In our Author Spotlight series, TDS editors chat with community members about career paths, writing, and sources of inspiration in data science and AI. Today we are pleased to share our conversation with Stephanie Carmer.
Stephanie is a staff machine learning engineer with nearly 10 years of experience in data science and ML. Previously, she was a higher education administrator and taught sociology and health sciences to undergraduate students. She posts monthly on TDS on social themes and AI/ML, and speaks nationally on ML-related topics. She will be speaking on Strategies for Customizing LLM Assessments at ODSC East in Boston in April 2026.
You learned sociology and the social and cultural foundations of education. How has your background shaped your views on the social impact of AI?
I think my educational background has shaped the way I look at everything, including AI. I learned to think sociologically throughout my academic career. This means looking at events and phenomena and asking questions such as, “What social inequalities are happening here?”, “How do different kinds of people experience this differently?”, and “How do organizations and groups of people influence this happening?” These are the kinds of things sociologists want to know, and we use the answers to understand what’s going on around us. They form a hypothesis about what is happening and why, and are eager to find evidence to prove or disprove that hypothesis. This is essentially the method of sociology.
You’ve been working as an ML engineer at DataGrail for over two years. How has the rise of LLMs changed your day-to-day work?
I’m actually writing a new article about this. I think the advances in code assistance with LLM are very exciting and are changing the way many people work in ML and software engineering. I use these tools to bounce ideas off, get critiques of approaches to problems, get alternative ideas for approaches, and for wrap-up work (such as writing unit tests and boilerplate code). But I think there’s still a lot of work for ML people to do. Specifically, applying skills gained from experience to unusual or unique problems. And all this is not to downplay the downsides and risks to LLMs, which are numerous in our society.
You asked if you could. ”Saving the AI economy” Do you think the AI hype has created a bubble like the dot-com era, or do you think the underlying utility of the technology is strong enough to sustain it?
I think this is a bubble, but I think the underlying technology is really not to blame. It was people who created the bubble, and as I said in that article, unimaginable amounts of money have been invested in the LLM technology under the assumption that it will produce some result that will yield commensurate returns. I think this is ridiculous. Not because LLM technology is useless in some important way, but because it is useless in over $200 billion. I think this could be a sustainable space if Silicon Valley and the venture capital world were willing to accept good returns on modest investments, rather than demanding huge returns on huge investments. However, the reality is that this is not the case, and I see no way out of this situation other than the bubble eventually bursting.
A year ago, you wrote about “.Cultural backlash against generative AI” What can AI companies do to rebuild trust with a skeptical public?
This is difficult. This is because I believe that the hype has decided the direction of the backlash. AI companies make outlandish promises because the next quarter’s numbers always need to show something great to keep the wheels turning. People who see that and feel that they are being deceived will naturally feel bad about the whole endeavor. It won’t happen, but it would go a long way if AI companies backed away from unrealistic promises and instead focused intently on finding rational and effective ways to apply their technology to people’s real problems. It would also be helpful if we could conduct a broader public education campaign about what LLM and “AI” actually are, demystifying the technology as much as possible. But I predict that the big players in this space won’t try to do something like that either, because the more people learn about the technology, the more realistic they become about what it can and cannot do.
Over the past few years, I’ve covered a variety of topics. How do you decide what to write next?
I often spend the month between articles thinking about how LLM and AI are showing up in my life, the lives of people around me, in the news, and talking about what people are experiencing when they see it. Sometimes you want to use a particular angle from sociology as a framework (power, race, class, gender, institutions, etc.) to look at a space, or sometimes a particular event or phenomenon gives you an idea to work on. Write notes throughout the month. When I come across something that I’m really interested in and want to research and think about, I choose it for the next month and dig deeper into it.
Are there any topics you would like to tackle in 2026 that you haven’t written about yet?
To be honest, I don’t have plans that far in advance! When I started writing a few years ago, I wrote down a huge list of ideas and topics, but I’ve completely exhausted them, so these days they’re only a month or two away at best. We welcome ideas from readers who would like to explore further topics that collide with social issues and AI.
To learn more about Stephanie’s work and keep up with her latest articles, follow her on TDS or LinkedIn.
