GoKawiilThe fear of missing out on the latest AI tools can be difficult to ignore.Credit: OsakaWayne Studios/Getty Images
My browser has 47 tabs open. I know this because it crashed this morning, and when I restarted it, the tab counter stared back at me like an accusation. Most of them are related to artificial intelligence: tutorials, preprints, model comparisons and launch announcements. I had opened each of them with the honest intention of reading them. Instead, I did what I always do: bookmarked them into a ‘To Read’ folder, closed the tabs and moved on with my day. The folder now has several hundred items; I have read only a few dozen.
I am a hospital pharmacist. My daily work is prescription review, checking that the medications that doctors order are safe, appropriate and correctly dosed for each person. I also do research in personalized medicine, trying to understand why a drug that works beautifully in one patient fails in the next. It is careful, slow and detail-oriented work — not glamorous, but important.
And yet, over the past year, I have felt a growing pressure to become a different kind of professional. Over the first few months of 2026, the US firms OpenAI, Anthropic and Google all shipped major updates to their AI models. Chinese AI companies arguably moved even faster, with Alibaba, Moonshot, StepFun and Zhipu all releasing new models in quick succession. Many major technology companies announced subscription plans for their AI coding assistants, and the US Food and Drug Administration published new guidance on AI in drug development. The implicit message was everywhere: the world is rushing towards AI, and you are standing still.
AI and science: what 1,600 researchers think
I got a taste of that in 2023 when, as a visiting scholar at Tsinghua University in Beijing, I helped to curate and verify biomedical training data for large language models (LLMs). I was not involved in designing or training those models themselves, but I was close enough to the frontier to see how fast things are moving.
Then came OpenClaw. OpenClaw is an open-source ‘agentic’ software tool that connects LLMs directly to your computer, letting them browse the web, read and edit files, take instructions through chat apps such as WhatsApp and Telegram, run code and perform multi-step tasks with minimal human intervention. Released last November as a tool for programmers, it was quickly adopted for a much wider range of tasks and exploded into a global phenomenon — particularly in China, where major tech companies rushed to build their own versions. A friend of mine who works in hospital administration, someone with no technical background, messaged me to ask how to install it. When your non-technical colleagues start asking about agentic AI, the FOMO (fear of missing out) becomes difficult to ignore.
So I tried to keep up. I installed OpenClaw on my laptop, two cloud servers and my phone — then left it unused. I signed up for an online course on machine learning and bookmarked tutorials I will probably never finish. Meanwhile, my actual research, a pharmacogenomics project I have been working on for two years, had made almost no progress. I was spending so much energy preparing for a future that might arrive that I had stopped doing the work that was in front of me.
Still, my anxiety increased. Far from easing my FOMO, my reading was feeding it — everyone, it seemed, was mastering AI except me.
Shifting sands
... continue reading