#TanukiTuesday 1/4/2022
What even is AI?
Menu
Definitions
Machine Learning, Artificial Intelligence, Deep Neural Networks, Algorithms, what? A lot of these terms are confusing but are thrown around like everyone knows what they’re talking about. It can be a bit confusing and a bit daunting to invest in a project that makes claims about words that have such general meanings, so I think it’s best we get familiar with these terms and how they interact with each other.
Machine learning is the core field that uses math and computer science to create models for prediction. Machine learning is very commonly used in Artificial Intelligence, which is intelligence that is made specifically for human interaction (whether knowingly or not), in order to create models that have some sort of mathematical guarantee of accuracy. Together along with neural networks, a specific type of machine learning model, our society has changed dramatically over the past few years.
The rise of Facebook, TikTok, and other social media was driven by artificial intelligence, using algorithms to determine what you want to see and optimizing for your viewing experience. I have the personal view that “algorithms” has been thrown around as a scary word in media today to make it feel impersonal, manipulative, or out of our control. This doesn’t have to be the case, and the way that happens it through education of what these are designed to do and what everyone gains from these additions to society at large. This is why at Tanuki communication is one of our top priorities, and as such we will be as open as possible about we are doing over the coming months.
Machine Learning
Machine learning has a deep history, and depending on who you talk to, even the simplest use of math can be considered machine learning, as long as we are trying to create a model to use in some form of prediction. Take for example trying to distinguish between dogs and cats. Certain attributes of both animals can be plotted against each other (think tail length or hair length). Doing this gives you a cluster of points, and sometimes the clusters are only dogs and sometimes the clusters are only cats.
Machine learning in general can be used here in many different ways. We could use a “supervised” learning algorithm which uses knowledge of each point being a dog or a cat to try best guess what is a dog or what is a cat. We can also run an algorithm that is “unsupervised” which tries to determine groups based on the chosen attributes, which may give us more or less labels (think different breeds of dogs and different breeds of cats). The latter approach has the added benefit of giving us insight into our data where someone might not be initially looking.
These algorithms can be as simple as drawing a best fit line and choosing based on which side of the line a point is on, all the way up to Deep Neural Networks, which use many attributes along while creating intermediate attributes by cascading these models (more on these next blog post!).
Artificial Intelligence
AI is used widely in society today, but I want to look at one of my favorite examples, and the example that is most commonly cited. Alpha Go was developed by Google and it revolutionized the way that we think about what machines are capable of and what we want to push next. Go was a game long thought impossible to better than human at because the amount of combinations was too much to be done analytically, and there weren’t good ways of estimating which move should be made. This was until Google developed a model that was able to estimate the the quality of its move choices along with another model to determine who would win the game after.
This was revolutionary as this was the first time a computer was able to win vs the best Go player in a series. AI is now widely used in every day life from medical devices, weather forecasting, and so many other applications that we all benefit from every day without even knowing.
Bridging the Gap
But wait… What does this have to do with crypto? What are you going to do to interact with people and learn?
Our goal is to create NFT’s using similar techniques and learning based off of market conditions what people want and how to connect art and artists to people who want to buy their NFT’s. Our first step is to have AI-generated NFT’s using modified current algorithms, and then branch off into more in depth AI-based smart contracts that can interact with the rest of Web3. After that we would like to have a marketplace that infuses learning and generation to deliver the best content to the people who want it, and allow access to services for creators themselves.
Our first goal is to drive organic growth by being welcoming to newcomers into our ecosystem and community.