“Is it fair?” “Is it explainable?” “Is it safe?” These are the questions commonly used to evaluate AI systems. In mental health, they seem especially relevant. Ethical guidelines, audit tools and compliance checklists promise trustworthy AI. But how helpful are these tools when emotional nuance and personal vulnerability come into play?

Imagine an LLM tailor-made for your cultural context such as where you live, the language you speak and the values most important to you. You can depend on it to proofread your emails for tone and social faux pas, respond in everyday, colloquial language and give you relevant recommendations on how to navigate your relationships. You can trust it not to give awkward responses that could put you in a compromised position and to understand the subtle nuances in human interaction vital to help you navigate everyday life.

Have you ever wondered how ChatGPT understands your sentences? The answer is hidden in two mathematical tricks that seem more complicated than they are.

Imagine this: You’re texting your friend about weekend plans. You type “Let’s meet at the park tomorrow” and send it. This is simple for you and your friend, right? But if AI were reading this message, it would be translated to something it can actually work with – numbers. Lots of these numbers format into what we call vectors.

Here we are in the world of embeddings and positional encoding, where words become vectors and positions, essential elements for AI to understand human languages.

Imagine opening up about your anxiety to a chatbot at 2 AM, something you usually don’t do with a human. It “listens”, responds, maybe even offers advice. But how does it know what to say? Who trained it, and what values guide its response? Is it really helping or nudging you towards emotional dependence?

These questions are not theoretical. They are at the heart of alignAI’s Mental Health use case.

It’s not so long ago that we thought of AI as belonging solely to the realm of computer scientists, mathematicians and technical experts. But it’s time to move away from this idea.

Who might need to be involved to create an AI system that helps us detect wildfire threats in different forests? Or creating an LLM which aids in the detection of cancerous cells? Or an AI that facilitates smart energy use in a home?

Have you ever wished ChatGPT a good morning? Asked it to “please” do something for you? Or thanked it when it did? Now ask yourself why. Maybe you wanted to be polite? Maybe it was a reflex? Or maybe you expected it to provide better results if you were friendly?

As AI systems (Large Language Models (LLMs) in particular) become more fluent, responsive, and seemingly empathetic, many users report a subtle shift in how they interact with them. ChatGPT remembers things from your past conversations, meets your energy if you are polite, and remembers your name. At times, it feels less like using a tool and more like speaking to a person.