I have been reading more and more about how Generative AI works ‘under the code’, and I think there are some real interesting questions to be asked from a brain science and Alzheimer’s perspective.
For those of you who may be looking for a simple introduction to Generative AI also known as GenAI, here we go.
GenAI identifies patterns and probabilities, and uses this knowledge to essentially build new content (thus the word generative). A simple example is – if the most common word after “I want to” is “eat”, and then followed by “sleep”, “run”, “walk”, etc., the system learns the likelihood of what the next word or phrase is following a given word or phrase. Of course if the system always gave you the most probable word or phrase it would still be a Generative AI but a big boring Generative AI. And to solve for this, GenAI models introduce some randomization to make it ‘creative’. When this creativity gets too far or creates content that is unlikely to make sense to us humans, the AI engineers call it hallucination. So there we go – that is a simple introduction to GenAI (and there is of course a lot more that happens that is perhaps fodder for another post!).

Now coming back to Alzheimer’s I believe that using a corpus of stored memories, new models can generate or recreate memories. If GenAI can create poetry, make art or even produce videos, I do not see why we cannot some day create memories. As scary as that might sound, I think that with the right rules and protocols, this could be a real blessing for people who suffer from memory loss.
What do you think? Are the days of generative memories looming ahead?
Questions will reign, like what are the ethics behind this. What if we accidentally create the wrong memories? Having a nuanced perspective is SO important because despite the name, neuroscience is not just limited to biology. We have to factor in computer science, engineering, ethics, philosophy, psychology, language study and SO many other fields.
(image source: National Geographic)
Leave a comment