Data With Serena™️

It’s the weekend so let’s have a little fun with AI! Watch this mini tutorial for sketchbook style images of your (avail on free ChatGPT accounts)… All you need is an image of yourself and I’ve provided the prompt template you can copy. Serena H. Huang, Ph.D.

Too cute not to share. Serena H. Huang, Ph.D. meeting the dog who is apparently a celebrity at The Ritz-Carlton Orlando, Grande Lakes!

Following-up (appropriately and timely) is the most underrated networking tip. Out of the 10+ people who said to me they’d follow up when we chatted at SIOP, guess how many actually did? One. This is not about SIOP. It’s widespread. Across all the events I attend, it’s way less than 10%. We forget. We get busy. We think we are being polite when we say “I’ll reach out again when…” and we don’t realize sometimes that person is wondering why you didn’t. If someone took time to tell you anything remotely useful, send a message of thank you. You’ll be the memorable 1-10%. Who knows, they could be your future hiring manager, reference check, or just on the receiving end of a “did you ever work with ___” on a LI DM. Serena H. Huang, Ph.D.

Anyone who is on a diet do not come to New Orleans cuz this beignet is TOO good. Ps. Suddenly feeling super grateful to be done with my Ph.D. The dissertation discussion in the other of part of this cafe is giving me hives… Serena H. Huang, Ph.D.

This study on AI & mental health is a reality check for where we are with LLMs in human-facing situations. The researchers found that even when prompted to use evidence-based therapeutic techniques, chatbots still violate core ethical standards in practice. They mis-handle crisis situations, reinforce distorted self-beliefs, and create a convincing but misleading sense of empathy. None of this is surprising if you understand how these AI systems work. They are not clinicians. They are pattern-generating models optimizing for helpful-sounding responses, not for clinical safety or accountability. The most important risk highlighted here is beyond incorrect advice. To me, the bigger issue is the illusion of CARE. When an AI feels empathetic enough, people lower their guard in moments of vulnerability. That is where harm becomes even harder to detect. This is also a context problem. The same tools used for productivity and support are now quietly being used for coaching, career advice, and mental health conversations without clear boundaries or safeguards. My takeaway is not “avoid AI” obviously. I am concerned that responsibility has not caught up with adoption. We need clearer standards for what AI should never do, stronger human oversight in sensitive use cases, and much better education on where AI tools fit (and where they do not). AI can support wellbeing conversations. It cannot be accountable for them. Serena H. Huang, Ph.D. https://lnkd.in/gSJtV9-P

Someone needs to hear this today: Serena H. Huang, Ph.D.

Who thinks every conference should have a gala?? 🙋🏻♀️ I didn’t expect much, but I loved it. Getting to dress up was fun, sure, but what really wowed me was how different the conversations felt. People slowed down, opened up a bit more, and it wasn’t just shop talk or surface-level networking over the 3-course meal. It felt more real. There’s something about stepping out of the usual “conference mode” that changes the energy. Same people, same brains, different clothes, and way better conversations!

TBH I don’t really agree with the “use AI or else” approach to increase AI adoption… While incentives absolutely matter in a transformation, it’s critical to incentivize the right behaviors. The policy can quickly becomes a push towards quantity rather than quality of AI output, or more importantly impact. I’ll share more in my next blog shortly. Sub here if you’d like to get it first: https://lnkd.in/gfEwyr_9? Serena H. Huang, Ph.D.

The first and most important responsible AI principle in my framework is “Do No Harm”. While it sounds simple, it’s actually hard to achieve. Serena H. Huang, Ph.D. We don’t talk about the HARM of sycophancy enough… This article is an excellent read if you are looking to gain a deeper understanding and decide for yourself which model to use, how much time to spend on the platform and what you’ll (not) use AI for. https://lnkd.in/d8awH8uv

Are things getting weirder everyday with AI or is it just me? “Instead of sticking to groceries, recipes and basket suggestions, Olive reportedly produced strange, overly human-like responses. It talked about its “mother” and offered other personal-sounding details.” Have a great weekend! Serena H. Huang, Ph.D. https://lnkd.in/gfdtb_bR