An ML scientist writes about the extreme strain of working in an AI industry collectively shaken up by ChatGPT, as he oscillates between motivation and burnout
“Every single person I know working in AI these days (in both the academy and industry) has been sparked by the ChatGPT moment. … Ahmad Anis : Be a goldfish. When things are moving so fast, it's good to remember that sometimes you'll waste a lot of effort or get scooped. … Tweets: Nathan Lambert / @natolambert : Almost everyone I know working in AI these days feels one step away from total burnout. I took the time to take you behind the curtain and know what people on the state-of-the-art AI are struggling with: https://robotic.substack.com/ ... @emeka_okafor : AI, an industry experiencing upheaval... https://twitter.com/... Oliver Groth / @omgroth : A very thoughtful piece by @natolambert on the state of affairs in A(G)I right now. Well worth a read if you also sometimes feel overwhelmed by the current pace of developments. My TLDR: Acknowledge the change, but look past the low-hanging fruits and keep asking tough questions. https://twitter.com/... Mario Filho / @mariofilhoml : ML friends, please remember to take care of your mental health Your 🧠 is a great asset and, if you hurt it, fixing it is VERY hard Slow down, take a breath 😉 https://twitter.com/... Murray Shanahan / @mpshanahan : “Almost everyone I know working in AI these days feels one step away from total burnout” https://twitter.com/... @mmitchell_ai : SUPER timely piece from colleague @natolambert. Captures a lot of the discussions in my research circles rn. “Clickyness is the driving trend in the last few months, which has such a sour flavor.” Curious abt what AI researchers are thinking/feeling right now? Check it out. https://twitter.com/... Dr. Marie Haynes / @marie_haynes : How about the SEO community? How are you doing in trying to keep up with all the change lately? https://twitter.com/... @gowthami_s : This is exactly how I've been feeling for the last few months. It is becoming almost impossible to keep up with the current pace of research. #phdlife #machinelearning https://twitter.com/... https://twitter.com/... Bec Johnson / @voxbec : Great read: what it feels like to work in the LLM space now. As a final year PhD working on the ethics of LLMs for several years, I've found it overwhelming of late how much extra activity there is now: papers, tweets, media, models, & oomplah from the bandwagon. 🙌 @natolambert https://twitter.com/... Jeremy Lu / @thecat : There is no “working in AI”, a few thousand people are doing real AI at most, the rest of us are calling an API. 嘻 居.有人如 此..說出實 話.是坦..好 討喜~🤗 https://robotic.substack.com/ ... Eric Jang / @ericjang11 : Great read by @natolambert on the simultaneous feeling of excitement and anxiety that is AI research & startups right now https://open.substack.com/... Bijay Gurung / @bg_learning : A thoughtful, reflective piece; reading it felt like taking a brief pause (lol) from the feeling of being both inundated/overwhelmed and awe-struck by the progress, or rather more generally the movement in the field... https://twitter.com/... Ryan Butner / @rsbutner : The burnout I feel stems from seeing the coming fight over AI: its already a culture war hot potato, and the discourse gets super dumb as the moment regulation is brought up. It will be a fight that doesn't produce a catharsis or conclusion, just another forever war here. https://twitter.com/... Anders Sandberg / @anderssandberg : This feels like what I have seen from the near sidelines. https://twitter.com/...
Context & Ripple Effects
Since ChatGPT detonated in late 2022, coverage of generative AI has mostly tracked capability and market effects — answer engines displacing search, models learning by watching experts — but Nathan Lambert's behind-the-curtain account supplies the human ledger of that same acceleration: nearly everyone he knows in AI feels one step from total burnout.
The strain predates the chatbot boom — ethical oversight in AI already fell to overworked peer reviewers back in 2021 — so the post-ChatGPT pace lands on a field whose guardrails were thin to begin with. Ahmad Anis's advice to 'be a goldfish' frames the coping mechanism: effort gets wasted and work gets scooped too fast to grieve it.
First-order effects
- ML scientists in both academia and industry are working under a scooping economy where being second to publish can erase months of effort, driving the motivation-burnout oscillation Lambert documents.
- Individual researchers like Anis respond tactically — treating wasted effort as disposable rather than protecting projects — which changes how science gets done day to day.
Second-order effects
- The same acceleration that exhausts credentialed researchers runs on a second, less visible workforce: companies like Scale AI hire subject-matter experts as annotators, forming what The Verge calls a hidden 'tasker' underclass bearing its own share of the load.
- Labs competing at the frontier face a retention problem of their own making, since the pace set by rivals is the pace that burns out their staff.
Third-order effects
- If the pattern holds, AI becomes a high-churn industrial discipline whose quality-control layer stays volunteer-grade — the peer-review oversight gap flagged in 2021 widens precisely as output volume explodes.
- The burnout-plus-underclass structure points toward the institutionalization debate now running through the field: whether AI settles into a 'normal technology' with sustainable labor norms or keeps operating as a perpetual emergency.
The trend: The post-ChatGPT race is converting AI research into an industrialized, high-churn profession whose unpriced costs — researcher burnout and hidden annotation labor — are becoming constraints the field itself must manage.