How many times have we wished we could hold everything at once? All the reports, all the notes, all the small details that seem important but are spread across files and moments. What if the system we use could actually “remember” more, not just a few pages, but entire histories?
Google’s Gemini is moving in that direction. With the ability to process up to 2 million tokens, it can take in what would roughly equal 1.5 million words. That could include therapy notes, assessments, videos, interviews, and research, all at once. For therapists, this touches something very familiar: our work is rarely based on one piece of information, but on how many pieces come together.
Think of a child followed over several years. Different professionals, different reports, different perspectives. We often move back and forth between documents, trying to connect them into a meaningful picture. A system like this could help bring everything into one place and highlight patterns or gaps. But even then, it is not “understanding” the child. It is organizing information, not sensing meaning.
Because in clinical work, more data does not equal deeper understanding. We listen, we observe, we feel the atmosphere in the room. We notice what is said, but also what is not said. AI can gather and structure, but it does not experience the relationship. It does not feel the hesitation, the resistance, or the shift in tone.
Used carefully, this kind of tool can still support us. We might ask it to compare reports, summarize repeated goals, or point out inconsistencies. It can help us prepare and save time. But the interpretation, the decision, the responsibility, all of that stays with us as clinicians.
There is also something tempting here. When answers come quickly, we might rely on them too easily. But clinical thinking takes time. It is built through questioning, reflecting, and sometimes sitting with uncertainty. If AI moves too fast, we risk skipping that process.
And beyond practice, there are ethical questions. What are we sharing? Do we have consent? Are we protecting the people behind the data? Even if AI can handle more information, we are still responsible for how that information is used.
In research, this expanded “memory” may help us review and organize large amounts of material. But again, it does not replace careful reasoning or methodological rigor. It can support the process, not guarantee its quality.
So maybe the question is not only whether AI can remember more. It is how we use that memory. Does it help us think more clearly, or does it think for us? In the end, what matters is that the person in front of us does not become just another piece of data, but remains at the center of our attention.