The life of a PhD researcher is sometimes romanticized – a world of quiet contemplation, groundbreaking discoveries, and academic accolades. While those moments certainly exist, the reality is often a whirlwind of intense focus, frustrating dead ends, and the constant battle against the relentless march of time. Today, I’m pulling back the curtain to share a glimpse into a typical day in my life, as I grapple with the fascinating, and sometimes maddening, world of autonomous vehicle forensics.
It all started with a seemingly simple suggestion from my supervisors. I’m working on developing an ontology – essentially a structured framework – to aid in digital forensic investigations of self-driving cars. For weeks, I’d been painstakingly piecing together the intricate relationships between the car’s sensors, control systems, and data logs. Think of it as creating a detailed map of how a car’s “brain” works, so we can understand what happened in the event of an incident.
Then came the feedback: “Have you considered using Large Language Models (LLMs) to automate some of this?”
It sounded brilliant! Imagine, I thought, simply prompting an AI to generate the classes, objects, and relationships needed for my ontology. It would save me weeks of manual labor! But, as any seasoned researcher knows, nothing is ever that simple. My supervisors’ advice, while promising, came with a crucial caveat: “You can’t just use an LLM. You need to justify its use. Understand the existing research, the strengths, weaknesses, and how it applies specifically to your field.”
And that, my friends, is where the rabbit hole began.
5:30 AM: The Eureka…and the Coffee
The enormity of the task hit me like a jolt of electricity. I woke up at 5:30 AM, my mind buzzing with questions. Forget gentle sunrise and birdsong; my morning began with a frantic dash to the kitchen, fueled by the urgent need for copious amounts of coffee.
My kitchen counter transformed into a makeshift command center. Laptop, dual-screen attachment (a recent birthday gift that has become my most prized possession), and a bottomless flask of coffee. On one screen, I had a Word document, bravely titled “Harnessing Prompt Engineering in LLMs for Automated Generation of Class-Object Relationships for Digital Forensics of Autonomous Vehicles.” Ambitious, I know.
On another screen, Google Scholar became my best friend and worst enemy. I was on a quest for peer-reviewed research on “prompt engineering,” the art and science of crafting effective instructions for LLMs. My supervisors, bless their demanding hearts, insist on sources no older than four years. This meant sifting through countless papers, each averaging 20+ pages, trying to discern the nuggets of gold from the mountains of academic jargon.
The third screen? That was for GitHub, where I desperately searched for any existing prompt templates that might offer a shortcut.
The Perils of Paper Trails
The next few hours were a blur of reading, writing, and rewriting. I’d devour a research paper, feeling a surge of understanding, only to stumble upon a counter-argument in the next, forcing me to revise my entire approach. It’s a humbling experience, realizing how much you don’t know, even about something you’re supposed to be an “expert” in.
I’d test my understanding by experimenting with different LLMs, the results varying wildly. I’d refine my prompts, try again, and slowly, ever so slowly, the pieces started to come together. It’s a relentless cycle: read, write, test, edit, repeat.
The World Outside (Barely)
Lost in my research, I became oblivious to the world around me. My wife, bless her patience, gave me a peck on the cheek and said good morning, but I was too deep in the prompt engineering abyss to register much beyond a vague sense of affection. She made breakfast, but the hunger for knowledge trumped the hunger for food.
Later, my mother popped in, commenting on my apparent immobility. “You’ve been stuck in that same spot since my trip to the loo!” she exclaimed. I tried to explain the complexities of ontology generation and LLM prompt engineering, but her eyes glazed over. She just wanted a bit of my time, but quickly realized I was in “do not disturb” mode and left me to my academic solitude.
The day wore on, marked only by the gradual depletion of my coffee flask and the growing ache in my shoulders. My wife returned from the gym, made lunch, and, understandably, gave me a wide berth, the untouched breakfast serving as a silent testament to my single-minded focus.
5:00 PM: Reality Check
It was 5:00 PM before I finally surfaced, blinking in the fading light. I looked down at myself: still in my night robe, coffee flask empty, hair disheveled from constant tugging, and a gnawing hunger in my stomach. I realized I’d been living on caffeine and academic adrenaline.
Enough was enough. I needed a break.
Tennis Therapy and Tunes
I traded my research uniform for tennis gear and dashed to the courts, arriving just as the matches were about to begin. Two hours of chasing yellow balls, channeling my inner Novak Djokovic (while my aging body screamed for mercy), and chatting with fellow tennis enthusiasts provided a much-needed mental cleanse.
As I drove home, the city lights twinkling, I put on I.K. Dairo’s “Ise Ori Ranmi Ni Mo Nse,” a timeless Nigerian tune that always lifts my spirits. The cool evening air, the music, the sights and sounds of Perth winding down – it was a welcome escape from the world of ontologies and LLMs.
Back to Reality (and Dinner)
Stepping back into the house, hunger pangs hit me with full force. I made a beeline for the kitchen, where the now-cold breakfast awaited its microwave resurrection. As I reached for it, my eyes landed on my laptop setup, a stark reminder of the unfinished work. For now, the hunger for food won the battle against the hunger for knowledge.
Dinner (formerly breakfast) warmed and ready, I switched on the TV, tuning into Al Jazeera. I needed a break from the research, but also wanted to stay informed. Western news outlets, in my opinion, often present a skewed view of the ongoing conflict in Gaza, so I sought a different perspective. Then, I flipped over to Channels TV to catch up on Nigerian politics, only to be further disheartened by the responses of a prominent political figure. It was a stark reminder that the world outside academia is just as complex and challenging as the world within it.
Finally, exhaustion won. A hot shower and the promise of sleep beckoned.
The PhD Journey: A Winding Road
As I reflect on my day, and on the broader PhD experience, I’m reminded of something I said earlier: “Knowledge extraction…it doesn’t come in a straight line. You learn a little, then you look for more research to support, or refute, what you’ve learned, and then learn more.”
It’s a messy, iterative process, full of detours, dead ends, and moments of profound frustration. But it’s also a journey of intellectual growth, of pushing the boundaries of human knowledge, and of contributing, in my own small way, to a deeper understanding of the world. And for that, I wouldn’t trade it for anything.
Well, maybe for a few more hours of sleep. And a self-refilling coffee flask, just in case you are thinking of giving me a gift.
Update – 12th May 2025 – “You are the best!”
After many demanding hours spent researching effective prompts for LLMs to help establish object-class relationships as a foundational ontology, I presented my findings to my supervisors during our weekly meeting.
I was absolutely thrilled by their feedback, which was not only positive but truly commendatory. It’s rare to hear that among the numerous Higher Degree by Research students in the Faculty, one’s work stands out as the best. I was deeply moved, realizing that the countless late nights and the exhaustion had yielded significant results
However, as is often the case, good work leads to more work! My supervisors encouraged me to delve deeper and further refine the prompts. The research paper is still in progress, but here is a draft as it currently stands.