before publishing anywhere else, is simply in Notion (what a wonderful application).
“Fellowship” of the Future? Invest your Hope
But there is a real sense of obligation I feel, to be candid. Much of my research interests and justifications for going into a PhD program lie on betting on future outputs. I’m not particularly interested in becoming another ML PhD who works at a Big Tech company to tweak your ads to better fit your demographics and niche interests. I wonder sometimes about the choices I’ve made, in terms of politics, or simply living a quiet life. But each time I come back to my passion being research, “academia”, or at least collaborative inquisition for knowledge, and seeking to use that to improve things. So what do I do in that context?
It feels in a way like there’s a context of justifying myself, like writing a grant application, or funding calls. But the funding I’m “asking for” is peoples hope and belief, and it’s strange. It’s strange, as someone who is now publicly being benefited by “the people’s tax-dollars”, and yes, I will even say by this country’s belief in myself to contribute to graduate studies and beyond.
Whatever the case, I do feel somewhat of a call of duty. Towards rising to a challenge or a set of standards. I don’t even know what they fully are, yet. But I think the elephant in the room for this fellowship, and for many of my generation and the next is — ok, what is your vision of the future going to be? A lot of people in my generation and the next are seriously talking about not bringing children into this world. I don’t have a fully formed opinion for my life, but I feel like, at best, this fellowship and the connections I’m making along the way are something to take seriously. I don’t want to “put the burden” on others here; getting into a PhD program is enough as it is. But I’m heartened by many things, even, going back to NeurIPS and some of the figures therein.
The workshops have been solid the last few time’s I’ve attended, and, while there is legitimate questionings of whether or not “NeurIPS is getting too big”, right now it is the big tent for all of “The AI Rearchers” to gather. The continued acceptance of innovative, socially aware, or critical voices (Timnit Gebru, Ramon Amaro, Ruha Benjamin) is heartening.
Even Turing Award Winner Yoshua Bengio, probably one of the most recognizable names in “AI” and machine learning, is an example. On social media, his posts are often about climate and about the bigger picture. I don’t know the man personally, but I’ve heard him speak enough and see him as someone caring about the future. There are many who are working on a great many projects, many aims, many perspectives. I feel like the most solid contribution across the board is providing hope and opportunity, so I will aim to do that as part of my fellowship and hopeful PhD work in the future (and ideally teaching or mentoring beyond then). As I’ve said before, I believe our best bet at a better future necessitates as many on-ramps, ladders, and other infrastructures to get as many educated minds and strong hearts seats at the table. I think that’s going to afford the best change at having “a” future, particularly one that has us live up to our potential.
Formally stated or not, I think part of the goal of everyone in higher education (or serving as a mentor or educator more broadly) is faced with the task of imbuing meaning outright, and the ability to perceive, support, and foster meaning in those we engage. (This is hard work: it is knowing yourself, and earnestly striving to understand your situation, and others’ as well.) There is a lot to be hopeful for and to ride inspiration towards, but the more intentional and careful we are about crafting paths for contribution therein, the better. I honestly do not see it as a “side-quest”, even though that’s how it is most often referenced: I think it’s directly related to having a fertile enough environment where necessary innovations may become realized.