The subject
For a while now I have been very fascinated by the topic of evolution. It is interesting to look back at history and study past life forms, their behaviour, what features we have retained, and what functions are fundamental to biological systems. It often makes me speculate about our future, especially with artificial intelligence and its systems becoming new, autonomous identities.
I started researching any ideas connecting evolution and AI – what can evolution teach us about AI? Are there any patterns unveiled from the past, about our motivations or design, that apply to AI as well? Is biological evolution even a useful guide to understanding AI’s foundation and future to begin with?
There wasn’t a plethora of research aimed at establishing a link between the two topics. However, I did find maybe a few papers where one discussed the effect of natural selection on future AI behaviour and another examined parallels between evolutionary pressures shaping organisms and machines.
https://www.lesswrong.com/posts/e6gg8S8AfJmwE8HLh/machine-evolution#Grounding_the_intuition
https://arxiv.org/abs/2303.16200
There were general reservations about not being able to reach a precise problem within this area as well as a lack of enough supporting evidence to with it. As a result, I brainstormed more ideas with my mother and she really helped me consider pivoting to another topic – brain and AI. One idea led to another and we somehow arrived at the point of discussing the brain, intelligence, and artificial intelligence. This area was also an extension of my original fascination with the area of biological evolution and its connection with AI, wherein neural networks are modeled on the human brain’s architecture (and there is plenty of research on the relevance of neurobiological principles/reasonings in AI’s further development).
I eventually discovered the problem of continual/lifelong learning within AI. It became very interesting to see the proposed approaches towards bettering machine learning and also developing AGI. Some studies borrowed from olfactory systems in fruit flies and underlying mechanisms of sleep in people, all of which was incredibly insightful to know. I had never probed into biology and AI’s intersection, let alone know the diverse, widespread applications of neural systems in organisms in artificial neural networks. As a result, I decided to mainly center my thesis on the use of neuroscience in advancing AI.
Final Product
For the final project itself, I currently know that I would like to develop an interactive/immersive website which raises awareness around neuroscience’s latest novel research in AI’s progress. I have some example websites that I find very enthralling to engage with, and want to design a similar visual experience with mine:
https://artsandculture.google.com/experiment/TAEPZtXK2s139g
https://artsexperiments.withgoogle.com/medusae/
I love the blend of information, interaction, and pleasing visuals in both the sites. There isn’t much complexity with the animations or elements present: both feel very clean and structured, and I did not feel confused about navigation or understanding the context of things. This not only emphasises the need of having a straightforward, simple, yet meaningful user experience, but also being skilled at captivative storytelling. I worry that I might be all over the place with my development; it is one of those things where it seems easy but is very hard to actualize in practice!
Upon trying further research of those websites, I could not find info on what code/libraries/tools went into the process, but I did use the developer tools feature in chrome to study the underlying code at work. There was quite a lot of raw javascript and some files had a vast amount of code, but I still could not get the whole picture with every component involved.
Notwithstanding this, I already knew about three.js and its ability to create beautiful, immersive 3D graphics on the web. It is very possible that similar (or even the same library) libraries were used in the makings of those sites; regardless, I want to begin the development of my website by experimenting with three.js. The examples given on the official website of other people’s work (see this: https://exp-infinite-passerella.lusion.co/ for instance) is already close to the vision and experience I see my project aligning with, hence, I am deciding to start here.
That’s it for now!
also, here is my pinterest board with some unorganised general ideas I liked seeing as projects: https://pin.it/cDNzavotw