Within the Creator Highlight collection, TDS Editors chat with members of our neighborhood about their profession path in information science and AI, their writing, and their sources of inspiration. Right this moment, we’re thrilled to share our dialog with Stephanie Kirmer.
Stephanie is a Workers Machine Studying Engineer, with virtually 10 years of expertise in information science and ML. Beforehand, she was the next training administrator and taught sociology and well being sciences to undergraduate college students. She writes a month-to-month put up on TDS about social themes and AI/ML, and provides talks across the nation on ML-related topics. She’ll be talking on methods for customizing LLM analysis at ODSC East in Boston in April 2026.
You studied sociology and the social and cultural foundations of training. How has your background formed your perspective on the social impacts of AI?
I believe my educational background has formed my perspective on all the pieces, together with AI. I realized to assume sociologically by way of my educational profession, and which means I take a look at occasions and phenomena and ask myself issues like “what are the social inequalities at play right here?”, “how do completely different varieties of individuals expertise this factor in a different way?”, and “how do establishments and teams of individuals affect how this factor is occurring?”. These are the sorts of issues a sociologist needs to know, and we use the solutions to develop an understanding of what’s occurring round us. I’m constructing a speculation about what’s occurring and why, after which earnestly in search of proof to show or disprove my speculation, and that’s the sociological technique, basically.
You will have been working as an ML Engineer at DataGrail for greater than two years. How has your day-to-day work modified with the rise of LLMs?
I’m really within the technique of writing a brand new piece about this. I believe the progress of code assistants utilizing LLMs is de facto fascinating and is altering how lots of people work in ML and in software program engineering. I exploit these instruments to bounce concepts off, to get critiques of my approaches to issues or to get different concepts to my method, and for scut work (writing unit exams or boilerplate code, for instance). I believe there’s nonetheless lots for individuals in ML to do, although, particularly making use of our expertise acquired from expertise to uncommon or distinctive issues. And all this isn’t to reduce the downsides and risks to LLMs in our society, of which there are various.
You’ve requested if we are able to “save the AI economic system.” Do you consider AI hype has created a bubble just like the dot-com period, or is the underlying utility of the tech sturdy sufficient to maintain it?
I believe it’s a bubble, however that the underlying tech is de facto to not blame. Folks have created the bubble, and as I described in that article, an unimaginable sum of money has been invested underneath the idea that LLM know-how goes to provide some sort of outcomes that can command income which might be commensurate. I believe that is foolish, not as a result of LLM know-how isn’t helpful in some key methods, however as a result of it isn’t $200 billion+ helpful. If Silicon Valley and the VC world have been prepared to simply accept good returns on a reasonable funding, as a substitute of demanding immense returns on a big funding, I believe this might be a sustainable house. However that’s not the way it has turned out, and I simply don’t see a method out of this that doesn’t contain a bubble bursting ultimately.
A yr in the past, you wrote in regards to the “Cultural Backlash Towards Generative AI.” What can AI firms do to rebuild belief with a skeptical public?
That is powerful, as a result of I believe the hype has set the tone for the blowback. AI firms are making outlandish guarantees as a result of the following quarter’s numbers at all times want to point out one thing spectacular to maintain the wheel turning. Individuals who take a look at that and sense they’re being lied to naturally have a bitter style about the entire endeavor. It gained’t occur, but when AI firms backed off the unrealistic guarantees and as a substitute targeted exhausting on discovering cheap, efficient methods to use their know-how to individuals’s precise issues, that may assist lots. It will additionally assist if we had a broad marketing campaign of public training about what LLMs and “AI” actually are, demystifying the know-how as a lot as we are able to. However, the extra individuals be taught in regards to the tech, the extra lifelike they are going to be about what it will possibly and may’t do, so I anticipate the massive gamers within the house additionally won’t be inclined to do this.
You’ve lined many alternative subjects previously few years. How do you determine what to write down about subsequent?
I are inclined to spend the month in between articles serious about how LLMs and AI are displaying up in my life, the lives of individuals round me, and the information, and I speak to individuals about what they’re seeing and experiencing with it. Typically I’ve a selected angle that comes from sociology (energy, race, class, gender, establishments, and so forth) that I need to use as framing to try the house, or generally a selected occasion or phenomenon offers me an thought to work with. I jot down notes all through the month and after I land on one thing that I really feel actually keen on, and need to analysis or take into consideration, I’ll choose that for the following month and do a deep dive.
Are there any subjects you haven’t written about but, and that you’re excited to sort out in 2026?
I truthfully don’t plan that far forward! After I began writing a couple of years in the past I wrote down a giant record of concepts and subjects and I’ve utterly exhausted it, so as of late I’m at most one or two months forward of the web page. I’d like to get concepts from readers about social points or themes that collide with AI they’d like me to dig into additional.
To be taught extra about Stephanie’s work and keep up-to-date along with her newest articles, you possibly can comply with her on TDS or LinkedIn.
